New three-monitor (4K + 5K + 4K) setup for my 2014 Mac Pro

Posted by Pierre Igot in: Macintosh
May 26th, 2017 • 5:42 pm

Like many other professional Mac users, I suspect, I was quite relieved when Apple finally shared, in early April, with a select panel of writers who cover Apple news, their plans for the Mac Pro and, more generally speaking, for pro and “prosumer” users of Macintosh computers.

When the new Mac Pro was announced in 2013, I was still using a workstation consisting of a 2009 Mac Pro with 12 GB of RAM, two SuperDrives (one for Region 1 and one for Region 2), and two GeForce GT 120 video cards driving two 30″ Apple Cinema Displays (each with a native resolution of 2,560 x 1,600 pixels). I had, in previous years, changed my startup drive from the conventional 1 TB hard drive that came with the Mac Pro to a 240GB Mercury EXTREME Pro 6G SSD from Other World Computing, and I had three conventional hard drives in the other three internal bays, as well as some external FireWire hard drives for backup.

I was reasonably happy with the 2009 Mac Pro, but of course, by that time it was getting a bit long in the tooth and I started lusting after the new Mac Pro. What I found particularly attractive was its promise of near-silent operation. While the 2009 Mac Pro was not outrageously noisy, it still was far from silent. Now, I knew, of course, that, with no internal bays for hard drives, even if I got a large internal SSD for the 2013 Mac Pro, I would still need external drives, and those could be quite noisy. But I also knew that Thunderbolt offered the possibility of a very long cable connection between the drives and the computer, which meant that I had the option to relegate the hard drives to a more remote location.

Because of past experience with problematic devices (the original Titanium PowerBook G4 with its very poor AirPort reception, and the disastrous mooing MacBook), there was no way that I was going to be an early adopter of the 2013 Mac Pro, so I waited until well into 2014 before purchasing one.

In July 2014, I bought a new Mac Pro with 32 GB of RAM, a 1-TB SSD internal drive, along with a 32″ 4K Sharp PN-K321 display (3,840 x 2,160 pixels), a miniDisplayPort-to-DVI adapter for one of my 30″ Apple Cinema Displays, and an external Promise Pegasus2 R4 Thunderbolt 2 unit with four 2 GB conventional hard drives, with a 30-foot Thunderbolt 2 cable to connect the Pegasus2 R4 to the Mac Pro. (It was an expensive order.)

When we moved to our new house in December 2014, I was able to move the Pegasus2 R4 to the basement, which meant that I now only had the Mac Pro and the displays on my desk in my office. It was close to perfectly silent operation most of the time. (Occasionally, when it gets quite warm in the room, the Mac Pro fan does kick in a bit, with a somewhat annoying high-pitched whine, but it’s fairly rare, and nowhere near as annoying as the constant whir of the fans in conventional hard drives.)

At first, it didn’t really bother me or worry me that Apple didn’t release upgraded versions of the 2013 Mac Pro. I was happy with my machine, and the lack of upgrades meant that I didn’t even have to worry about being tempted by a new Mac. But then, of course, after a couple of years, the complete lack of upgrades caused me to start worrying, like many others, about the long-term future of the Mac Pro and about the future of the Mac platform in general. (My dismal experiences with OS X on the Mac Pro, especially with Yosemite, did little to help dispell my concerns.)

Then I had another bad experience with OS X, this time with El Capitan and the random temporary video freezes it would cause on my Mac Pro. As reported earlier on this blog, the problem was initially misdiagnosed by Apple as a hardware problem, which forced me to leave the Mac Pro at the Apple store for about 10 days back in December 2016. This unfortunate situation indirectly led me to make the somewhat impulsive decision to purchase a new iMac 5K. The idea was that I would use it as a replacement for the Mac Pro for 10 days, and then, when I got my Mac Pro back (presumably fixed), I would give the iMac 5K to my wife to replace her own desktop machine, which was my older 2009 Mac Pro. (Performance-wise, the older Mac Pro was more than enough for her, but of course, she too dreamed of a quieter machine, and the 2009 Mac Pro was now seven years old…)

My idea was that not only it was time to get a new Mac for my wife anyway, but also that I would get to test-drive the iMac 5K as a workstation for my own office for 10 days. At that stage, in December 2016, there was still complete radio silence from Apple about the future of the Mac Pro, and even about the Mac more generally speaking. I figured that, if indeed they were to abandon the Mac Pro altogether, I might have to fall back on a beefed-up iMac as an alternative, and I wanted to see if it was feasible.

My conclusion after 10 days was that it was indeed feasible. It would not be ideal, because the performance level on the iMac 5K with a 2 TB Fusion drive was not quite on par with the 2014 Mac Pro, but it was not too slow, and I would always have the option to get an iMac with an SSD drive instead of a Fusion drive.

I must admit that I was also curious to see what it would feel like to work with a Retina monitor for the first time. I had first-hand experience with Retina on the iPad and on the iPhone, but not on a desktop computer. (I had never owned a Retina laptop either.) My view had always been that it was more important for me to have lots of screen real estate than to have a very high resolution.

I was able to set up the iMac 5K with my 4K Sharp PN-K321 monitor as a secondary display, and thus get a display setup that was somewhat similar to what I was used to on the Mac Pro (with the Sharp PN-K321 display as my main display and a 30″ Apple Cinema Display as a secondary display) — except with a better-quality picture all around.

Then I got my Mac Pro back (with a replacement video card that did nothing to fix the random video freezes — which turned out to be fixed by upgrading to Sierra later in December), I had to give up the iMac 5K. I didn’t really miss it, except… that I had been bitten by the Retina bug. There was just no denying that things looked much better in macOS with Retina, and so I started experimenting with my existing setup on the Mac Pro and exploring my options.

The first that I did was that I tried using the Sharp PN-K321 display in “scaled” resolution, i.e. in Retina mode. That meant sacrificing quite a bit of screen real estate, because I was now down to an effective resolution of 1,920 x 1,080 pixels (as opposed to the native 3,840 x 2,160 pixels), but I also had to admit that, even with the native 3,840 x 2,160 resolution, I was only able to have two document windows open side by side, with a high zoom setting (around 200%, to make the text readable at a comfortable level). Now I could still have two documents side by side, with a more normal zoom setting (a little over 100%), but the text looked undeniably better.

The one major drawback in this approach was that, since the Sharp PN-K321 monitor was my primary display, the whole macOS user interface was now much bigger and taking more space. I could use some settings in macOS itself or in third-party tools for tinkering with hidden macOS settings to try and minimize the impact of the switch, but there were still things that I could do nothing about. The menubar itself, for example, was now pretty big, with much less room for menu extras on the right. (I was already a Bartender user, but I had to move more items to the secondary bar.) Also, all the controls in individual applications were bigger. For certain things, it was not too much of a problem. Dialog boxes might have been larger, for example, but they didn’t have to stay open all the time. For other things, which had to stay open and visible all the time, it was more of a problem. The Word 2016 user interface and its “Ribbon” for example, now had much less space available to display controls, so that fewer controls were visible at any given time, and some only in compact form, as opposed to the more expansive form that I was used to. Similary, I had to reorganize my DragThing docks, my BBEdit palettes, my Adobe InDesign palettes, and so on.

Still, I was willing to live like that for a while, with the Sharp PN-K321 monitor in Retina mode as my main display and with the 30″ Apple Cinema Display at native resolution as a secondary display. (It also helped that I was able to fairly quickly adjust my Keyboard Maestro macros for shuffling windows around and resizing them automatically.)

My long-term goal, however, was to seriously investigate the possibility of acquiring a 5K display for my Mac Pro. I must say, however, that I was quite disappointed in the lack of results when I tried searching for information about this online.

For one thing, Apple’s own documentation on this is lacking. If you check the page titled “Use multiple displays with your Mac Pro (Late 2013)”, for instance, it still reads:

You can connect up to six of the following properly-configured displays to your Mac Pro (Late 2013).

  • Six Apple Thunderbolt Displays (27-inch), Apple LED Cinema Displays (27-inch), or third-party Mini DisplayPort displays.
  • Three 4K displays: two connected via Mini DisplayPort and one connected via HDMI.
  • One 4K Ultra HD TV or 4K display using HDMI and four Apple Thunderbolt Displays (27-inch), Apple LED Cinema Displays (27-inch), or third-party Mini DisplayPort displays.
  • Two HDMI (HD or 4K) devices: one connected via HDMI and one connected via Mini DisplayPort with an HDMI adapter.
  • Six DVI displays. This configuration requires an active DVI adapter.

Not a single mention of 5K displays on this page! Granted, 5K displays weren’t even available when the Mac Pro first came out, but they became available soon after, and there were enough mentions of people using a monitor such as the Dell UP2715K with a Mac Pro online to indicate that this was feasible. Of course, when you are talking about hardware products by companies other than Apple, the situation with the documentation is even more abysmal. Even today, if you do a Google search for “Dell UP2715K mac pro”, the very first result you get is… a knowledge base article from Dell titled “Issues Apple Mac platform users may encounter using the UP2715k Dell UltraSharp 27 Ultra HD 5K Monitor”. Not an article about how to do it, but an article about the problems you may encounter if you try to do it!

To make an already long story not too much longer, after reading all this and other stuff, I decided to take the plunge and order a Dell UP2715K for the Mac Pro. Based on what I had found online, the monitor itself would come with the dual DisplayPort-to-Mini-DisplayPort (DP-to-mDP) cable required to connect the monitor to the Mac Pro.

Unfortunately, when I looked at Dell’s own shop pages, I found that there was a page about the monitor, but it was listed as being discontinued and being replaced with a new product. (This was in January 2017.) But when I clicked on the link to see the new product supposed to replace it, it took me… nowhere. (I can no longer show you all this, because this page for the monitor in Dell’s own online store no longer exists. All that is left on the Dell web site is the support page for the product.)

I still found the monitor for sale at Amazon.ca and at NewEgg.ca. Amazon had several offers for it, including one where the monitor was sold by Amazon itself and a few others where it was sold by third parties through Amazon. Even though the Amazon.ca offer said “Usually ships within 1 to 2 months”, I was not in a hurry, and I have had nothing but good experiences with purchases from Amazon.ca, so I decided to go for that option. (It was also a bit cheaper than the other offers.)

Two months later, however, I got an automatic notice that the shipment was delayed by another month. I still wasn’t in a hurry, so I decided to wait. Once month later, same thing: shipment delayed by another month. At that stage, I decided that there was a good chance that it might never ship, and there were no further news about the “replacement” product from Dell either, so I decided to go with NewEgg.ca. By now, NewEgg had several offers from third parties, including one that had a pretty good price for the monitor (cheaper than the original Amazon offer) and a reasonable fee for shipping. I cancelled the order from Amazon and ordered that one instead.

One week later, the Dell UP2715K was delivered by UPS. It had everything in the box that it was supposed to have, including the dual cable that I needed. Of course, I knew that, it being a Dell product, it was probably not going to work without some effort and configuration. And sure enough, when I tried to connect it, nothing happened. I had disconnected the 30″ Apple Cinema Display with the mDP-to-DVI adapter and tried to connect the Dell while the Sharp was still connected, but I didn’t get anything.

I tried fiddling with cables, rebooting, etc. to no avail. Then finally I realized that the Dell was probably not smart enough to automatically detect what it was connected to, so I fiddled with the Dell’s own buttons on the side of the screen. Sure enough, even though I had used the two available DisplayPort connections on the monitor, for some reason it was configured to use something else as the “input” by default. (I cannot remember now whether it was HDMI or something else.) Once I told the Dell to use the correct source as input, I finally got something! But then all of a sudden I could no longer get my Sharp PN-K321 monitor to work. I had the 5K monitor as my main display, but the Mac Pro refused to see the 4K Sharp monitor as a secondary display.

Then, because of the lack of documentation about this online, I started to worry that maybe the Mac Pro was not able to handle both a 5K monitor and a 4K monitor at the same time via Thunderbolt/miniDisplay.

I thought that it was time to get on the phone with Apple and try and get some clarification on the whole situation. (Because I bought the Mac Pro in July 2014 with the three-year AppleCare coverage, I am still under warranty until July 2017, so I figured I would take advantage of it before it was too late.)

The first representative I talked to was predictably useless. But she was also humble enough to realize that this was beyond her level of expertise and she immediately referred me to a “special advisor” named Alan. Alan was pretty good and immediately realized that I knew what I was talking about. He didn’t have too much expertise on the Mac Pro with multiple displays himself, but he was able to do some digging and find another web page on the Apple site that seemed to be more up to date, and actually said that the 2013 Mac Pro had the following specifications:

Display Support

  • Connect up to:
    • Three 5K displays
    • Six Thunderbolt displays

That was more like it! But now the question was: If the Mac Pro did indeed support up to three 5K displays, how come I could not get it to work with one 5K display and one 4K display?

First of all, Alan drew my attention to the original page on the Apple web site titled “Use multiple displays with your Mac Pro (Late 2013)”. While the information was obviously outdated (compared to the other page) the diagram did indicate that, while the Mac Pro had six Thunderbolt/miniDisplayPort connections, all six were not interchangeable, and the underlying technology was based on three Thunderbolt buses only. Two ports were for Bus 1, two ports were for Bus 2, and the last two were for Bus 0, with the Mac Pro’s HDMI port also being on Bus 0.

What this meant is that, if, say, you did want to try to connect three 5K displays to the Mac Pro, you’d have to connect each display (with its two required mDP connections) to a different bus. It also meant that, even for a single 5K display, the two mDP connections for the display had to be on the same bus, and that presumably would free up the other two buses for other displays.

So we tried that while I was on the phone with Alan. I put the 5K display on Bus 1, and tried to connect the Sharp 4K display on Bus 2. It still didn’t work. Ditto with Bus 0. Alan had no idea why. Finally, he suggested trying to connect the Sharp 4K display via HDMI instead, just out of curiosity. I did have a spare HDMI cable, so I tried and… it worked! Success! I had two displays again… except that HDMI on the Mac Pro doesn’t support 60 fps for such a high resolution, so I had to use the 4K Sharp at 30 fps instead, which didn’t feel very smooth. It was tolerable for a secondary display, but not great.

Alan agreed with that, and said that he would have to talk to his sources in engineering to find out why I wasn’t able to connect the 4K display via mDP.

Then, out of curiosity (but also because I had always dreamed of having not just two, but three displays on my desktop), I asked Alan if we could try connecting the 30″ Apple Cinema Display with the mDP-to-DVI adapter back to the Mac Pro as well. But we got nothing. That didn’t seem right, since that monitor wasn’t even a 4K display. But something appeared to be seriously wrong since I had connected the 5K display.

Alan said to leave it with him and he would get back to me.

Meanwhile, because I am an experienced troubleshooter, I decided to try a couple more things. First of all, I noticed that, when I connected the 30″ Apple Cinema Display with the mDP-to-DVI adapter to the Mac Pro, the small white LED on the Cinema Display wouldn’t even come on, even when I pressed the Power button. (The Cinema Display has one of these frustrating touch Power buttons that have no physical state and no moving parts. You just tap on it and assume that it registers it as meaning “ON”. Tapping on it in this case, even repeatedly, produced nothing.) I also noticed that the connection between the power brick and the cable was a bit frayed. I had never noticed this before, but I wondered whether things were hanging by a thread and my moving things around one more time had finally caused the physical connection to give up the ghost, even though it only looked like the insolation around the electrical wiring was a bit exposed.

So I brought my old 2009 Mac Pro back up from the basement and tried connecting the Cinema Display to its DVI port directly. And… it worked just fine. The monitor came on, with the small white LED glowing, and everything was peachy. So the damaged physical connection to the power brick was not an issue, and the monitor itself worked fine. Could there be something wrong with the mDP-to-DVI adapter, then? After all, when I checked the reviews for the product on Apple’s web site, they were far from glowing, and several people complained about failures after a few months…

Well, I had one way to find out: I could try plugging the Cinema Display with the mDP-to-DVI adapter to my wife’s iMac 5K upstairs. I tried that, and… it didn’t work. The Cinema Display had the same symptoms as with my Mac Pro: No picture, no white LED coming on, no matter how many times I pressed that frustrating touch power button.

So it started to look like the mDP-to-DVI adapter had all of a sudden died on me, at the same time I had decided to set up a new 5K display for my Mac Pro. Too much tinkering for its taste?

