The Finder’s GUI tax can be very expensive

I agree there should be visual feedback, but maybe more of a compromise. E.g. bundle all unzip tasks into 1 single progress bar and rotate text strings for informing which task is underway.

The "dancing" modal is quite ridiculous.

nayuki• 10 years ago

This is basically what happens in third-party tools. For example if you use 7-Zip on Windows to unzip multiple archives, there is only one progress window and progress bar, and it will display the names of each archive being extracted one after another.

wruza2• 10 years ago

If operation takes more than human's average reaction time, progress should be shown. Usually 250ms. Jef Raskin and others wrote a bookshelf about that a decade ago, but I still see strange solutions everywhere.

pvdebbe• 10 years ago

J Nielsen would disagree with you. A subsecond operation could very well operate without visual feedback and only after that start showing the progress. Does finder have a way to show messages in status bar or some such? For quick ops it would be ideal to say, "operation completed".

kalleboo2• 10 years ago

Apple disagrees - the Finder already handles this without any visual feedback for file copy operations - you don't get a file copy progress dialog if it takes less time than you need to react (just an audio confirmation).

TheAceOfHearts• 10 years ago

If someone is looking for an alternative with a GUI, I use The Unarchiver [0] and I'm generally happy with it. I don't usually have to expand lots of small files like the OP, so I don't know how it compares with that task.

[0] http://unarchiver.c3.cx/unarchiver

curun1r• 10 years ago

Seconded. The Unarchiver is always one of the first things I install on a new Mac.

The same author is also responsible for another program, Xee, which is a great, no nonsense image viewer. The 3.x version costs a few bucks, but the last 2.x version can still be downloaded for free.

syrrim• 10 years ago

The unarhiver also handles .rar files, whih archive utility can't do.

Normal_gaussian• 10 years ago

If we are assuming the underlying expansion is as efficient then the cost comes either in the boilerplate or mistakes made when parallelising. In the case of the boilerplate the solution would be embedding some more of the linked code and reducing IPC/other unnecessary bottlenecks. However I suspect the error is in the parallelisation, which is not surprising, especially as the authors likely optimised and tested for large, not small, files.

As it is, removing the GUI is perhaps the worst thing you could do to the user. Closely followed by inducing epilepsy with that ridiculous expanding dialog.

cesarb• 10 years ago

a window with a single progress bar for the entire task would be OK, but would still slow operations down.

There's a simple way to do a visual progress bar with almost zero slowdown: run the task in a separate thread from the progress bar, and update the progress through lock-free shared variables. Make the progress bar read the shared variables only a couple of times per second, sleeping between its updates.

fredsir• 10 years ago

It's not only the communication between the task and the GUI that is slowing things down (if it is at all, which I doubt). Even more, I think, it's the fact that macOS, the OS in question, has some conventions for windows, controls and so forth, one being that it takes X seconds for the window to appear and X seconds for it to disappear. If it was being shown and hidden again as fast as the task starts and finishes, it would be just be a flicker on the screen, which would confuse the user. It also doesn't make sense to spend a second to show a window, if the progress bar it embeds has completed before the second has elapsed, so there has to be a delay there, and so forth. The task is so quick it can really not be presented to the user without the presentation being slower than the actual task.

I think it comes down to good UI+UX, which in this particular case becomes bad UI+UX.

oneeyedpigeon• 10 years ago

I think an 'in-line' style of alert can work well in this situation, like a status bar but closeable. It can be dealt with at a time most convenient to the use, still provides feedback. It could even just display the output of a CLI command with a given flag, even for a small subset of 'blessed' commands if necessary.

userbinator• 10 years ago

If it was being shown and hidden again as fast as the task starts and finishes, it would be just be a flicker on the screen, which would confuse the user.

Maybe the really inexperienced users, but it makes sense that something which happens quickly, will mean a window that also appears and disappears quickly. This is why Windows has options to disable animations and other effects, and everything does feel noticeably more responsive when they are disabled --- windows and menus appear and disappear instantly.

waterhouse• 10 years ago

If they're simply following UX recommendations blindly, they could go for this one:

"Use a progress indicator for any action that takes longer than about 1.0 second."

https://www.nngroup.com/articles/progress-indicators/

You don't necessarily know in advance whether it'll take 1 second, but I think a strategy that would work is: after 1 second, if the task is not done, open a progress bar, and if the task completes immediately after that, display the completed progress bar for 500 milliseconds or something before closing it.

__jal• 10 years ago

They should make the UI idiom feeding your file a mushroom and then double the icon size on completion.

robotmlg• 10 years ago

only if they want to get sued by Nintendo...

headcanon3• 10 years ago

Nintendo got it from Alice in Wonderland, so perhaps prior art?

DominoTree• 10 years ago

The window animations in Mac OS itself can make things feel very sluggish as well, although the velocity-based ones seem better with Sierra.

module0000• 10 years ago

Just want to add, `unzip` from your terminal window can be used as well. If opening those gz's every month is a pain, you could handle it with something like: `unzip -r my_stuff_{1..99}.zip`

proaralyst• 10 years ago

The files in the article aren't zips, they're individually compressed files. Thus, unzip won't work.

jcoffland• 10 years ago

Windows explorer unzip is even worse. Recently in a VM I tried unzipping boost 1.62.0 using explorer in Windows 10. It said it would take over 3 hours. The same file was decompressed on the command line in about a minute using 7zip on the same VM.

Piskvorrr• 10 years ago

For a tool that's supposed to be on most desktops by default, Explorer has some strange inefficiencies, such as the estimation and the loooong "preparing to copy" phase. Perhaps they are carried over from the old Windows NT codebase?

wruza2• 10 years ago

Of course they are. MS is biggest visual theme seller on the planet.

3r1kB• 10 years ago

Reminds me of this XKCD haha https://xkcd.com/612/

Someone• 10 years ago

Windows also always unzips to a fixed scratch directory before creating the real files.

That scratch directory is %TMP% or %TEMP%, I think, but it definitely always is on the same disk, typically C:

If you unzip files to another disk than where that directory lives, it unzips to its scratch directory, then _copies_ the extracted files to their destination, and finally deletes the scratch files.

That's bad in itself, but doubly so if the disk where it unpacks doesn't have room for the unzipped files.

Mac OS had a system call "Give me the temp directory on this disk" (FindFolder) that made it possible to extract to a temp directory and then move the result to the destination. That call made it into Carbon. I don't think macOS still has it, but it seems UnArchiver does the right thing, anyways.

aidenn02• 10 years ago

Internet explorer did the same thing with downloads; I discovered this when I had a 1GB C: drive and a 6GB D: drive and tried to download something that was too large to fit on the C: drive.

theandrewbailey• 10 years ago

For the longest time, I searched for a GUI archiver that didn't use temp directories at all.

patio11• 10 years ago

I wonder whether this is not