Then my troubleshooter’s mind continued to whir and I started wondering about the cable connection between the Sharp 4K and the Mac Pro. One thing that made me a bit suspicious at that stage was that, for the past year or so, I had been experiencing weird problems with my Sharp 4K display and the Mac Pro: Sometimes, when I would wake the display from sleep, only half of it would come back on, and the other half would stay black. When that happened, even rebooting the Mac Pro didn’t help. The only solution that I had found was to physically unplug the DP-to-mDP cable from the Sharp to the Mac Pro on the Sharp’s side and plug it back in. I could even do that without rebooting the Mac Pro. (Thunderbolt/Mini DisplayPort is plug-and-play.) The only problem is that, of course, when I did that, macOS would reshuffle all my windows, because I would be temporarily back to one monitor only. So I would have to manually reorganize my windows each time this happened, which was an annoyance. Then in more recent months I had noticed a new problem with the Sharp 4K display and the Mac Pro: Sometimes, while I was working, half of the Sharp screen would go black for a few seconds and come back on, or the entire screen would go black and then come back on. I could go days without a single incident, or sometimes it would happen several times a day. Rebooting the Mac didn’t seem to make any difference. Since this particular had started more recently, I had good reason to believe that both issues (the screen staying half black after sleep, and the screen going half black or completely black randomly) were actually software bugs, the second one having been introduced by a more recent system update. (I know from experience that bugs that only affect a small subset of users can fly under the radar for a long time, so nothing surprises me anymore.)

But now, in light of these new developments coinciding with my purchase and installation of the new 5K display, I started wondering whether there was something else going on, and maybe it was a hardware problem with the Sharp 4K itself or with the cable. The only way to find out for sure was to test the Sharp 4K with a different cable… except that I didn’t have another mDP-to-DP cable. I only had the dual mDP-to-DP cable for the Dell 5K, and that single mDP-to-DP that originally came with the Sharp 4K and now looked somewhat suspicious to me. I could probably use one half of the dual mDP-to-DP cable with the Sharp 4K, but then I couldn’t use the Dell 5K at the same time. And since I could no longer use the Cinema Display with mDP-to-DVI adapter with the Mac Pro either, I was kind of stuck… I could order a new single mDP-to-DP cable online, of course, but it would take at least a few days to arrive, and I wanted to test things right away.

So I did the only thing that I could do, which was to take my Sharp 4K display upstairs and try to use one half of the dual mDP-to-DP cable from Dell to connect it to my wife’s iMac 5K. And guess what? It worked like a charm.

This seemed to indicate that the Sharp 4K itself was not the culprit, and that the problem was indeed with the single mDP-to-DP cable that originally came with the Sharp display, which might have been defective from the start or at least for a long time (hence the problems I had been having with the Sharp 4K and the Mac Pro, with half of the screen going or staying black) and might just now have died altogether.

There was only one way to prove this conclusively, which was to order a new mDP-to-DP cable from NewEgg. I figured that, while I was at it, instead of ordering a new mDP-to-DVI adapter from Apple, which was expensive, I would also order a cheap DVI-to-HDMI adapter from NewEgg and try to use that to connect the Cinema Display to the Mac Pro (via the HDMI port, which is on Bus 0). I could have the Dell 5K with the dual mDP-to-DP cable on Bus 1, the Sharp 4K with a new single mDP-to-DP cable on Bus 2, and the Cinema Display with an HDMI adapter on Bus 0. That should work.

I ordered the stuff. It came within a week, and I was able to confirm everything, i.e. that the single mDP-to-DP cable from Sharp was indeed toast, and that the Cinema Display worked with the Mac Pro via HDMI… except that this flavour of HDMI (at least) is limited to a 1920 x 1080 pixel resolution, which of course is significantly lower than the Cinema Display’s native resolution.

I got back to Alan with all this information (which meant that I had effectively done his work for him, but never mind…). We were both a bit shocked by the very weird fact that both my single mDP-to-DP cable from Sharp and my mDP-to-DVI adapter from Apple seemed to have failed at the same time, and also at the same time that I decided to purchase and install a new 5K display for my Mac Pro! I mean, what are the odds? (Maybe engineers familiar with these technologies are aware of vulnerabilities that can come to the surface when trying to fiddle with stuff. I don’t know…)

Finally, I had a decision to make. I now knew for a fact, not only that I could have both the 5K and the 4K displays with my Mac Pro (which was already great), but also that my Mac Pro could support even more and could actually have a third display as well, which has always been a dream of mine. (As far as I am concerned, the more screen real estate, the better.) But the Cinema Display via HDMI had a limited resolution. In order to get the full resolution, I would have to get another mDP-to-DVI adapter. It wasn’t a cheap solution, and the reviews about Apple’s adapter were not too reassuring. So I took another leap of faith and decided to order yet another new monitor for my Mac Pro.

Of course, I wasn’t going to order another 5K display. For one thing, it would be too expensive. Then it would use up two more Thunderbolt/Mini DisplayPort connections on my Mac Pro, leaving me with only one left for hard drives. (The displays don’t come with their own Thunderbold/mDP ports for daisy chaining.) I wanted to have at least two connections free for hard drives (one for my drives in the basement and one for portable drives that I need to connect from time to time). Finally, a 5K resolution is only really important for the main, central display that I am looking at most often. For displays on the side, which I only look at more occasionally, it is not nearly as significant. In an ideal world, of course, all my displays would be 5K. But in the real world, with the state (and cost) of technology as it is today, having two or three 5K displays is overkill and makes little sense.

I settled for a Dell P2715Q instead. Physically, it’s almost identical to the 5K display. It has a 4K resolution. It had good reviews. And it wasn’t too expensive. I thought that, as another secondary display, it would do just fine.

I went to order it from NewEgg and… noticed that, while they still listed the Dell 5K, all their offers were “currently out of stock”. So I didn’t really have the option to get another 5K display anyway… (I’ve since seen indications online that the only current option for a 5K display for the Mac is the recently released LG 5K display, which already has a history of problems that does not inspire confidence, and — officially, at least — only works at 5K with recent MacBook Pro laptops anyway.)

So I ordered the Dell 4K fro NewEgg.ca, and it came within a few days. I was pleasantly surprised to discover that it had a more old-fashioned matte finish, as opposed to the glossy finish of the Dell 5K. Glossy is fine for my main display (even though I don’t particularly like it), but I realized, upon installing this third display, that if it had had a glossy finish, that would have been a bit of a problem, because this third display on the left had to be on an angle and would therefore catch some of the light coming from my office window. So I was quite pleased that it had a matte finish.

As expected, this third display worked with a single mDP-to-DP cable on the Mac Pro’s third bus. I had to tinker with the resolutions of the two side displays for a while in order to find the optimal setup, but I finally ended up with this arrangement:

3displays2017

The central display is the Dell 5K at the “Default for display” resolution, which is Retina, i.e. an apparent resolution of 2560 x 1440 pixels (actually 5120 x 2880 pixels).

The Dell 4K on the left is at a scaled resolution of 2560 x 1440 pixels, which is not Retina, since the display’s native resolution is 3840 x 2160 pixels, but still looks almost as good as Retina when looked at from an angle at a greater distance from the eyes than the Dell 5K at the center of my desk.

And the Sharp 4K on the right is also at a scaled resolution, of 3008 x 1692 pixels, which, again, is not Retina, since the display’s native resolution is also 3840 x 2160 pixels, and is also not the same as the 2560 x 1440 resolution of the Dell 4K, because the Sharp 4K is physically bigger and can display more. Again, it doesn’t look as good as Retina, and for small font sizes there is a bit of eye strain on occasion, but on the whole, for most real-world applications, it works fine, and does look fairly close to Retina from an angle and at a greater distance.

The most important thing for me in this setup is that there is no significant change in the size of the UI elements when moving from one screen to the other. With these resolutions, it really does look almost like one extra-wide continuous desktop. I can easily move things from one screen to the next either manually or with Keyboard Maestro macros. It is definitely a significant improvement over my previous two-display setup, not just because of the improved quality of the displays, but also because there is much less of a discrepancy between the main display and the other displays.

In spite of my initial fears about buying non-Apple displays and about the risks associated with purchasing equipment whose use with a Mac Pro was not very well documented online, I must say that the setup was relatively painless. (The major pain described above was caused by the defective cable and the defective adapter. Without these defects, the transition would have been significantly smoother.) The only slight problem I encountered was that the default configuration of the Dell 5K included some sleep setting that didn’t agree with the Mac’s own sleep feature. I was sort of prepared for this by what I had read online, so I knew right away that I probably had to tinker with the Dell 5K’s own internal menu feature. I used that feature to turn the Dell 5K’s own sleep feature off, and now the display seems to go to sleep properly whenever I use macOS itself to put all my displays to sleep (via a hot corner).

The Dell 4K didn’t have any such problem and worked properly with macOS’s sleep feature right out of the box.

And how does it feel to work with these displays on a daily basis? Well, I have to admit that there was a period of adjustment when I first got the Dell 5K display. I was underwhelmed by the brightness levels that it was capable of (compared to the Sharp 4K) and the glossy finish took a little while to get used to. I had to turn the brightness up to a level that is very near the maximum that it is capable of, and I also had to turn the brightness of the Sharp 4K down a bit. The Dell 4K is pretty close to the 5K in term of its brightness. The whites on the Sharp 4K are a bit bluer than on the Dell displays, but I was able to calibrate things so that all three displays are pretty close in that respect, and indeed they are much closer to each other than my Sharp 4K and my Apple Cinema Display ever were. (The Apple Cinema Display’s whites were always much yellower than the Sharp 4K’s, no matter which calibration I used.)

Now that the Dell 5K has been discontinued, I am very happy that I managed to buy one when I did. Apple might come up with its own new range of displays in the future, but I am far from convinced that they’ll make any effort to support older Macs when they do. And since no other option is currently available, I am really glad that the reseller that I bought the display from through NewEgg.ca still had one in stock.

Thanks to this hardware upgrade, and to the relatively painless upgrade to macOS Sierra in early 2017, it really feels like my 2014 Mac Pro has been given a new lease on life, and this is almost as good as getting a new computer. It means that I certainly won’t have any problem waiting for Apple to come up with a new Mac Pro next year and won’t feel any pressure to be an early adopter of that new machine (as long as nothing breaks in this Mac Pro).

The 2013 “trashcan” Mac Pro might not have been a great success, but it works well for me, with the adaptations that I was able to make, namely with the use of external Thunderbolt hard drives relegated to the basement via a 30-foot Thunderbolt 2 cable so that I don’t have to hear them, and now with the use of a three-display setup that includes one central Retina display with a decent amount of screen real estate (unlike a 4K display used in Retina mode, which really doesn’t provide enough screen real estate) and two side displays with near-Retina resolutions.

None of this was cheap, but in my line of work, my computer is my main business expense and it is worth spending the extra amount to get something that I am really comfortable working with.


Editing a formula in Numbers: an example of macOS’s failure to buffer quick user actions

Posted by Pierre Igot in: Macintosh
May 4th, 2017 • 3:20 pm

Here is a very simple example of something that is fundamentally wrong in macOS in its current incarnation.

STEP 1: In the current version of Numbers for macOS 10.12 (Sierra), which is Numbers 4.1.1, select a cell:

numbers4-speed1

STEP 2: Now press the = key on your keyboard, which is a shortcut for inserting a formula in the cell:

numbers4-speed2

(In passing, note the mouse cursor, which is not right either. In its current position, the mouse pointer should not be an I-beam, but an arrow. But that’s another long-standing issue in macOS, which has been well documented, not that it has made any difference so far.)

STEP 3: Now click on the cell above. This inserts a reference to the cell in the formula that you’ve started editing:

numbers4-speed3

Great! Everything is peachy, right?

Er, not quite. See, there is a slight problem if you happen to be a fast user who makes full use of his/her two hands to avoid wasting time on repetitive tasks. If you go too fast between STEP 2 and STEP 3, Numbers completely fails to insert the cell reference in the formula. It completely ignores your mouse click in STEP 3. It’s as if you never did click.

How on earth is this possible? Can I really be too fast for my computer? How can a 2014 Mac Pro with tons of RAM be incapable of keeping up with my very human motor skills, no matter how fast I am?

If, for whatever reason (e.g. because of a background process), Numbers was slowed down and had trouble keeping up with me, at the very least it should be able to record my actions in a buffer and execute them once it’s back to its normal state. Yet there’s nothing significant going on in the background (I can reproduce this at will, under any circumstances), and it clearly is not even buffering my actions!

There’s something fundamentally wrong with this. And, before you ask, it’s not an isolated problem, a problem that might be limited to Numbers and to this particular scenario.

I constantly encounter situations, in my daily use of my system, where macOS is unable to keep up with my user actions, whether it is while I am typing text on the keyboard or executive a sequence of actions with the mouse or keyboard shortcuts — and where it also fails to buffer the actions that it appears to be unable to keep up with.

The end result is that macOS is constantly losing track of my user actions and forcing me to repeat them — i.e. to delete the phrase that it mangled (by missing some letters even though I did type them) and retype it, to undo accidental mouse actions that happened because macOS missed a step or two, etc. — thus severely disrupting my workflow.

How can this be acceptable? How can Apple not have noticed this pervasive problem? How could this even be treated as “good enough” by Apple?

My guess is that, as always, there is a confluence of reasons explaining this situation.

First of all, there is the fact that it is intermittent, and impossible to reproduce at will. Right there, it means that the bug, no matter how serious it is, will not be a top priority for Apple. There’s a flaw in their very system for identifying bugs and dealing with them that makes intermittent bugs, no matter how serious they are when they do occur, much harder to report, get processed, and get fixed.

And not only is it an intermittent bug, but it’s also hard to report, even by experienced bug reporters such as myself. How do you provide evidence of user actions that the system does not even seem to have detected? When I type a phrase in a word processor and I know that I have typed it properly, without skipping any letters and without making any typos, and yet what appears on the screen does not match what I typed, how do I prove this to Apple? Nobody is here next to me filming my fingers on the keyboard and ensuring that I did indeed type what I said I typed. Similarly, nobody is there monitoring my every mouse click or keyboard shortcut, and keeping an exact chronological log of all the user actions that my hands are executing.

I also suspect that part of the reason why this problem is underreported and not treated seriously enough by Apple might be that today’s hardware is, paradoxically, fast enough to mask the problem fairly well, so that it doesn’t happen often enough, or in a way that is egregious enough and common enough to prompt action on Apple’s part. It’s telling to me, for instance, that the application where it happens most often for me is… Microsoft Word 2016, which is, generally speaking, the most sluggish, most unresponsive piece of macOS software that I have to use. So the underlying general sluggishness of that particular application appears to exacerbate the problem and, sure enough, that’s the application where it occurs most frequently. But it’s by no means the only application where it occurs. I see the problem even in Apple’s own macOS applications, including Mail, TextEdit, and Safari, only less often than in Word. And I bet there aren’t many engineers at Apple who use Word on a regular basis. (Let’s not even wonder here how many Microsoft engineers use Microsoft Word for macOS for any serious work. If they did, they would probably die of shame.)

In addition, I believe that the fact that I use certain third-party tools, such as Typinator, to partially automate text entry in all my applications, causes me to be more vulnerable to the issue. Typinator lets me type abbreviations and automatically replaces them with expansions, and it also automatically corrects my most frequent typos. I have hundreds of abbreviations memorized and corrections recorded, and I use them all the time. It makes me much more productive. But I cannot deny that I often notice hiccups in the flow of text on the screen during those automatic expansions and corrections. And typically, when this happens, while the expansion or correction eventually happens after a second or two, macOS inserts a few more of the characters that I typed after the abbreviation, and then there is a whole bunch of characters missing, which macOS somehow failed to buffer. I have discussed the issue with the Typinator developers at Ergonis and they argue that they are unable to reproduce the problem themselves, and that, in any case, even if some hiccups occurred, macOS should buffer the rest of the characters properly and spill them out eventually. I don’t think that the Ergonis folks are trying hard enough to reproduce my problem, for instance by using hundreds of abbreviations and corrections all the time like I do, while typing very fast, but I also believe that they are fundamentally right and that, even if it makes the problem more obvious, their software cannot be blamed for what appears to be essentially a system-wide failure to buffer user actions — including text input and keyboard shortcuts and mouse actions — properly.

But of course, how can I convince Apple that this is serious, in light of all the above factors? I am quite sure that there is not a single macOS developer at Apple who uses Typinator like I do, in Mail, TextEdit, Safari, Microsoft Word, Nisus Writer Pro, BBEdit and all kinds of other applications. In fact, I am quite sure that there are very few people among Apple’s engineers or even Apple’s testers who are actual writers and who strive to be as productive in their writing as I do. Some of them might have noticed something a bit odd here and there, but because it’s not exactly a deal breaker (like an application crash or a system freeze) and it’s so hard to circumscribe properly, it’s not particularly surprising that nothing is being done about it (as far as I can tell).

That is why I chose to start with the simple, obvious, easily reproducible problem in Numbers described above. It’s far from being the most pressing issue for me (I don’t execute this particular sequence of actions more than a few times a day maximum, and most of the time I am not fast enough to encounter the problem). But it’s symptomatic of a larger problem, and I had to start somewhere. Of course, there is a risk that the problem in Numbers will be seen by whomever reads this as isolated and relatively unimportant, but that’s beyond my control (although such a reaction would be symptomatic of what’s wrong with the Mac today, where “attention to detail” has become little more than an old myth).

It’s possible that, if I stopped using Typinator altogether in my daily computing, I would experience the buffering problem much less frequently. But how can this be justified? Typinator is an excellent tool that greatly enhances my productivity, by enabling me to type thousands of words per day. I wouldn’t be nearly as productive without it. It works fine most of the time. Also, I just cannot accept the fact that a machine as powerful as my Mac Pro, with all its processing power, cannot handle a task as simple as properly buffering all my user actions without exception. This, to me, is a fundamental flaw in macOS, and one that must be fixed.

I just don’t know how to get Apple to acknowledge it and do something about it.

I know for a fact that I am not the only one experiencing hiccups in text input. Just the other day, I was discussing it with a friend of my wife’s, who was staying with us for a week. She is retired and has been using an iMac for several years. When I mentioned my frustration about the fact that I had such a powerful machine and that it couldn’t even keep up with my typing, she confirmed that she too had noticed unexplained hiccups in text input on hers. And she doesn’t use any kind of third-party utility like Typinator! She’s not a power user by any stretch of the imagination. She doesn’t type lots of text all day long. And yet she too has noticed it… (I didn’t clarify with her whether her observations also included buffering problems, i.e. not just hiccups, but actual characters failing to be recorded by the system and vanishing into the ether. One cannot expect ordinary users to be so in tune with what’s going on in their system to clearly notice and understand such things.)

How many other Mac users are there out there who are aware of this problem and as annoyed by it as I am? It all started in Lion, if memory serves, on my 2006 Mac Pro, and that’s a long time ago. Still, nothing has been done to fix it, as far as I can tell, and, two other Macs and several macOS upgrades later, it’s still there.

What is it going to take?


macOS 10.12 (Sierra) on Mac Pro: A solid release

Posted by Pierre Igot in: Macintosh
January 24th, 2017 • 2:53 pm

As you know if you’ve read my recent blog posts, I’ve had a bit of a torrid time with my 2014 Mac Pro in the past few months, because of a recurring problem with random video freezes.

It turns out that, contrary to what I had believed earlier on, the problem was not hardware-related, but software-related (and possibly related to a recent security update, more specifically).

While my Mac Pro is working more or less normally again now, this whole ordeal had two very concrete consequences for me.

First of all, it (indirectly) led me to purchase a new iMac 5K for my wife as a replacement for her aging 2009 Mac Pro, which gave me the opportunity to discover first hand the joys of desktop computing with a Retina display. This prompted me to experiment with my own existing hardware setup (the 2014 Mac Pro driving a 4K screen and an older 30″ Apple Cinema Display), by switching the 4K monitor from its native 3840 x 2160 resolution to the scaled 1920 x 1080 resolution that effectively gives me a Retina display for my main screen.

It took me a little while to adjust to the reduced screen real estate and to fine-tune my virtual work environment accordingly, but I am quite happy with the results. That said, I am still planning on purchasing a 5K monitor to replace my current 4K + 30″ setup with a 5K + 4K setup that is entirely Retina. I was considering the Dell UP2715K, but it looks like it’s being discontinued and replaced by something else, although what that “something else” is going to be is not clear at all. (The site says, “Please see our recommended replacement product”, but then it does not show what that replacement product is, as far as I can tell.)

I feel that the switch to Retina will give my 2014 Mac Pro a bit of a new lease on life. It looks like the most egregious bugs related to Retina have been addressed, and the computer has enough power to handle the new setup. I’ll be able to say more when I have made the full switch.

The second major consequence of this mishap with the Mac Pro is that it forced me to upgrade to Sierra sooner than I was planning to (by a few weeks at least, since the 10.12.3 update has actually just come out).

After my very traumatic experience with early versions of Yosemite and my much smoother transition to El Capitan, which I only upgraded to when it had reached 10.11.3, I was planning on waiting for 10.12.3. But the video freeze issue in El Capitan forced me to upgrade sooner.

I am happy (and relieved!) to report that the upgrade from El Capitan to Sierra 10.12.2 was quite smooth, and that I am positively surprised by the overall solidity of the new system. After the upgrade, I was able to run my machine for more than a week without having to restart it a single time for any reason, and most things run as well, if not better, than they did in El Capitan.

I was a bit worried about the PDF-related issues reported with Sierra, because I use all kinds of PDFs on a daily basis in my work, and Preview is my preferred PDF viewer. I must admit, however, that I personally haven’t noticed any significant differences or encountered any bugs. On the contrary, it looks like a bug introduced in Yosemite’s Preview in 2015 when resizing a window that contains a PDF document too fast has finally been fixed. This does not mean that the PDF bugs affecting other people are not important, of course. It just means that, for the types of tasks that I have to complete with PDF documents, Sierra seems to be working fine.

I am also happy to report that Sierra fixes other problems that had plagued my computing environment for a long time. It fixes several sleep-related issues that I had. For example, even with Power Nap enabled, I was unable to get El Capitan’s Mail to keep checking for new mail all through the night. Typically, I would wake my computer in the morning and find that it still had to download a bunch of messages that had arrived through the night. This is important, because SpamSieve needs to have Mail checking for new mail regularly on my desktop computer in order to filter out spam on my iOS devices.

I also had, for years, an issue where Typinator would be unable to reliably play some specific sound effects after my computer had been put to sleep. This meant that, for years, I had to quit and relaunch Typinator each and every morning. (I had a Keyboard Maestro macro for this, of course, but it was still annoying.) This audio problem appears to finally be gone in Sierra.

I didn’t encounter any new incompatibilities in Sierra, even with old software that I am still using and that will never be updated again, such as NetNewsWire 3, from 2012, and the CD-ROM-based French dictionaries Le Grand Robert and The Collins-Robert French Dictionary, also from 2012. (Time will tell how far into the future these continue to work, but I really do not feel like giving the Le Robert company any more of my money for a subscription to its online versions, based on the very poor customer service that they offer and the poor quality of their software in the first place. I only use these dictionaries because they are valuable as dictionaries. As for NetNewsWire, I suspect that one day I’ll have to find another RSS reader, but I am OK with delaying that process for as long as possible.)

I was quite disappointed to see that Sierra does not fix a number of long-standing issues in the Finder. The text-label highlighting in background windows is still broken in column view when dragging and dropping files. I don’t know how much more obvious this bug needs to be for Apple to finally do something about it. Similarly, another bug that is much harder to reproduce and that sometimes causes the Finder to completely fail to highlight the destination of a drag-and-drop operation altogether is still there. Since that one is intermittent, I suspect that Apple is still in complete denial about its very existence.

I am still occasionally experiencing significant problems in Sierra with the mouse pointer failing to change when it is supposed to. This can affect all applications systemwide. But it’s also an intermittent problem, and so, until I can find a 100%-reliable scenario to reproduce it, I am afraid I am going to have to continue to live with it.

In fact, after I posted about my disappointment with the persistence of these issues in Sierra on Twitter, I got a reply from @AppleSupport asking me to contact them by direct message to further explore the issues. (I didn’t direct the original tweet at @AppleSupport. I only used the handle @apple to refer to Apple in the tweet. I guess that’s enough to prompt a canned response.) I did respond to the invitation, and ended up having to explain things to a “senior advisor” on the phone. He did agree that he could reproduce the highlighting issue in the Finder and said that he would forward an “FYI” (his word) to engineering about it. He wasn’t able to reproduce the mouse pointer issue, so he asked me to try and reproduce it in a clean user environment and phone him with the results. Of course, I cannot reproduce the problem reliably, so that’s a dead end.

The process was rather time-consuming and yielded dubious results. I won’t go through it again. @AppleSupport needs to do a better job of distinguishing between calls for help and reports on bugs. Since they don’t seem to be able to do so, I guess I shall refrain from referring to Apple by its Twitter handle in the future. I’ll stick to Bug Reporter, even though that’s time consuming too (those huge uploads of system data!) and also has dubious results.

I am also disappointed to report that the hiccups that have been affecting text input on the Mac since Lion still haven’t been eliminated. I regularly encounter them in applications such as Mail and TextEdit, and they are particularly bad in Microsoft Word 2016, which is such an atrocious piece of software to begin with.

Microsoft seems to be simply unable to produce an application for the Mac that is both powerful and lean. Word 2016 might be powerful, but it’s a clunky monster and it’s slower than any other Mac application for the most basic stuff, including text input. Things are so bad in Word that the hiccups that are clearly a flaw in the operating system itself can significantly interfere with your typing and can even cause the application to completely lose track of what you’ve been typing altogether, effectively letting clusters of letters vanish into the ether. (They are not even buffered properly!) The combination of text input flaws in the operating system and Word’s own clunkiness is simply disastrous.

While things are not worse, in that respect, in Sierra than they were in El Capitan (where they were marginally better than in Yosemite), they are not significantly better either, and whenever I have no choice but to work in Word, I fear the worst, most of the time with good reason.

It continues to boggle my mind that, with all the power that we have under the hood in a machine such as a 2014 Mac Pro, we still have to deal with responsiveness issues of the most basic sort. How on earth is a 2014 Mac Pro not fast enough to process my typing in real time, no matter how fast I type — I’m only human, after all — and no matter what other processes are happening in the background at the same time?

This remains, to me, one of the most shocking developments in recent years in Macintosh computing. Clearly, Apple’s own priorities lie elsewhere, and they don’t care enough about the needs of Mac users whose primary use of their computer is to compose text by typing it.

Siri on the desktop? I couldn’t care less. I don’t even a microphone for audio input! (It’s also worth noting that even disabling Siri doesn’t stop Sierra from wasting resources on things such as background “speech recognition” processes…)

Basically, what I would like is a system that always processes text input in real time, with no exceptions, and lets me view text in full Retina glory on large, bright monitors. Once I get my 5K monitor, I might be closer than ever to that ideal, but until Apple finally addresses text input responsiveness issues in a real way in macOS, I will still have to endure a life of ongoing frustration with the poor use of hardware resources by my software tools.

That being said, Sierra remains, as far as I am concerned, a solid release. It’s no solid enough to make me fully trust Apple again, and I will certainly remain wary of future system updates and upgrades, but at least my system today is as good, if not slightly better, than it was a year ago. And I guess that, the way things are going, that’s a minor miracle.


Video freezes on Mac Pro: It was software after all

Posted by Pierre Igot in: Macintosh
January 21st, 2017 • 6:35 pm

I have an update on the situation with my Mac Pro that I blogged about in December

Back then, I indicated that the unpredictable video freezes that I had been experiencing for a while appeared to be gone after I had had one of my video cards replaced by Apple under warranty.

If you follow me on Twitter, you know that this is not where the story ends.

In fact, after the video card swap on December 11, the Mac behaved properly for exactly two weeks and then… out of the blue, I started experiencing video freezes again, on December 25. (Yes, on Christmas day!) The symptoms were exactly the same again: normal behaviour for a long time, and then, all of a sudden, a cascade of temporary video freezes, accompanied by the display of this error message in the system log:

Mac-Pro kernel[0]: stalling for detach from AMDTahitiGraphicsAccelerator

The video freezes would last several minutes, and then things would go back to normal… until the next time, which could be 15 minutes later, 12 hours later, or 72 hours later. There was no way to predict it.

How was this possible? This clearly indicated that the problem was not fixed by the video card swap, and that the “purple tinge” that the Genius Bar staff member had noticed with my video card when using his advanced testing tools was unrelated to the actual problem. Swapping the video card might have fixed the “purple tinge” (which I never experienced myself under normal conditions), but it didn’t fix the video freezes.

Needless to say, since the video freezes started happening on Christmas Day, I didn’t do anything about them right away. Apple’s services were closed on December 25 and 26 anyway, and I figured that they would be swamped with calls on December 27 from people with trivial problems with their new Apple-made Christmas gifts. So I knew I had to wait a few days.

I also was on holiday and didn’t exactly relish the thought of spending several hours of that time off on the phone with various Apple representatives trying to convince them to do something about the problem.

So I just continued using my computer normally, trying to put up with the freezes and not let them drive me mad. Then, the next time they happened, I noticed something that I hadn’t noticed before. One of the ways that the problem first manifested itself was not necessarily with a complete video freeze of the entire interface (except the mouse pointer), but with a web page failing to load in Safari. A page that I had just visited a moment ago and wanted to go back to wouldn’t load. Instead, I would be faced with a blank window, with no explanation. My Internet connection was still working fine, but Safari would refuse to load/display the web page. Somehow other web pages in other tabs or windows would still work fine for a while… And then eventually the video freezes would extend to the entire interface, and I would have to go through the entire thing again.

When this particular version of the problem occurred, it jigged my memory, and I remembered something that I hadn’t thought about in a while, which was that I had noticed a similar problem (web pages failing to load for no apparent reason) long before I started experiencing the video freezes affecting the entire user interface for extended periods. It had been fairly rare, and at the time, I had not tried to troubleshoot the problem, precisely because it was fairly rare and I had simply filed it under the “General Crappiness of Apple Software These Days” category (subcategory: “Impossible-to-Reproduce Crappy Software Problems for Which Bug Reporting Would Be a Waste of Time”).

But now, this new episode led me to think that, maybe, just maybe, the video freezes that I had been experiencing lately were just a more severe and intrusive manifestation of a long-standing problem… which in turn led me to think that, maybe, just maybe, it wasn’t a hardware problem after all.

I obviously couldn’t know for sure, but the theory that the video freezes were a software problem was worth revisiting…

But what could I do about it if it was a software problem? Well, I was still running El Capitan at the time, because my experience with early versions of Yosemite back in 2015 had made me extremely wary of being an early adopter of major system upgrades. My original plan had been to wait until version 10.12.3 of Sierra before trying it on my Mac Pro. Sierra was currently at 10.2.2. What if I took the chance and tried to install Sierra on my Mac Pro now, in order to see if, by any chance, the upgrade might, inadvertently or not, contain a fix for the video freezes?

I figured that I didn’t have too much to lose at that stage, and that if I didn’t do it now, there was a good chance that Apple’s representatives would ask me to try to do that anyway as part of the process of further troubleshooting the freezes.

So I took all the usual precautions (full backup, application upgrades that hadn’t been done yet, reading about potential problems with Sierra) and then I took the plunge, on December 26.

Well, today is January 21, and I am happy to report that I have not experienced a single recurrence of the video freezes accompanied by the “stalling for detach” error messages in the system log.

So it looks like Sierra (at least version 10.12.2) does contain a fix for the problem. As you can imagine, it’s a huge relief for me. I don’t have to go through the time-consuming process of trying to get Apple representatives to acknowledge the seriousness of the situation and provide me with a solution. I have a computer that works well again, and does not have a hardware flaw after all.

There is an epilogue to this story, however. Since December 26, I have heard from a fellow Twitter user who has been experiencing similar problems, albeit on a completely different machine (a MacBook Pro). She too has been experiencing video freezes, along with “stalling for detach” error messages in the system log. (In fact, it’s through a Google search for this phrase that she found my blog post about my problem.)

In her case, her MacBook Pro has video freezes and displays the following message in the log:

stalling for detach from AMDVerdeGraphicsAccelerator

Same message with a different flavour of graphics accelerator… We now had a clearly strong sign that the problem was indeed software and not hardware. (In retrospect, the very fact that the freezes were accompanied by error messages in the log was a sign that it was a software problem. Hardware problems typically occur at a level that the system is not even aware of, and cause freezes or kernel panics without a trace of the reason why in the logs.)

Still, like me, she was told by Apple that her video card might need to be replaced. She had just got it replaced, and based on my experience, I now warned her that there was a good chance that it wouldn’t fix the problem. Sure enough, two days later, she had more video freezes.

I told her that I had upgraded to Sierra and that it seemed to have fixed the problem, but that it was still too soon for me to tell for sure. (Now that nearly a month has elapsed, I am quite sure.)

She decided to stick with El Capitan for a while, because she was wary of the PDF-related issues reported with Sierra. (To be honest, the reports of bugs with Preview had also worried me quite a bit before then. However, while I do use PDFs in my work quite a bit, I haven’t really noticed any significant problems in my workflow with Sierra yet, and I have been back at work for two weeks now. Of course, your mileage may vary…)

And then…

On January 17, Apple released an updated version of Security Update 2016-003 for El Capitan, dubbed “Security Update 2016-003 Supplemental”. And guess what? The release notes indicated that the update to the update addressed “a kernel issue that could cause a Mac running El Capitan to occasionally freeze and become unresponsive”. Mmm…

Of course, when my own video freezes started getting serious, in November 2016, I did wonder whether any recent system update might have been the culprit. But the only update that I had installed during that time was an early build of that very security update, which I got as part of my involvement in the Apple Seed program. And since it was “only” a security update, and since my problem clearly seemed to be a video problem, I figured that there was no way that the two were linked. (I actually remember mentioning that update to the Apple senior representative who had taken care of me in December. He didn’t think there was a connection either.)

Now, of course, in light of these new developments, I do have to wonder whether my own video problems were not caused by that update. The timing seems to confirm it, except for the fact that I do remember experiencing the occasional blank page in Safari long before last fall… Still, it is also possible that there was an underlying problem before the security update, but that the security update just made it way, way worse, at least for some Macs.

Also, when you look at the details about the security update that Apple provides on this page, you can see that it contains fixes for security bugs that do involve things like “AppleGraphicsPowerManagement”, “CoreGraphics”, and “CoreMedia External Displays”. So it is not too far-fetched to think that the security update might have introduced (or amplified) low-level issues that did involve the video part of the system.

My Twitter friend has now installed the Supplemental version of the security update and, so far, she hasn’t experienced any further problems. It’s probably still a bit too soon to tell for sure. (I’d give it a whole month myself to be certain.) But it now appears quite likely that the video freezes that we both experienced, accompanied with these “stalling for detach” error messages in the system log, were not hardware problems, and were indeed either caused or made much worse by the original version of Security Update 2016-003, and that Security Update 2016-003 Supplemental does fix them.

Still, both she and I had to go through the process of dealing with (relatively) clueless Apple representatives and had to waste time getting our computers serviced for nothing. We might have got brand-new video cards out of the deal, but that does not, of course, compensate for all the grief caused by the bug.

And, as usual, I cannot help but wonder whether there is really nothing that Apple can do to improve its process for dealing with serious bugs. If this problem was indeed caused by Security Update 2016-003, why were the Apple representatives that we dealt with not aware of it? Why did they instead suggest useless hardware repairs? And why was the bug fixed silently without telling us? Where is the knowledge base article on Apple’s web site that describes the problem that we experienced in full detail, with clear reference to the exact symptoms and to the “stalling for detach” error messages in the system log?

What if other people are still experiencing the exact same problem and are looking for help today? How will they know to install Security Update 2016-003 Supplemental, especially if they are inclined to avoid installing new updates right away, precisely because they don’t trust that Apple can produce bug-free updates in the first place?

Something is rotten here, and other reports online about this lack of transparency tend to confirm it. Apple needs to realize that bugs cost us time, therefore money, and that, unlike their own representatives and senior representatives, we don’t get paid for the time we spend trying to work around these bugs or get them fixed.

Will we ever get a new attitude at Apple that not only transparently communicates about such crucial bugs, but also apologizes for the significant inconvenience that they cause?


Fix for recurring Microsoft Word 2016 document opening/saving bug

Posted by Pierre Igot in: Microsoft
December 22nd, 2016 • 4:02 pm

Ever since I upgraded to Microsoft Office 2016 on my Mac, I have been experiencing this recurring bug.

Word 2016 works OK for many days. (I say “OK” and not “great” because Word never works great, even at the best of times.)

Then all of a sudden, for no apparent reason, it becomes unable to open or save documents. The application starts displaying all kinds of error messages, like the following:

word2016-cantopen

Alternatively, if you are in the process of working on an existing document that you’ve been saving regularly using Command-S until now, all of a sudden Word starts complaining that it can no longer save the document (which might leave you stranded if you made lots of changes since the last time you saved the document).

Sometimes, weird problems like this one can be remedied by quitting and relaunching Word. Not in this case, though. If you try to quit and relaunch Word, you then get the following message:

word2016-cantopentemplate

In other words, at this stage it looks like Microsoft Word 2016 has become unable to open any file, including its own Normal template file. To make matters even more scary, if you click on “OK” to continue and then try to quit Word the normal way, when trying to quit, Word displays the following error message:

word2016-cantopensavetemplate

Don’t click on “Yes”! (It won’t actually do anything, since Word is currently unable to open or save anything, but it’s still not a good idea.) If you click on “No”, Word still fails to quit. So at that stage the only option is to force-quit Word.

You can try this whole cycle multiple times just for fun if you want to, but whatever option you choose in those misleading error dialog boxes, nothing will work, and Word won’t go back to normal.

The whole problem happened to me again this afternoon. Normally, when it happens, I have found that the only solution was to reboot the entire computer. Today, however, I had a little time, so I decided to do more sleuthing.

My first reflex was to go to the Console and see if I could notice any error messages that might help better circumscribe the problem.

I first noticed that, even time I tried to relaunch Word, the system log would include something like this:

16-12-22 14:50:16,765 Microsoft Word[50336]: ApplePersistenceIgnoreState: Existing state will not be touched. New state will be written to /var/folders/sp/fgbr368s4_q5_rbzb_4k5w740000gn/T/com.microsoft.Word/com.microsoft.Word.savedState

So I figured that the problem might have to do with some kind of conflict between Word and OS X’s built-in application resuming feature, which normally causes an application to reopen the windows that were left open the last time the application was quit.

Of course, the irony here would be that OS X’s Resume (or “Persistence”) feature has never been supported by Microsoft. Word 2016 has never been able to reopen the windows that were left open the last time it was quit, even when the general setting in OS X’s System Preferences (under “General”) is as follows:

osx-resumesetting

Based on this message in the Console, Microsoft’s lack of support for this OS X feature clearly does not prevent OS X itself from trying to apply it to Word 2016, even if it does not do anything.

So my first troubleshooting step was to try and see if I should clear whatever “saved state” might linger in my system for Word 2016 and relaunch the application. I went to:

~/Library/Saved Application State/com.microsoft.Word.savedState/

and cleared everything inside that folder and tried to relaunch Word, but that didn’t work.

I then went to the exact location mentioned in that message in Console:

/var/folders/sp/fgbr368s4_q5_rbzb_4k5w740000gn/T/com.microsoft.Word/com.microsoft.Word.savedState/

and I cleared everything that was there and tried to relaunch Word, but that didn’t work either.

Finally, I tried to launch Word 2016 while holding the Shift key down. This did make a difference, in that Word 2016 no longer displayed the message saying that it was unable to open the Normal template. But that’s easily explained if you consider that starting Word up while holding the Shift key down effectively launches Word in a so-called “safe” state, for troubleshooting purposes. In that “safe” state, Word no longer tries to open the Normal template, in case the template itself is the source of whatever trouble you are trying to address. (It is not the source of the trouble in this case, which can be demonstrated by the fact that the whole situation is resolved by simply rebooting the entire system. After that, Word 2016 works fine again, with the same Normal template.)

After launching Word 2016 successfully in a “safe” state, however, I found that it was still unable to open files. So the problem was not fixed.

Basically, this whole process suggested that the problem had nothing to do with OS X’s Resume/Persistence feature and couldn’t be fixed by fiddling with the folders involved in that feature, and that the error message in the Console didn’t really mean anything significant. (I was able to confirm this later on by finding that the error message still appeared upon launching Word even when Word was working normally again.)

I then paid closer attention to the other references to Microsoft Word in the system log in the Console, and found multiple occurrences of another suspicious-looking message:

16-12-22 15:43:17,268 Microsoft Word[61532]: NSAllowAppKitWeakReferences=YES

I had no idea what it meant, but I decided to try and find out if anyone online had ever referred to this “NSAllowAppKitWeakReferences” variable in the context of troubleshooting Word 2016.

I first found a thread on Apple Discussions, but that thread discussed a different problem (Microsoft Error Reporting quitting unexpectedly) and nothing is the thread was helpful in my situation.

Then I found this thread on the Microsoft Community forum for Office 2016 for Mac. It contained lots of the usual bullshit about things being Apple’s fault, Microsoft allegedly having fixed (or not fixed) the problem, Microsoft allegedly having noted that multiple users were reporting the problem and being supposedly working on a solution, Microsoft not being involved in these “user-to-user” discussion forums even though a user named “Erik J” had “[MSFT]” appearing next to his name, etc.

And then I saw the very last message in the thread, from that Erik J who might or might not work for Microsoft, who mentioned the following troubleshooting steps:

  1. Close Word.
  2. Click the magnifying glass icon in the top right corner of your screen, type Terminal, and press Return.
  3. In the Terminal window that opens, type the following and press Return: defaults write com.microsoft.Word ForceTempFilesInAppContainer 1
  4. To ensure that this setting worked,
    • Open a Finder window, and press Command+Shift+G.
    • Type the following path: ~/Library/Containers/com.microsoft.Word/Data/Library/Application Support/Microsoft/Temp
    • Press Return. Upon launching Word, this folder will be created if it doesn’t exist, and Word will create a temp file inside it. Quitting Word will then cause this temp file to removed, which tells us that the new setting has taken effect.

I don’t really know what this does, but I tried it this afternoon on my machine after Word 2016 started acting up and refusing to open or save any file, and… it fixed the problem!

After typing the above-mentioned defaults command in Terminal and relaunching Word, I found that it was back to normal, able to open the Normal template without trouble and able to open and save files as expected.

Given that Erik J does not give any further information and that it is the last post in the thread (dated August 19, 2016), I don’t know if it’s a permanent fix or a temporary one, or if it’s even meant to be a fix. All I know is that it seems to work on my machine for my copy of Word 2016, and that this seems to indicate that, if the problem occurs again, I might be able to fix it without having to reboot my entire machine, which is of course a significant relief.

One ironic aspect of the whole process is that, after all this, I discovered that, even when Word 2016 was back to working normally again, it still caused the recurrence of the NSAllowAppKitWeakReferences=YES error in the system log. So maybe that error in itself doesn’t mean anything either, but actually just guided me to the right forum.

I will post a link to this blog post on the forum at the end of that discussion thread, and maybe Erik or someone else at Microsoft will notice it and maybe (yeah right) get back to me with a request for more information or a confirmation that this is a fix (and that it will be included in a future update). Whatever happens, I figure the most useful thing that I can do for fellow Word 2016 users is to write about the whole thing in a blog post, so there you are.


Video freezes on Mac Pro: Video card replacement required

Posted by Pierre Igot in: Macintosh
December 18th, 2016 • 5:19 pm

I don’t know it is bad luck or something else, but I have certainly experienced my share of hardware flaws and failures with Apple products over the years. I am not talking about more or less predictable failures here, like conventional hard drives failing after a few years, or power adapter cords becoming frayed (although I’ve certainly had my share of those as well, of course). No, I am talking about hardware issues that were due either to clear design flaws or to defective parts.

The first significant one that I can remember is the problem with wifi reception on the original Titanium PowerBook G4 back in 2001. (While this was never officially acknowledged, the titanium shell of the laptop effectively acted as some kind of Faraday cage and had a significantly detrimental effect on wifi reception on the laptop, which, as you can imagine, was a massive disappointment.)

Then, in 2003, I had a Power Mac G4 MDD that was affected by a major noise problem with its power supply unit. The problem was so bad that an online petition was started, and eventually Apple had to offer a replacement program for those affected.

That G4 MDD was replaced by a Power Mac G5 in 2005, and that machine worked well for a while, but then after less than a year I started experiencing all kinds of random freezes. Apple was never able to fix the problem, and eventually offered to replace it with a Mac Pro.

Meanwhile, in 2006, I also purchased a new black MacBook for my wife and, for the first time ever, I had to send the machine back and ask for a refund, because it had unacceptable noise issues that were treated by Apple as “normal” or “working as expected”.

In 2007, I ended up purchasing a 17″ MacBook Pro instead, and it worked OK for a while, and then the machine started developing a problem with a bulging battery.

Still in 2007, I started having kernel panics with my 2006 Mac Pro that required a logic board swap.

In 2009, I started experiencing kernel panics on my 2009 Mac Pro with dual video cards (to drive two monitors). These kernel panics were clearly due to a software issue, but since the software issue only affected a small subset of users (those with two video cards), Apple never did acknowledge it, and I ended up opting to purchase an expensive replacement video card of a different kind just to work around the problem.

That 2009 Mac Pro worked fine after that (and is, in fact, still going strong today in late 2016, with a third-party SSD as its system volume and with that replacement video card), but I eventually replaced it with a 2014 Mac Pro as my main work machine.

While I had major responsiveness issues with the 2014 Mac Pro when running the early versions of Yosemite (a very annoying software problem), I didn’t have any really significant hardware issues for the first couple of years. (This Mac has some occasional Thunderbolt-related glitches, and there is also a noticeable high-pitched whine at times when the machine is running a bit hot, but these are relatively minor.)

Then, a few weeks ago, I started experiencing something new. For no apparent reason, out of the blue, the Mac Pro started experiencing video freezes. During these freezes, everything on the screen would be completely frozen except for the mouse pointer, which would continue to move normally, although nothing would respond to mouse clicks. Sometimes, the mouse pointer would eventually turn into the Spinning Beach Ball of Death and nothing would happen for several minutes, so I would end up forcing a hard reboot, by holding the Mac Pro’s power button down for five seconds.

Most often, however, each freeze would last about 20 or 30 seconds, and then things would become responsive again, but only for a few seconds. OS X would try to execute some of the mouse actions that I had attempted while the screen was in the process of becoming frozen (which had obviously been stored in a buffer by the system), but then after a few seconds, everything would freeze again. The same cycle would repeat a few times, and then eventually things would go back to normal.

In addition, sometimes, at the same time the screen would freeze, I would see some video corruption, typically in some part of the main Mail Viewer window in Mail:

2014macpro-videocorruption

Like the freezes, the corruption would eventually clear up.

Luckily, this problem happened at a time when I hadn’t done any significant system updates in a while, except for a security update for El Capitan. So I was able to relatively easily rule out a software problem. (I still did suspect the security update for a while, although it was very unlikely, because I am still part of the AppleSeed program, and had installed early builds of it before the final, official release.)

From the very beginning, my 30 years of experience troubleshooting Mac computers for myself and for clients suggested to me that this was a hardware issue. I still went through all kinds of routine troubleshooting steps (zapping the PRAM, emptying caches, etc.), especially since, at the beginning, it looked like the problem would typically happen right when I would attempt to load a graphics-rich web page in Safari.

Of course, the problem was intermittent and impossible to reproduce reliably. It would just happen seemingly randomly. Sometimes I would go a couple of days without a single freeze. Sometimes I would get two or three freezes in a single day. It didn’t seem connected to what I was doing at the time either, except for the fact that it most often occurred while browsing the web with Safari (a very common activity, of course).

I tried to run the Mac Pro’s hardware test from the recovery partition, but the test failed to identify any issue (which didn’t prove anything, since the problem was very intermittent).

I ended up getting on the phone with AppleCare. (I had purchased the three-year warranty for the Mac Pro back in July 2014, so it was still covered.) The phone representative was relatively competent and made me execute a few typical troubleshooting steps, including some that I had already done and some that I had not. She said to give it a try and get back to them if it didn’t solve the problem.

It didn’t. I did some more research and found that the system log in Console would, around the times of the freezes, contain multiple occurrences of an error message looking like this:

Mac-Pro kernel[0]: stalling for detach from AMDTahitiGraphicsAccelerator

I tried to find references to this online and was not very successful. But I did find this thread on the MacRumors forums, which appeared to confirm that this kind of problem was either with the GPU driver software or with the GPU cards themselves. (I also tried some low-tech troubleshooting steps, like trying to get rid of whatever dust might have accumulated inside the Mac Pro, and also switching Thunderbolt cables around, in case the problem was some kind of weird glitch with the video signal over Thunderbolt. Nothing helped.)

After a few days, I was on the phone again, and this time I didn’t have too much trouble getting the person to understand that I knew what I was talking about and that I had effectively “done my homework”. I told her quite bluntly that a trip to the Apple Store would be a four-hour round trip for me and that I hoped she would try and minimize the need for such trips, by ensuring that they were as useful and constructive possible.

Still, she couldn’t guarantee that the Apple Store folks in Halifax (Nova Scotia) would readily accept my request to swap the graphics cards. All she could do was to set up an appointment so that I could bring my machine to someone at the “Genius Bar” and get them to have a look at things.

With winter and nasty, unpredictable weather approaching, I took the next available appointment, which a few days later, on a Tuesday, at 10:40 am.

I arrived on time for the appointment, but the “Genius Bar” was obviously very busy with all kinds of folks getting one-on-one lessons on fairly straightforward stuff (as far as I could tell from the conversations I overheard). I had to wait for about 10 minutes, and then finally got to talk to someone about my problem. To her credit, she quickly realized that the issue was beyond her level of expertise and she summoned someone in the back, who showed up after a few more minutes.

He seemed to know his stuff, and proceeded to run a number of more “advanced” tests. Fortunately for me, he quickly noticed something odd, which had nothing to do with what I was experiencing: Whenever he tried to run his special testing software, the whole screen would take on a purple tinge. He asked me if I had ever noticed such a purple tinge while running the Mac Pro. I told him that I had not. But he was definitely able to reproduce this (and not my freezes) reliably while running his tests and this, to him, was reason enough to initiate a repair process that would involve ordering two replacement video cards and a replacement logic board (since he said the graphics problem could be located at the site of the connection between the video cards and the logic board).

They would have to order the parts and then try them in the machine to see if it made any difference (to the purple tinge problem, obviously). He said the whole process should take three to five days.

Unfortunately, since I only had consumer-level warranty coverage, my only option was to either go home with my Mac Pro and continue to use it for a few days, then come back in when the parts were in and leave the Mac Pro with them for who knows how long. I didn’t exactly relish the prospect of multiple round trips to Halifax, especially because of weather issues and also because of the typical shopping rush before the holidays.

Then I had to make a quick decision. I thought that my wife was currently still using my 2009 Mac Pro, which would turn eight years old in a few months. As I said, that 2009 Mac Pro with an SSD (bought from Other World Computing several years ago, and actually replaced once under warranty since the original purchase) and with a replacement video card was still working fine, but there was no denying that it was getting old. Also, my wife never liked the fact that she had to deal with a big tower, which was, of course, too powerful for her needs and, most important, rather noisy (relatively speaking) for her home office environment. So I figured that now was as good a time as any to replace her own computer with a more recent Mac, and the obvious (albeit rather expensive) option was to get her a 27″ iMac 5K. Provided that they had one in stock, I would leave the Mac Pro at the Apple Store and use that new iMac 5K as a replacement for my work machine until the repair was complete.

Granted, the iMac 5K model itself had not been upgraded by Apple in a year, and I would be spending a significant amount of cash on a model that was not exactly brand new and didn’t have the very latest technology (USB-C/Thunderbolt 3, etc.). But my wife didn’t really need the very latest technology and, besides, my experience as an early adopter (of both hardware and software) in the past 15 years has been decidedly mixed. It would also give me a real chance to put the iMac 5K to the test in my own work environment, to see if it might ever be considered as a potential replacement for my 2014 Mac Pro (given that the Mac Pro has been completely neglected since 2013 and that there is sadly a real chance that it will eventually be discontinued altogether).

The Halifax Apple Store did have the 27″ iMac 5K in stock, but only with Fusion drives (an SSD-only iMac can only be had through the built-to-order process) and only with 8 GB of RAM. I asked if they could at least upgrade the iMac to 16 GB, and they were able to do that in under an hour. So we left my Mac Pro at the Apple Store and went home with a new 27″ iMac 5K with 16 GB of RAM and a 2 TB Fusion drive.

For the next 24 hours or so, I attempted to use the iMac as my work machine without restoring my full work environment from my backup, by only installing what I thought I really needed, since it was only for a few days. That was a mistake. My work environment is heavily customized, with tools such as Typinator, Keyboard Maestro, LaunchBar, and so on. In addition, some other tools that I use on a regular basis, like the Grand Robert French dictionary on CD-ROM, are very poorly supported by the developer, and I soon found that, while the CD-ROM still works fine in my customized work environment, I couldn’t get it to work in a brand new environment. (The software would crash every time I would attempt to “authenticate” it by inserting the original CD-ROM. There was no hope of getting any help from the developer on a timely basis.)

So I ended up restoring my work environment from the backup just the same, even though it was a backup of the startup volume for my Mac Pro, and not of a startup volume for an iMac. The process took a while, and there was a scary moment at the end, when I attempted to boot the iMac from the restored work environment and I just got a black screen. I managed to solve the problem by rebooting with the Shift key down, forcing the iMac to start in Safe Mode and, while that was quite slow, it did work, and after that I was able to boot normally.

I was pleasantly surprised to find that the iMac was perfectly able to drive both its internal 5K display and my external 4K monitor as a main display via Thunderbolt. This meant that it was relatively easy for me to restore a fully functional work environment with two displays, the iMac effectively replacing my second (and older) 30″ Cinema HD display. (Even my Keyboard Maestro macros for moving windows from screen to screen still worked, because the “virtual” resolution of the 5K display obviously matched the normal resolution of my older 30″ display.)

And I must admit that I was more than pleasantly surprised by the quality of the 5K display itself. Now that I have experienced it, I can say without any reservations that my dream work environment right now would be to have a new Mac Pro that is able to drive two large 5K displays side by side. Sadly, such a machine does not exist, and we might have to settle for an improved iMac 5K that is able to drive one or two external 5K displays — if such a machine ever gets made.

I have no idea whether Apple is slowly abandoning its desktop computer lineup or not. It certainly feels that way these days. Needless to say, it would be news of catastrophic proportions for us professional users. The irony would be that it would happen just as the hardware capabilities in the industry are finally getting where we want them to be (i.e. affordable computers able to drive multiple affordable 5K screens).

The other concern I had about the iMac on my desktop was about the noise. The Mac Pro is near silent (except for that occasional high-pitched whine) and all my conventional storage (with hard drives) is in another room, with a 30-ft optical Thunderbolt connection through the wall. The iMac on my desktop brought a conventional hard drive (the conventional part of the Fusion drive) back within earshot. The noise was definitely noticeable, but I must admit it wasn’t really significant.

Of course, I didn’t really try to do anything hardcore, like exploiting the computer’s multiple cores to encode a large video file, for instance. So I don’t really know how noisy the iMac 5K can get. But I do know now that, under regular circumstances, doing the stuff that I regularly do, the noise level remains low and is quite acceptable. It’s not perfectly silent, but then nothing really is. It is very quiet, and if I had to choose between this constant low-grade white noise and the occasional whine coming from my Mac Pro, I am honestly not sure which I would choose.

So, I did eventually manage to work with the iMac as my main work machine for a few days, and, on the fourth day (a Saturday), I got a call, around… 8 pm, from the Apple Store, letting me know that they had swapped one of the video cards, and that the problem, as far as they were concerned, appeared to be fixed (which, presumably, meant that the purple tinge was gone).

I knew that a friend of mine was going to Halifax the next day, and he had already offered to pick the machine up for me if it was ready. So I told the Apple Store representative on the phone about this, and he said that it wasn’t a problem, that I could just give him my friend’s name and he would add it to the file as a person authorized to pick up the computer on my behalf.

The next day (Sunday), I got another call from the Apple Store, telling me that they had someone there to pick up my computer, and… that person was not authorized. With all the high tech that they have for managing their support calls and repair jobs, they obviously still are vulnerable to human error, or whatever the reason for this mix-up was. In any case, I repeated the name again, and it was OK.

My Mac Pro was back at home later in the evening, and I set things up again the following day. Everything was in working order, and I just had a backlog of a few days of daily tasks (filing emails away, transferring newer files, etc.) to complete before I was fully back in business.

I didn’t get any video freezes on that first day, but I figured I’d better wait for a few days at least before drawing any conclusions. It’s now been seven days, and I still haven’t had a freeze. I have also checked that there is no longer any occurrence of the “stalling for detach from AMDTahitiGraphicsAccelerator” message in the system logs.

So currently the signs are good that whatever problem in the video card caused the freezes was also what caused the purple tinge, and that fixing one fixed the other. After all this, I guess I can count myself lucky that my intermittent freezing problem also happened to have a non-intermittent side effect, which conveniently made the issue obvious and undeniable to Apple’s technicians.

Overall, I must admit that my experience with the Apple Store in Halifax for this repair was good. If I hadn’t purchased the extended AppleCare warranty back in 2014, the repair would have cost me nearly $400. With the warranty, it cost me nothing except the time wasted trying to troubleshoot the problem and then dealing with the Apple Store and travelling there to get the problem looked after. (I am obviously not counting the purchase of the iMac 5K here, but I must admit things would have been significantly more complicated for me and my work if I hadn’t seized this opportunity to replace my wife’s aging computer.)

I just have one last observation to make about the Apple Store staff. Everyone on the floor in the store does everything on iOS devices (iPhones, iPads): look up information, process payments, record data, etc. I was quite impressed with the staff’s typing skills on their iOS devices, which were obviously much better than mine (with far fewer typing mistakes, as far I could tell).

However, I couldn’t help but notice that, when using iOS for things other than typing, like tapping on buttons, links, etc., even with all their obvious experience using these devices, they still very often have to tap on buttons twice or even three times (or even more times) to get the buttons to respond to their taps. This definitely matches my experience, and I was secretly pleased to observe that I am not the only one, and that even experienced users have this problem.

I realize that the mouse is not a perfect pointing device either and that there is a small percentage of missed clicks, missed UI targets, etc. But as far as I am concerned (and, based on my observation, as far as a lot of users are concerned, even experienced ones), the percentage of misses is nowhere near as high as it is with tapping on buttons on a touch screen. Based on this observation alone, I simply don’t see how anyone can argue that a touch screen is an appropriate interface for doing serious work. The rate of wasteful tactile interactions is just far too great. (I have little experience with trackpads, but I suspect that they would fall somewhere in between a mouse and a touch screen when it comes to the error rates.)

And let’s not even discuss the replacement of real keyboard keys with “virtual” keys on a miniature touch screen… How often does a press on a keyboard key actual fail, as opposed to a tap on a button on a touch screen?

I don’t really know what to think (the worst?) of recent developments when it comes to Apple’s desktop hardware. But I won’t deny that I am quite worried about the long-term future of my use of a Mac computer with two large screens for my work. Will it at least remain viable until I retire (which is not for another 15 years or so!)? I tried to share my concerns as a “business” user with the Apple Store staff in Halifax, who, of course, are quite interested in trying to sell me more “business” products and services, but unfortunately that didn’t get very far

However, at least now I know that, in the immediate future, if worse comes to worst, I do have the option of replacing my Mac Pro with an iMac 5K. It wouldn’t be ideal and I would still be disappointed, but I could make it work. So that, at least, is partly reassuring.


OS X 10.11 (El Capitan): New mouse-related bugs

Posted by Pierre Igot in: Macintosh
July 23rd, 2016 • 5:53 pm

One of the very few new features that I was (moderately) excited about when I upgraded to OS X 10.11 (El Capitan) was the “Shake mouse pointer to locate” option in System Preferences > Accessibility > Display:

elcapitan-cursorshake

After a few months with El Capitan, I can now report, with relatively conclusive evidence, that once again, Apple has proven incapable of introducing a new (and potentially useful) feature without introducing all kinds of new bugs at the same time. (Does this remind you of anyone?)

As soon as I upgraded my 2014 Mac Pro to El Capitan, I noticed the following problem. Occasionally, when waking my computer from sleep in the morning, I would be faced with a user environment with no mouse pointer anywhere on the screen. No amount of mouse shaking or fiddling with the keyboard (which seemed to work as expected) would bring it back. The only solution was to unplug the mouse (I still use a old, wired Apple “Magic Mouse”), wait a few seconds, and then plug it back in. Then the mouse pointer would finally magically reappear.

I upgraded to El Capitan back when the 10.11.3 update came out. We are now at 10.11.6 and nothing has changed. The problem still occurs from time to time. Of course, it can’t be reproduced reliably 100% of the time using a specific sequence of steps, so there’s basically no point in even trying to report the bug to Apple. Even if someone over there did care, they would never have the chance to devote time and resources to trying to reproduce the problem themselves and then doing something about it. It simply is not the way it works with Apple. Either you have a 100% reproducible scenario, preferably for a bug that affects lots of people, or you’re out of luck and trying to get the bug acknowledged and fixed is an exercice in futility.

Last week, several months after upgrading my own Mac, I took the time to upgrade my wife’s Mac, which is a 2006 Mac Pro tower — an old machine, I know, but one that has been upgraded with a faster video card and with a solid-state drive. That Mac Pro still works fine, and still supports the latest system, so we have little incentive to replace it with something more current. But… that also means that we are now more likely to encounter new bugs with more recent versions of the system that Apple doesn’t know anything about and doesn’t care about, because there are now too few users running the OS on such old machines, and because they probably don’t do any testing in-house whatsoever on such old machines themselves.

Guess what? As soon as I upgraded her Mac to El Capitan, I noticed another mouse-related problem.

On her Mac, the problem is the following. Every time the Mac Pro (the entire machine, not just the screen) goes to sleep, when I wake it up, things seem to be working OK, but then after a few seconds the mouse pointer freezes completely. The Mac continues to respond to keyboard input. (Application-switching with command-Tab still works, for example.) But the mouse pointer itself refuses to move, and some other UI events seem to be stuck. Then after twenty seconds or so, El Capitan snaps out of it, the mouse pointer starts responding to mouse movements again, and other UI events stuck in a queue (system beeps, notifications) are spurted out in rapid succession.

Things work fine after that, but whenever the Mac Pro is put to sleep, either manually via the Apple menu or because of its Energy Saver settings, the same problem occurs a few seconds after waking the machine from sleep.

This time, the problem is 100% reproducible. I have even been able to reproduce it in a separate user environment with no customizations. So I could report it to Apple. But do I really think it’s worth the effort? I am far from convinced. Even if someone did pay attention to such a bug report, the likelihood of the bug getting fixed would be very small. (I’ve searched through the forums, and I haven’t found too many reports of the exact same problem. People have other freezing problems, but they are not temporary, and they don’t occur a few seconds after waking the Mac.)

So, what’s the solution? I can either ask my wife to get used to the 20-second delay every time the machine wakes from sleep, or I can change the Energy Saver settings so that the machine itself never goes to sleep. I can still get the screen to go to sleep and the hard drives to go to sleep, but I can set the Mac itself to “Never”. Of course, the environmentalist in me is not too happy about this, because it means more energy wasted for no reason (and also possibly a shorter lifespan for the machine). But what else can I do, really? I don’t get any sense that Apple really cares about such issues anymore (if they ever did).

The environmentalist in me is doubly annoyed these days, because my own 2014 Mac Pro also has a power-related problem in El Capitan. It now refuses to go to sleep altogether, even when I ask it to. It does go to some kind of semi-sleep, as evidenced by the fact that it stops checking for new mail after a while (even though I do have enabled Power Nap, for whatever that is worth), but the Mac itself never switches to the low-power mode with the pulsating light and zero fan noise. It stays on, and so does my external Thunderbolt enclosure of conventional hard drives in the basement.

This particular problem was introduced in the 10.11.5 update and is still there in 10.11.6. Earlier versions of El Capitan had no problem letting the Mac Pro go to sleep (and the Thunderbolt unit would automatically go to sleep itself after a few moments). I have reported the bug to Apple, and they have responded, by asking me if I can still reproduce the problem… in Developer Preview 3 of macOS Sierra! Really helpful… I guess there’s no hope of that one getting fixed any time soon either, then.

And up goes the wasteful power consumption in the Igot household thanks to Apple bugs…

I don’t believe this last bug has anything to do with the mouse, but I thought I’d mention it as yet another example of new, annoying bugs creeping up all the time in Apple software these days.

That is not all, however, as far as the mouse is concerned in El Capitan on my Mac Pro. By far the worst problem is another one, which I will attempt to describe now.

From time to time, seemingly out of the blue, the movements of the mouse pointer on my screen start getting jumpy. That’s the best way I can describe it. I move the mouse in a straight line, either dragging something along or just to click on something, and in the middle of the straight line, all of a sudden, the mouse pointer jumps to a different location on the screen.

It is very discombobulating and of course, causes me to make all kinds of errors, by clicking on the wrong target or dropping something in the wrong place. And it seems to be random.

I have been able to capture an example of the behaviour with ScreenFlow, which you can view in the following video clips. (The clips are in my Dropbox folder. You can preview them on the web or download them to play them on your desktop.)

To begin with, here’s a video clip of me dragging a window up and down in the Finder with smooth, responsive movements:

ElCapitan-NOTJumpy.mp4

Now here’s a nearly identical video clip, but this time with some “jumpiness” in the mouse movements:


ElCapitan-Jumpy.mp4

Hopefully you’ll notice the difference. In this particular clip, the jumps in question occur when going back up with the mouse.

This is just one particular example. The jumps are totally random and are sometimes much more significant, which makes it all the harder to keep track of the mouse pointer’s position.

Unfortunately, once this starts happening on my Mac Pro, the only way I have found to make it stop is to reboot the entire system — which, as you can imagine, is a major pain in the neck.

I haven’t really found much help through online searches. All I know is that the problem started occurring as soon as I upgraded to El Capitan. It cannot be a coincidence!

At some point, I suspected Bluetooth, but I don’t use any Bluetooth devices. Indeed, the upgrade to El Capitan, the system unilaterally reactivated Bluetooth on my Mac Pro, so I had to manually turn it off again. And for a while I didn’t have the jumpiness problem. But then it started happening again, even with Bluetooth off. So that wasn’t the culprit.

I am still investigating (and rebooting when I need to get work done). Right now I am experimenting with turning the “Shake mouse pointer to locate” option off altogether — even though I actually find it somewhat useful. (I have two large screens. Add to that the fact that Apple still has not fixed the on-going problem with the mouse pointer not changing reliably depending on the context, as it’s supposed to, and you can easily end up with the wrong mouse pointer on a background that makes it very hard to see.)

Of course, I cannot say with 100% certainty that this new problem is directly linked to the introduction of the the “Shake mouse pointer to locate” option in El Capitan. But I cannot help but wonder… and despair, once more, of Apple ever getting its house in order and shipping an OS that actually fixes a whole whack of bugs without introducing a whole whack of new ones.

Fun times…


OS X 10.11 (El Capitan): Problem with vanishing keystrokes still not solved

Posted by Pierre Igot in: Macintosh
May 24th, 2016 • 1:35 pm

In March, I wrote about my experience with OS X 10.11 (El Capitan), in relatively glowing terms. After my nightmarish experience with OS X 10.10 (Yosemite), the relatively smooth upgrade to El Capitan and the significant improvements in responsiveness were like a breath of fresh air. I even found long-standing bugs that were significant for me and that had finally been fixed…

Now that a few months have elapsed, however, I am afraid I need to revise my judgement somewhat. El Capitan is still a vast improvement over Yosemite, but I can now report that there are significant issues with OS X that were introduced several years ago (the consensus being that many of them first appeared in Lion) and are still not fixed in El Capitan.

First and foremost, there is still a major issue with keystrokes disappearing into the ether. I first reported on such a problem in OS X’s Mail back when Lion came out, in 2012. Since then, in every single version of OS X that I have used, I have noticed the problem, not just in Mail, but in various other applications, both Apple applications and third-party applications. It is, of course, an intermittent problem that is, for the most part, impossible to reproduce reliably.

I recently wrote a post about a particular scenario in OS X’s Finder where I can reproduce the problem 100% reliably. But this is just a particular scenario. There are many other scenarios (composing messages in Mail, writing texts in Word 2016 or Nisus Writer Pro, etc.) where the problem with disappearing keystrokes occurs randomly and unpredictably. It’s like OS X suffers from responsiveness “hiccups”, and these hiccups are enough to cause it to lose track of what’s currently being typed by the user.

The feedback I got after writing that post was that it probably had something to do with OS X handling stuff “asynchronously”… I am afraid I don’t know enough about the innards of OS X to understand what’s going on here. All I know is that responsiveness problems in OS X are nothing new and that, in the past, as far as I can remember, OS X always seemed to have some kind of internal buffer that would prevent it from losing track of what was being typed. Yes, the flow of text being typed out would sometimes be interrupted by hiccups, but these hiccups wouldn’t actually cause OS X to lose keystrokes. They would be processed eventually, after the hiccups had passed.

Now, since Lion at least, OS X basically has hiccups that not only slow you down, but actually can cause some of your keystrokes to disappear altogether without bring processed. How is this acceptable?

As usual, problems in OS X manifest themselves in clusters, and there is no denying that the problem with lost keystrokes tends to be worse in applications that also have their own responsiveness issues, most notably in Word 2016. This effectively makes Word, an application that is already of very poor quality to begin with, even worse. The problem with vanishing keystrokes also tends to get worse overall when there are several background processes that are taking place, and actually slowing down the entire system for obvious (and understandable) reasons.

But let me be clear about this: The hiccups and the vanishing keystrokes can only happen in OS X even when there is no obvious reason for the system to become less responsive, even only temporarily. My system is complex enough that I cannot fully know, at any given time, exactly everything that is taking place, but I always keep an eye on CPU levels on my Mac, and the hiccups and the vanishing keystrokes can happen even when CPU levels are quite low and there is no apparent reason for my machine to experience a slow-down.

Worse still: The hiccups don’t just affect text entry. If a hiccup occurs while I am in the process of typing a sequence of keystrokes to execute a series of actions, i.e. a sequence of keyboard shortcuts in an application, it can actually affect the execution of the actions. And of course, unlike text entry, which is visible on the screen, missing keystrokes in sequences of keyboard shortcuts are not easy to spot. It can be rather maddening!

I find it rather sad that, in 2016, on a very powerful machine (a 2014 Mac Pro), I find myself having to monitor activity levels on a regular basis just in order to try and better understand why my system is not more responsive or in danger of losing some of my keystrokes, and adjust my own behaviour accordingly.

When exactly will Apple finally produce a computer that does not force fast users like me to wait or adjust their own speed? As I said, overall, El Capitan is indeed much more responsive than Yosemite, at least on my 2014 Mac Pro, with the same array of both Apple and third-party applications that I use on any given day. But it still isn’t as good as a system such as OS X Snow Leopard used to be, and one of the main reasons is that responsiveness issues (which have always been a fact of life in OS X) are now compounded by the very serious problem of randomly vanishing keystrokes.

I would probably be much more tolerant of hiccups if they didn’t cause keystrokes to disappear. I would be able to simply ignore the hiccups then. But as it is, I cannot help but notice them, because whenever they happen, there is a risk that I might lose keystrokes and have to retype, in part or in full, what I have just typed all over again.

Am I really the only one who’s bothered by this?

(Things can get even worse. From time to time, I experience a hiccup that is so bad that it causes the foreground application to somehow “seize up” completely when it comes to keyboard input. When that happens, even though the application itself remains responsive — I can still navigate its menus, dialog boxes, etc. — it refuses to process keyboard input, instead playing a system beep or doing nothing. When that happens, things don’t get back to normal by themselves. I have no choice but to quit and relaunch the application altogether. You won’t be surprised to hear that the application most often affected by this is Microsoft Word 2016, but I believe I remember experiencing the problem in other applications as well, even if it’s much rarer.)


OS X 10.11 (El Capitan): Finder fails to record first few keystrokes typed immediately after creating new folder

Posted by Pierre Igot in: Macintosh
March 26th, 2016 • 2:33 pm

Last week, I wrote about how pleased I was with my upgrade from Yosemite to El Capitan.

I am still pleased, but daily use of El Capitan is starting to reveal things that I hadn’t noticed in my first few days of playing with the new OS.

As I noted in my post, I am a fast user. Yosemite was very frustrating because of all kinds of small responsiveness issues that caused it to experience “hiccups” and to fail to register some of my keystrokes or mouse clicks if I went “too fast” for it.

El Capitan’s responsiveness is much better. However, it’s not perfect. And I have a very simple example of a very common situation where, like its immediate predecessor, it fails miserably.

In OS X’s Finder, when I want to create a new folder, I tend to use the keyboard shortcut, i.e. command-shift-N. This is supposed to create the new folder and then make its default name (“untitled folder”) editable and highlighted, so that I can start typing the new name over the default name right away.

The key problem here is what “right away” actually means. In my mind, it means what it means in everyday English, i.e. without any delay.

In OS X 10.11, however, it means something else. It means “very soon after pressing command-shift-N, provided you give OS X a little time to recover from the tremendous strain of having to create a new folder and make its name editable and highlighted”.

In practical terms, it means that, after I press command-shift-N on my machine (a 2014 Mac Pro with 32 GB of RAM, with a 1 TB SSD as the startup volume), I cannot start typing the name right away. I have to wait for a fraction of a second before I do so. If I don’t, then the first couple of letters I type fail to appear in the folder name that I typed.

Yes, you read this right: I, Pierre Igot, am a supernaturally fast typist, with whom a powerful machine such as the 2014 Mac Pro is not able to keep up.

I am joking, of course. There’s nothing supernatural about my typing, and I am quite sure the Mac Pro itself is not at fault. The problem is, as usual, with Apple’s software and, in this particular case, with El Capitan’s Finder.

I’ve actually created a very simple Keyboard Maestro macro to reproduce the problem:

ElCapitan-Finder-NewFolder1

This macro does exact the same thing I do when I create a folder and type its name (“My new folder name” in this example), except, of course, that the macro is executed by the Keyboard Maestro engine, which is much faster than I, lowly human, can ever hope to be.

And guess what? El Capitan’s Finder completely fails to record the entire typed text. When I execute this macro on my Mac Pro, all I get is a new folder titled “untitled folder”. There’s no sign of the typed text (“My new folder name”) that the macro has input as typed text.

After experimenting a bit, I have found that, in order for El Capitan’s Finder not to ignore the typed text, there needs to be a pause of at least 0.3 second between the first step (the creating of the folder) and the second step (the typing of the folder name):

ElCapitan-Finder-NewFolder2

With that pause, everything works fine. If I only put a pause of 0.2 second, I get a folder with the truncated name “older name”. And if I only put a pause of 0.1 second, I get… nothing.

Of course, my own human hands are totally incapable of typing this fast, and in reality, when I execute these steps manually in El Capitan’s Finder, the worst that happens to me is that I am missing the very first letter of the name that I typed. But it happens often enough that it is very frustrating, especially since I only notice it once I’ve typed the rest of the name, and so I have to backtrack and edit the name manually in order to restore the missing letter at the beginning of the folder name.

What is really unbelievable to me here is not so much that the Finder needs a fraction of a second after creating the folder. It is that there does not seem to be any kind of text input buffer that keeps my keystrokes in memory until the OS is ready to process them. The keystroke(s) that the Finder fails to register simply disappear into the ether, as if the characters had never been typed.

How on earth is this acceptable in a modern operating system? It’s one thing to have speed/responsiveness issues. It’s quite another to fail to include standard features such as a keyboard buffer in order to minimize the impact of such issues on the actual usability of the software.

Last week, I was tempted to call El Capitan the best OS X release in a long time. It still is way better than Yosemite, and probably also better than Mavericks, Mountain Lion, and Lion. But beyond that? I don’t remember having basic keyboard buffer issues in the old days of OS X. What the hell is going at Apple? In the rush to add new features and simplify computing for the masses, have they somehow completely forgotten the basics?

Two words, Apple: Keyboard. Buffer.

It is that simple.

PS: If you are as bothered by this as I am, please report (rdar://25375085).

UPDATE: BR closed by Apple as duplicate of BR #23108215 on 2016-04-12.


My experience with OS X 10.11 (El Capitan) on my 2014 Mac Pro

Posted by Pierre Igot in: Anti-Aliasing Hall of Shame, Macintosh, Mail, Microsoft, Technology
March 18th, 2016 • 6:16 pm

I haven’t written on this blog since last year and my very traumatic experience with upgrading my 2014 Mac Pro to OS X 10.10 (Yosemite).

It’s not that I don’t have anything more to say about OS X or other topics. It’s just that I find it less time-consuming and more immediate to post things on Twitter, in spite of the limitations of the medium.

I also have been feeling rather discouraged about the whole computing thing. It’s something that’s been building up for years. From my atrociously noisy Power Mac G4 MDD to my Faraday-cage PowerBook G4 that couldn’t get a wireless signal to the mooing MacBook disaster to my kernel-panicking Power Mac G5 to the seemingly never-ending list of long-standing bugs in Apple’s software and its general untrustworthiness as a purveyor of quality, reliable productivity software (hello, Pages 5), I can’t help but feel that the number of on-going problems with my computing setup outweighs whatever enthusiasm I might still have in me about technology in general and computing with Apple products in particular.

And so the reserves of energy and time that I am willing to devote to writing about these things have become somewhat depleted.

That being said, unlike some people, I cannot stay behind the times for very long. There’s just no way that I can continue to work and play with an operating system that is several years old and no longer updated.

After the trauma of my experience with Yosemite, you won’t be surprised to hear that I decided to take my own sweet time with the upgrade to El Capitan. With Yosemite, I had waited until the official release of OS X 10.10.1 before upgrading from Mountain Lion. And it’s only when OS X 10.10.3 came out, several months later, than my major responsiveness issues with the Mac Pro were finally resolved. So I decided to wait until at least OS X 10.11.3 before upgrading to El Capitan.

Don’t get me wrong. It’s not that I was happy with my Yosemite-based system. While the major responsiveness issues were solved by the OS X 10.10.3 update, there were still many less obvious, but still extremely irritating responsiveness issues that were never fixed in Yosemite, even with the release of the OS X 10.10.5 update (the latest available). And there were, of course, a number of other bugs that were quite annoying to live with, not to mention the very questionable choice of Helvetica Neue as the system font.

Apple touted El Capitan as primarily a “bug fix” release, even though, of course, its marketing material focused on yet more new features — most of which held little interest for me (except for the promise of “across-the-board acceleration” and a Mac that would “feel more fluid and responsive”).

However, after the initial release of El Capitan, I saw a number of reports online on yet more new bugs introduced by this latest OS. Of course, it’s always difficult to get a proper sense of how bad things really are when you read about other people’s experiences online. But several specific things that people reported about El Capitan made me quite worried, and conforted me in my choice to wait things out.

Finally, a week ago, I completed a big project that had kept me busy for a whole month, and I decided that my “reward” for completing it would be to take some time to install El Capitan. At that stage, OS X 10.11.3 was officially out and AppleSeed had already released several iterations of the 10.11.4 update, so I thought it was a reasonable time to give it a try.

Responsiveness and Fluidity

I took all the usual precautions, of course, and made sure that I could easily fall back on my Yosemite-based setup if things turned out to be disastrous again, as they were with OS X 10.10.1 for me back in December 2014. I also made sure that all my third-party software was perfectly up-to-date.

And then I took the plunge. The upgrading process itself was rather worrisome, because there were long periods of time where nothing (absolutely nothing) was happening on the screen and I was really worried about the upgrade process having become stuck. (The progress bars and status messages were, on the whole, pretty much useless on my machine.) Thankfully, I managed to be patient enough to ignore those worrying signs and wait it out, and finally, after something like 90 minutes, the upgrade sorted itself out and I was back up and running.

My initial impression was fairly positive. There were only a few relatively minor glitches. There were no new obvious “big bugs” that might have been deal-breakers or at the very least very painful problems to have to live with for a while. And the whole system did indeed feel significantly more fluid and responsive on my Mac Pro than Yosemite ever did.

Now, a week later, I can confirm that there are indeed significant improvements in that department. I am afraid that, in Yosemite, Apple really got too far ahead of itself in the revamp of OS X’s UI and built something that was far from optimized for a setup such as mine (i.e. a 2014 Mac Pro with two monitors, a conventional 30” Apple Cinema display and a more recent 4K Sharp display), especially when it came to graphics acceleration and overall visual fluidity. There were far too many unpleasant “hiccups” in the Yosemite UI to which I — as a user who tends to work very fast (fast typing, fast use of the mouse and keyboard shortcuts) and uses several tools to automate a large number of tasks — was far too sensitive. Those hiccups were impossible to get used to for me, and were one of the main reasons why, in the end, I was quite anxious to upgrade to El Capitan and see if it wasn’t “all in my head”, and if Apple had indeed taken the task of making the OS more responsive for users like me seriously enough.

El Capitan is decidedly more fluid than Yosemite. Most of the hiccups that would be so annoying to me in Yosemite are gone. There is still the occasional noticeable delay in this or that, but not often enough to be a significant annoyance for me. I no longer notice times when the OS takes a fraction of a second to draw a menu after I’ve clicked on its heading, for example. (The weird way that OS X draws the blue highlighting for the “Help” menu heading, which I noticed first in Yosemite, is still there in El Capitan, however, but I don’t use that menu — which is rarely all that helpful — often enough to really be bothered by it.) And I no longer notice hiccups when dialog sheets (like the “Save As…” dialog) drop out of window title bars.

In fact, El Capitan’s responsiveness improvements are such that they have made even using this horrible kludge of a piece of software called Microsoft Word 2016 (something that I cannot escape in my line of work) not as painful as it was in Yosemite. Word 2016 is still a horrible application and it still has all kinds of responsiveness issues of its own, but at least they are no longer compounded by OS X’s own responsiveness issues, which means that they are somewhat more tolerable (but still unacceptable, of course).

(Another obvious change in Word 2016 in El Capitan is that now, when I open a Word document from the Finder, Word first displays a big black rectangle, and then fills it with the document window. It’s rather unsightly, although, of course, quite harmless, and only lasts for a fraction of a second. Microsoft has never had any shame about such subpar finish in its software, so I don’t expect that this purely cosmestic issue will be fixed any time soon.)

Even more important, the problem with keystrokes occasionally disappearing in the ether (instead of being buffered, and eventually properly processed) because of responsiveness issues with OS X seems to be gone, or at least greatly reduced. (It’s hard to reproduce such an intermittent problem.)

I was also pleasantly surprised by how few Keyboard Maestro macros I had to update after upgrading to El Capitan. My experience with the upgrade to Yosemite had been that several macros needed to be adjusted because of responsiveness issues and other changes and glitches. In El Capitan, I only had to make one significant change, to my macro for sending messages in Mail (which automatically adjusts font size for rich text messages), because Apple changed something in in the GUI scripting architecture (via System Events) for that application.

Mail

In fact, OS X Mail was one of my main areas of concern for the upgrade to El Capitan, because I had read reports about new bugs in the application, and I rely on it too much to be able to live with too many bugs. Fortunately, while many of Mail’s existing, long-standing bugs still aren’t fixed in El Capitan, I have yet to encounter any significant new bugs that really affect me in my use of it. The only notable ones so far are the following.

Mail loses the focus on the file list when I use the “Attach Files” dialog box to add attachments to a message that I am composing and when I take the extra step of Quick-Looking the files to check them before attaching them (a reasonable enough habit, I believe!). But I’ve submitted a bug report about it, and Apple has already responded by closing it as a duplicate, so the early signs are encouraging.

Also, there is a significant problem with the toolbar in the window for composing messages. Unlike toolbars elsewhere in Mail, this toolbar no longer displays the text labels under the button icons, and the option to force the toolbar to display these text labels is gone from the “Customize Toolbar…” dialog sheet:

mail-send-toolbar1

mail-send-toolbar2

mail-send-toolbar3

I hope that this is a bug that was introduced accidentally, and that it’s going to be fixed soon, because it’s kind of annoying, even for someone like me who pretty much knows what each button icon means. I still rely of the text labels as well for faster targeting.

One other noticeable change (for me) is that the information provided in Mail’s “Activity” window is now much more limited, to the point of being almost useless. I’ve always relied on that window to get a better sense of what was going on behind the scenes in Mail, especially when my bandwidth was much more limited than it is today. When things work fine, I don’t really need it, of course. But I don’t know what’s going to happen the next time I start having some serious problem with a mail server. (Also, Mail no longer has an optional pane for a more summary display of its background activities at the bottom of the mailbox list. Instead, it just displays a small status line from time to time at the bottom, on top of the list, whether you want it or not.)

I still find it rather irritating to have to deal with Mail’s very clumsy and buggy mechanism for removing attachments from messages before archiving them. The problems introduced in Yosemite (especially when removing attachments from emails in discussion threads, a.k.a. “conversations”) are still there in El Capitan, and the even older problems haven’t been addressed either. There has been absolutely no progress on that front, even though removing attachments is an important tool in order to try and keep the total size of one’s email archive reasonably low. (Mine currently weighs over 2 GB. I hate to think what it would weigh if I had left all these attachments in the messages when archiving them.)

And Mail’s text rendering engine still gleefully treats non-breaking spaces as regular, breakable spaces, which is quite problematic in French typography (although there’s probably no one at Cupertino who really cares about this).

On a positive note, however, Mail no longer freezes for 20 seconds every time I try to access the “Accounts” pane in its “Preferences” dialog, which I am quite relieved about. Also, the long-standing issue that I had with Mail failing to flag my sent messages as having attachments (when they did contain attachments) and therefore preventing me from removing said attachments (because it didn’t see them) appears to be gone.

And Apple has even fixed the long-standing bug with selection highlighting in Mail when it was hidden by the “Hide Others” menu command while in the background. It still took them five bloody years to fix it, but at least it’s finally gone (as is the similar selection-highlighting problem in various other applications that I use, especially stand-alone, site-specific Safari-based web browsers created using Fluid).

Other OS X Applications

In other OS X applications, some long-standing problems remain, such as Preview’s inability to keep up with very fast window resizing (especially problematic if you use Keyboard Maestro macros to resize your windows).

The Finder still suffers from that annoying glitch where OS X fails to invert the text of the folder name when dragging a file onto a folder in a background window:

ScreenFlowSnapz002

(And I’ve also already encountered the glitch where the Finder fails to highlight the target of a drag-and-drop altogether, so that’s not completely fixed either.)

Apple has changed something in the system that causes the font smoothing in Numbers 3.6 to be somewhat better for some fonts (Helvetican Neue, for one), but it still does not use subpixel antialiasing like it does in another applications, at least on my computer:

NumbersSnapz001

Spotlight still fails to automatically index new “.docx” documents that I add to my system (whether by creating them from scratch or by saving them as attachments from email messages), which is rather annoying. I was hoping that this was going to be fixed in El Capitan, because the last response I got from Bug Reporter about this problem last year (when I was still running Yosemite) was a request to confirm whether the problem was fixed in… the latest build of El Capitan at the time, which of course was very unhelpful, because there was no way I was going to upgrade to El Capitan back then (which was still then in beta form), just in order to check if this particular bug was fixed.

I gave up on Spotlight as a serious search tool long ago, and tend to rely on the (rather expensive, but effective and reliable) FoxTrot Professional Search application by CTM Development for my searches, but the chronic failure to automatically index new “.docx” documents is still an annoyance.

I actually ended up following the advice in this thread at Apple Discussions (I am obviously not the only one with this problem) to create a Terminal command for forcing Spotlight to index new “.docx” documents, and using CronniX to add this command to my system as a task to be automatically executed every day.

But really, should I, as a Mac user, really have to go to such lengths to work around your problems, Apple?

Another thing that I was looking forward to in El Capitan is the ability to shake the mouse to magnify the cursor, when I lose track of it on my large screens… It works, but it doesn’t work as well as I had hoped. I occasionally get cursor magnification during regular mouse movements, when I don’t want or need it, and when I do want the magnification, I find that the amount of shaking required is a bit excessive. I hope the feature will be fine-tuned in the future. (It might work better with a wireless mouse, but I still use the wired Apple Mouse.)

Sadly, the problem with OS X failing to change the mouse cursor reliably based on the context, which was introduced in Yosemite, is also still there in El Capitan. This problem is brilliantly illustrated by Eli Schiff in a YouTube video here. (The video was made with Yosemite, but as far as I can tell the problem still exists in El Capitan.)

The new San Francisco system font is indeed much nicer than Helvetica Neue, and we should never have had to live with Helvetica Neue for a year (and I really feel for those who are obliged, for whatever reason, to stick with Yosemite). Like other commentators, however, I do find that the lack of distinction between the opening and the closing single quotes and double quotes is quite problematic.

The problems with OS X’s architecture for “resuming” applications (i.e. reopening the windows that were left open when the application was last quit) are still there in El Capitan. TextEdit still seems unable to resume properly after a system restart, and I’ve already had an incident where I accidentally interrupted a system restart triggered by a software update and, after manually triggering the restart myself, I ended up with a complete failure of the resume feature in the Finder, which meant that I had to reopen all my Finder windows and tabs manually.

Conclusion

That’s about all that I can say for now, after one week. Overall, I am quite pleased with having upgraded to El Capitan. The singlemost important improvement for me is the responsiveness and fluidity of the UI. It’s not perfect, but it is much, much better than it was with Yosemite on the same hardware. The upgrade is worth it just for that.

I am also quite pleased to find that Apple has indeed fixed some long-standing bugs that had been a daily annoyance for me for years. I can actually now eliminate some of my Keyboard Maestro macros, because they are no longer needed to work around those bugs.

Unfortunately, there are still many issues that remain unaddressed, in Mail and in other OS X applications, which make the upgrade less impressive as a “bug-fixing” release than I had hoped it would be. Given that Apple’s recent history indicates that no significant additional bug fixes will come in subsequent incremental updates for El Capitan (other than for bugs introduced in El Capitan itself), and that the next version of OS X might not have as much a focus on bug fixing as this one, I fear that I might have to continue to live with those remaining annoyances for many more years.

El Capitan does not significantly alter my current feelings about computing in general. Sure, it’s not as painful to use as Yosemite was, but it’s still far from being as polished as I’d like it to be, as some of the problems highlighted above demonstrate.

Will the next version of OS X be another step forward or a new step back? Time will tell… But if you haven’t upgraded to El Capitan yet and needed to know more about what it brings, I hope that the above addresses at least some of your questions and concerns.


OS X 10.10.3: Fixes major responsiveness issues on 2014 Mac Pro

Posted by Pierre Igot in: Macintosh
April 15th, 2015 • 6:03 pm

This has been a very traumatic few months for me as a Mac user. As indicated in several previous blog posts, I purchased a new Mac Pro along with a new 4K display back in July 2014. The machine shipped with Mavericks and worked pretty well until I upgraded to Yosemite.

I waited until OS X 10.10.1 came out before upgrading, but this was still not cautious enough. I soon realized that, on my Mac Pro, Yosemite suffered from significant responsiveness issues that were not tied to any specific piece of software or to a particularly high level of CPU activity.

When I say “significant”, I really mean it. These were not just slight delays that one can accept to live with in a 1.0 or 1.0.1 version of a new software product until the developers get around to further optimizing the software. These were massive, outrageous delays in seemingly simple tasks, such as switching apps, clicking on menu headings to pull them down, resizing windows, etc.

As an experienced troubleshooter myself, I went through all kinds of steps to try and isolate the problem and hopefully find a solution. Nothing worked. There was not a single application that I could identify as the culprit. It soon became quite clear to me, with my extensive experience as a troubleshooter, that this was a problem with the OS itself.

In desperation, I even went so far as to do a brand new “clean install” of my system and applications — something that I hadn’t done in years. When you use a fairly advanced setup such as mine, with a number of third-party applications and lots of customizations (none of which would qualify as “hacks”, although Apple probably does not do much in-house testing to ensure that utilities such as Default Folder X, Typinator and Keyboard Maestro always work well with their latest system software), doing a clean install is a very tedious, time-consuming process, because you not only have to migrate all your files back and reinstall all your applications, but also restore all the customizations one by one, without relying on existing application support files that might have somehow become “corrupted”.

But I did it, and even before I had completed the process of restoring all my customizations, I knew that the clean install had not solved the issue.

The problem with responsiveness issues is obviously that they are somewhat subjective. I am a fast typist and, generally speaking, a fast GUI user. I even use Keyboard Maestro to automate a number of repetitive interactions with the GUI using macros; those macros effectively mimic the behaviour of an extremely fast GUI user. So I am particularly sensitive to responsiveness issues. And let me tell you, on my machine at least, the situation was a nightmare.

I had all kinds of Keyboard Maestro macros that no longer worked properly because the system was not responsive enough, so I had to artificially add all kinds of built-in delays to the macros to try and make them work with the OS. (And even that only went so far, because the system’s own delays were not consistent and predictable.)

But even in my own “manual” interactions with the Yosemite GUI, I was hampered by all kinds of undesirable delays that would cause me to miss targets, accidentally trigger things I didn’t mean to trigger, etc. It was maddening!

Given that my Mac Pro was a recent purchase and that this was clearly an unacceptable level of responsiveness on a supposedly fast machine, I felt that I had no choice but to go through Apple Care. I won’t get into all the details, because telling them in full would be far too long, but suffice to say that I “burned through” four different “senior representatives”, who all agreed that what I was experiencing was unacceptable, but were at a loss to explain it, let alone reproduce it themselves or get their engineering counterparts to acknowledge the seriousness of the situation and do something about it.

Things were so bad that there was talk, at some point, of downgrading to OS X 10.9, or even of trying to get my Mac Pro’s video cards replaced, in case that might somehow miraculously fix the problem — as if it wasn’t obvious enough that this was a software problem and not a hardware problem.

I spent a lot of time executing troubleshooting tasks for the senior representatives, or for the engineering people that they were talking to about my issues, all the while knowing very well that none of them would do anything to fix the problem.

What was particularly maddening, of course, was that responsiveness issues are hard to capture, and also tend to be intermittent in nature, which means that you often cannot reproduce them on demand. However, I did end up purchasing a piece of software (ScreenFlow) that enabled me to create video captures of some of the most egregious problems, and I was also able to narrow down, to some extent, the circumstances under which the responsiveness problems would occur.

Unfortunately, I also wasted a lot of time on what can only be described, in retrospect, as wild goose chases, because I was seeing various behaviours that seemed to be connected but ultimately proved to be unrelated (or loosely related at best) to my responsiveness issues. For instance, I couldn’t help but notice that Yosemite was eating up the 32 GB of physical RAM on my Mac Pro like candy, and that responsiveness issues seemed to be getting worse the closer I got to 32 GB of memory used. The last senior representative that I talked to seemed to be unable to find out whether this kind of RAM usage was normal in Yosemite or not, and couldn’t explain, for instance, how a single web page left open in Safari could end up using 1.5+ GB of memory all by itself, on top of the memory used by the Safari process itself.

While Apple Care’s representatives proved to be unable to help me in any significant way, I was fortunate enough to get a variety of suggestions and tips from readers of this blog and my Twitter feed. This is how, for example, I found out that having turned the “Reduce transparency” option on in the Accessibility settings in System Preferences was actually causing a significant menu-flashing bug on top of my general responsiveness issues, and I ended up using the “Increase contrast” option in the same Accessibility settings for a while, because it helped reduce the severity of this bug and also, somehow, seemed to slightly alleviate the general responsiveness issues in Yosemite on my machine.

I, of course, had mentioned to the Apple Care representatives that I was also a member of AppleSeed and, as such, had access to early builds of the next system upgrade, OS X 10.10.3. (OS X 10.10.2 had come out early on with no improvements.) I asked whether it was a good idea, in my circumstances, to give those early builds of 10.10.3 a try. The representative that I was in touch with at the time was obviously at a loss himself and figured that I didn’t have much to lose. So I did proceed with the installation.

The very first 10.10.3 build available through AppleSeed included the new Photos application, but did nothing to alleviate my problems, so I was not particularly hopeful that things would get better in subsequent builds (worried as I was that the focus would be mostly on this new Photos application and not on fixing bugs).

But then two things happened. I got a tip from a reader about this post on Reddit, which suggested a link between responsiveness issues in Yosemite and a specific kernel extension (AppleGraphicsPowerManagement.kext). I was intrigued, because I had always suspected (with my limited knowledge of the engineering underpinnings of OS X) that, in the effort to minimize battery consumption for laptop computers, Apple had gone too far in implementing mechanisms that put all kinds of things to sleep automatically when idle. Certainly, several of my responsiveness issues looked like problems where, when I was clicking on something, the computer system appeared to be somehow “waking up”, and only responded to my clicks after a few seconds. And of course, my responsiveness issues had everything to do with the GUI, i.e. with “graphics”, and I had been suspecting problems with graphics drivers all along.

And then within a few hours of that discovery, AppleSeed released a new build of the OS X 10.10.3 update, which I installed, and which included a number of updated versions of kernel extensions, including this particular one (and several other graphics drivers)!

I soon realized that this update was a major step forward, that it clearly addressed several of the issues that I had been experiencing. For instance, the menu-flashing bug was completely gone. There would still be the occasional slight delay between a click on a menu heading or button and the appearance of the actual menu, but there was no longer a problem with the menu flashing and disappearing before I had a chance to make a selection.

Most important, however, was the fact that the general problem with the massive delays that I had been experiencing on my Mac Pro was gone.

I am not sure I can convey how relieved I felt at this point. I immediately communicated with the Apple Care senior representative that I was in touch with. He, of course, had no inkling that this particular build might feature any bug fix that would address the problem that we had been discussing for over two months. He was just as pleasantly surprised to hear about this.

While I was expressing my huge relief to him, I also made it quite clear that I didn’t feel that this outcome made everything alright and that everything that had come before then was forgotten (let alone forgiven). For one thing, I had no guarantee (nor did he) that whatever this particular build fixed would not be broken again in subsequent builds. Since no one on the engineering side had communicated with him about the fact that this build might contain a fix, we had no idea whether the fix was even intentional — and not simply an accidental by-product of some other improvement or optimization.

In fact, I was so worried about this that I refrained from installing any subsequent builds of the 10.10.3 update until the official update came out last week. I wanted to make sure that I had a system that worked, and given that I was otherwise very busy, the easiest solution was to do nothing. Then when the update finally came out, I made sure I had an up-to-date backup of my existing system that was updated right before I installed the update, and I made sure I immediately tested the final version of the update to confirm that the responsiveness problems had not come back. (They had not.)

I also made it quite clear to the senior representative that I felt that there had been massive communication issues between the engineering side and the customer service side at Apple, and that this had caused me and him to waste lots of time on wild goose chases and other useless (in retrospect) troubleshooting steps. While he himself was at least getting paid for dealing with me, I, on the other hand, had no way to get any compensation for the countless hours I had spent on troubleshooting this problem. If at least Apple’s engineers had said, early on, “Yes, we know that there is a major responsiveness issue for some users, and we’re working on a fix, which should be available around such and such a date”, then we would not have wasted so much time!

There is a part of me that still hasn’t forgiven Apple for this, and maybe never will. The whole experience has been very traumatizing, and I am really angry at Apple for having caused me to waste so much of my time on trying to fix a problem that I could do nothing about. As I said to the senior representative, “the impression of callousness on the part of Apple’s engineers will take a very long time to dissipate”. I realize that this particular problem probably only affected a small minority of users, but when someone goes through four different senior representatives, provides reliable scenarios for reproducing specific issues, provides graphic evidence in the form of screen captures (the engineers even asked me to record a video of my monitor with my iPhone at some point!), and clearly demonstrates that he is no fool and knows what he is talking about, the issue has to be treated more seriously and professionally, with a more acceptable level of… responsiveness in the communication itself.

I cannot help but feel, at times, that Apple’s engineers think that they are above all this and incapable of making such egregious mistakes that cause significant hardship for some of their users. While I agree that customer complaints might not always be entirely legitimate and that some users are fussy and moan about things that, in the big picture, are really rather small potatoes, this was not one of those cases. The issues I had were very real, and the very fact that they were suddenly fixed by a single system update proves that they were.

Because of the lack of communication, I still don’t know if Apple’s engineers even realize that they have fixed those significant responsiveness issues for me (and hopefully others). Part of me is still afraid that I might encounter a regression in a future system update. But at least now I had an official version of Yosemite that I can fall back on if things go awry again.

In closing, I should also stress that this is not the end of my complaints about responsiveness in Yosemite. There are still multiple aspects of the GUI in Yosemite that don’t feel right to me. The responsiveness issues are no longer egregious to the point that I find them unbearable, but there are still problems, including several bugs that I identified in the process of trying to further circumscribe the issues that were specific to my Mac Pro and mistakenly took as manifestations of these issues, when in fact they were just made more visible and noticeable by the massive delays that my responsiveness issues were causing.

I still often notice, for example, a slight delay between the time I click on a menu heading in the toolbar and the time the blue highlighting behind the heading comes on and the menu below gets drawn. It’s a small fraction of a second, but it’s noticeable. It doesn’t affect my ability to use the menu, but it makes the whole process feel slower.

The particular problem that I identified in this blog post about the blue highlighting for the “Help” menu is still there. It’s cosmetic, but it looks sloppy.

The bug in OS X’s Preview that causes it to fail to keep up with fast window resizing (especially when using a Keyboard Maestro macro) is still there.

And there are, more generally, multiple situations where you can almost witness the process of Yosemite “building” the user interface in real time. For example, yesterday I posted on Twitter a link to this video recording that I created and that shows what happens when I play back, in slow motion, a video recording of System Preferences drawing the “General” preference pane in real time in Yosemite. This video clip, which is effectively the recording of the playback of a recording, shows that there is a delay of 0.07 second between the time the contents of the “General” preference pane appear and the time some of the blue highlighting that should be part of the UI appears (the blue highlights on the radio buttons, the blue highlights on the arrows for the pop-up menus, etc.).

Why is there such a delay? It is not caused by some background activity that would happen to be slowing my system down at the time. The 0.07-second delay happens each and every time I access the “General” preference pane. It’s a small delay, but it’s noticeable. And it certainly contributes to the feeling that Yosemite is slow, even if it’s not really that slow.

I don’t know what Apple did to the GUI in Yosemite. They clearly rebuilt a lot of things (the font change to Helvetica playing an important part in this process), and in the process they created a lack of smoothness, fluidity, and zippiness that is really quite disturbing. It makes you feel like you using a system that is somehow more “fragile”, less reliable, less solid and could fall apart at any point — something that Apple’s engineers have decided is “good enough” for most people, but that more demanding, more attentive users will find rather disappointing and unworthy of Apple’s legendary attention to detail (which is becoming more and more of a legend and less and less of a reality with every passing year, I am afraid).

I cannot finish this post without thanking all the Betalogue readers and Twitter followers who have written to me to share their own stories and offer suggestions, tips, or simply commiseration. Unlike Apple’s customer service, they were actually able, collectively, to make a difference for me, and I can only hope that, with this blog post, I in turn might able to make a difference for other users.


OS X Activity Monitor: Yet more evidence of Apple’s sloppy work in Yosemite

Posted by Pierre Igot in: Macintosh, Mail
March 13th, 2015 • 5:00 pm

Here’s another example of Apple’s sloppy work in OS X 10.10 (Yosemite) that is very easy to reproduce (at least on my Mac Pro):

  1. Open the Activity Monitor application (in the Application/Utilities folder).
  2. Option-click on your desktop to hide the foreground application, i.e. Activity Monitor, and switch back to the Finder.
  3. Click on the Activity Monitor icon in the Dock or press command-Tab to switch back to Activity Monitor.

Notice anything? The toolbar now looks like this:

yosemite-activity-monitor1

Normally, when Activity Monitor is in the foreground, the toolbar should look like this:

yosemite-activity-monitor2

In other words, somehow Yosemite “forgets” to switch the background chrome of the window toolbar back to its foreground version. Yet the window is indeed in the foreground, as indicated by the three coloured buttons on the left, which are coloured (and not greyed-out), i.e. in the state they are supposed to be in when the window is in the foreground.

And even the other toolbar buttons that look like they are in the background and disabled are actually active and you can click on them. (This actually forces them to switch back to their normal foreground state visually, one by one.)

So the issue is purely cosmetic. But it’s still very sloppy on Apple’s part.

I haven’t been able to reproduce this problem with any other application. I also cannot reproduce the problem when step #2 in the process above involves choosing “Hide Activity Monitor” in the “Activity Monitor” application menu. But I can also reproduce it when I switch to another application and use “Hide Others” to hide all other applications, including Activity Monitor.

Fixing the problem involves switching to another application and then back to Activity Monitor. This forces a “refresh” of the window display and then Activity Monitor uses the appropriate chrome for the foreground window.

This is not the first bug I have encountered in OS X with the system’s commands for hiding applications. Long-time Betalogue readers might remember that I reported, back in 2012, on a selection-highlighting bug that was first introduced in Lion and has been with us ever since. Because of this bug, after I use “Hide Others” to hide all applications other than the foreground application, if the applications thus hidden include Mail, when I bring Mail back to the foreground, it continues to use the background (grey) colour to highlight messages in the message list, and it also fails to make the I-beam cursor visible in messages that are in the process of being edited.

The same bug also affects, for some reason, stand-alone Safari-based web browsers created with the Fluid application. I have several of those and they all suffer from the same problem as Mail. If they are hidden by “Hide Others” and then brought back to the foreground, the selection highlighting is all wrong.

Here again, fixing the problem involves switching to another application and then back to Mail or the affected stand-alone browser app. In fact, it’s such an annoyance that I have created Keyboard Maestro macros to execute this fix each time I switch to those applications, because I use “Hide Others” all the time and so my stand-alone browser apps often end up in this undesirable state.

This selection-highlighting bug has now been with us for three years. And now, thanks to Apple’s sloppy work in Yosemite, we can add a new one to the list, which only affects Activity Monitor, but is definitely easy to reproduce (at least on my system).

Sigh.


OS X 10.10 (Yosemite): Preview fails to keep up with fast window resizing (continued)

Posted by Pierre Igot in: Macintosh
March 9th, 2015 • 1:31 pm

Since my original post about a bug with window resizing in Preview in Yosemite, I have received feedback indicating that not everyone is able to reproduce this bug on their system.

This leads me to suspect that the problem might be limited to specific hardware configurations, like the other responsiveness issues that I have be experiencing since upgrading to Yosemite, which seem to all involve issues of drawing/refreshing the contents of windows, menus, etc. on the screen.

In my original post, I indicated that, in order to reproduce the problem, you really had to use very quick mouse movements in Preview, which will cause it to fail to keep up with your window resizing and only partly resize its contents to fit the frame.

I also indicated that the issue was particularly noticeable when using Keyboard Maestro macros, because these macros effectively mimic very fast mouse movements — in fact faster than what the human hand can achieve.

This means that, while reproducing the problem with a human hand can be challenging, with a Keyboard Maestro macro I am able to reproduce the bug reliably 100% of the time.

So in the interest of further identifying the circumstances under which this Yosemite bug occurs, here is a scenario involving Keyboard Maestro that makes it possible to reproduce the bug without fail:

  1. If you don’t have it yet, download and install Keyboard Maestro on your system.
  2. Create the following macro (or download it here and import it) or a similar one:
  3. km-resize

  4. Open any PDF document in Preview and switch the view mode to “Single Page” (in the “View” menu).
  5. Make the Preview window containing the PDF fairly small.
  6. Execute the macro.

If you have the same bug I have on my system (a 2014 Mac Pro), then the window will resize properly, but Preview will entirely fail to scale the contents to fit the resized window, as it normally should.

The only way to force Preview to scale the contents of the window is to adjust the window size a second time. This will force it to “refresh” the contents and redraw the page at the size it should normally be in a window of that size.

If you can reproduce this bug on your system, I would be very interested to know about it. Please use the link in the sidebar to contact me and send me your hardware configuration.

Thanks!


OS X 10.10 (Yosemite): Preview fails to keep up with fast window resizing

Posted by Pierre Igot in: Macintosh
March 6th, 2015 • 5:56 pm

Here’s a typical example of the many things that are wrong in Yosemite in terms of performance and responsiveness. When you first upgrade to Yosemite, you might notice all kinds of different responsiveness issues and it can be all quite confusing, giving you a general impression of unresponsiveness even though you cannot quite clearly pinpoint what is wrong.

Over time, however, it gradually becomes easier to identify specific behaviours in specific applications that help explain this general impression. Here is one such specific behaviour.

It has to do with the Preview application and with what happens when you view a PDF in “Single Page” view mode and you resize the window to make it bigger.

In order to reproduce the problem you need to do the following:

  1. Open a PDF file (any PDF file will do) in Preview.
  2. In Preview, go to the “View” menu and select the “Single Page” option (as opposed to the default “Continuous Scroll”).
  3. Then grab the bottom-right corner of the window and resize it to make it quite small.
  4. Then grab the bottom-right corner of the window again and drag it to the right and down as fast as you can with your mouse.

The speed is important here. If you drag it at a “normal” speed, you won’t encounter any problems.

If you drag it fast enough, on the other hand, sooner or later this will happen:

(Click here to get the MP4 video on DropBox.)

Note the situation in the video around the 02:00 mark: Even though the window’s bottom-right corner is near the edge of the video frame, the bottom-right corner of the PDF page is nowhere near that edge. It’s because I’ve resized the window too fast for Preview, and it has failed to keep up with my mouse movement.

But I’ve kept my mouse button down. And around the 03:00 mark, I nudge the mouse pointer a little bit more, which forces Preview to finally refresh its display of the PDF page so that it properly fills the space available inside the window’s edge (allowing for a tiny bit of padding for esthetic reasons, of course).

How pathetic is this? My hardware is not to blame: I have a very powerful 2014 Mac Pro with tons of RAM, and there’s nothing running in the background that would excuse such a shortcoming. (In actual fact, the ScreenFlow application is running and recording all this in real time, of course, but this behaviour is perfectly reproducible even when ScreenFlow is not running.)

Now, you might be tempted to say: “Yeah, but who resizes at that speed with the mouse anyway? If you resize at a more ‘normal’ speed, there is no problem here.” And that might be what Apple’s engineers think. (In other words, things are “good enough” for most users here.)

Well, for one thing, this is a pretty sad attitude to have. There is no question that my Mac is powerful enough and that Preview should not be struggling with this. Any self-respecting engineer who strives for excellence in his or her work should be dissatisfied with this and insist that the problem be fixed.

And then there’s the fact that it effectively penalizes users for being too fast! In what world is that considered OK? We live in an age of superfast hardware that should have no trouble keeping up with human interactions, especially when everything is local and no networking is involved. Yes, I am a fast user. So what? Does it really make me a bad user?

Finally, there’s one more thing: This shortcoming has a significant impact on people who, like me, might use Keyboard Maestro to automate various tasks, including the moving and resizing of windows in OS X. I have two large monitors and I have several macros that enable me to automatically move and resize my windows in any OS X application, including Preview, so that I can easily organize my document windows side by side on either screen.

I often have four different documents open in four different windows side by side and my large screens enable me to me to view these documents at a fairly low zoom level while maintaining legibility. But the two screens do not have the same size/resolution, and so when I switch a document window from one screen to the other I need to resize it. If that resizing involves a PDF document open in a window in Preview, then guess what happens?

Well, for window manipulation, Keyboard Maestro’s macros effectively behave like a superfast user, mimicking mouse movements at the fastest possible speed. So of course they are going to be affected by this bug in Yosemite’s Preview application. Sure enough, every time I use a macro that includes a step like this one:

km-move-resize-window

and the window involved happens to be on my second screen, where the window size has to be smaller, the macro correctly moves it to the main screen and correctly resizes the window to fill the bigger screen, but Preview fails to keep up with it and so the PDF page in the window is not resized properly.

In fact, Keyboard Maestro is faster than I’ll ever be with my mouse, and it’s so fast that Preview fails to resize the PDF page itself altogether! The window is bigger, but the PDF page keeps the same size it had on the secondary screen.

The only solution that I have found to work around this problem is to add an extra step to the macro, just for Preview:

km-move-resize-window-nudge

This extra step effectively mimicks the little nudge illustrated around the 03:00 mark in the video above, which forces Preview to refresh its display of the PDF page so that it finally fills the space available inside the window.

But it’s not an ideal solution, because any extra step like this increases the risk that something might go wrong during the execution of the macro. If Yosemite experiences any kind of “hiccup” during the execution (which it tends to do), then the nudge might not work as expected and I will still have to resize the window manually in order to force Preview to refresh its display of the PDF page.

This is a new problem in Yosemite. Preview in Mavericks never had such a problem, and I had been using my Keyboard Maestro macros with it for a long time without a hitch.

When people talk about Apple’s sloppy work in Yosemite, I believe that this is the kind of thing that they are thinking of. It took me a while to narrow it down because of all the other responsiveness problems I was having with Yosemite, but this, as far as I can tell, is an easily reproducible problem for anyone running Yosemite, even people who don’t have the egregious responsiveness problems that I have been experiencing.

When will Apple get around to fixing it (if ever)? Unfortunately, because the application works “well enough” for the majority of users, I am not too optimistic, which is why I’ll have to keep this extra “nudge” step in my macros and in my own dragging movements for quite a while, I fear.

rdar://20069961


Word 2011 for OS X: Easily confused

Posted by Pierre Igot in: Microsoft
March 5th, 2015 • 3:44 pm

If you’ve ever felt like confusing the heck out of Word 2011 for OS X just for fun (and who wouldn’t, really?), try this:

  1. In the Finder, create a folder called “Confused Word” on your desktop.
  2. In Word 2011, create a new document and save it as “Test1.docx” in the “Confused Word” folder. Leave the document open in a document window in Word.
  3. In the Finder, select the “Test1.docx” file in the “Confused Word” folder, make its name editable, and change it to “Test2.docx”.
  4. Switch back to Word 2011 and confirm that the document window you left open now has the document title “Test2.docx” instead of “Test1.docx”.
  5. Switch back to the Finder, select the newly renamed “Test2.docx” file and duplicate it to create a second file in the same folder called “Test2 copy.docx”.
  6. Make the file name “Test2 copy.docx” editable and change it to “Test1.docx”.
  7. Double-click on “Test1.docx” to open it in Word 2011.

The first strange thing that happens is that the “Test1.docx” document opens as “Read-Only”. Why? I have absolutely no idea.

But now hit command-S in Word 2011 to save the “Test1.docx” document that is labelled “Read-Only”.

Since it’s (allegedly) read-only, you’d think that Word 2011 would complain about your attempt to save it and ask you to save it under a new name or something, right?

Think again. Not only does Word 2011 not complain, but it actually changes the document title in the title bar to “Test2.docx (Read-Only)”!

As they say on the Interwebs, WTF?

It gets even funkier if, before hitting that command-S shortcut, you actually make some changes to the contents of the “Test1.docx” document. In that case, when you hit command-S, Word changes the document title in the title bar to “Test2.docx”, without the “Ready-Only” part. And guess what happens to the “Test2.docx” document that you’d left open in the other document window in Word 2011?

It gets renamed to something like “Word Work File D_xxxxxx.tmp”, where “xxxxxx” is a sequence of numbers. Admittedly, this is a very temporary situation, because as soon as you switch to another app and then back to Word 2011, the document title of that other document window changes back to “Test2.docx” as well.

So now you have two document windows with the same document title/file name and two different states for the content (since you typed something else in that other document window before hitting command-S, remember?).

And now guess what happens when you close those two document windows and return to the Finder?

Well, there are no longer two files inside that “Confused Word” folder on your desktop. There’s only one file left, called “Test1.docx”, and its contents do not include the changes you made to the document before hitting command-S.

Nice to know you’ve got our backs, Microsoft.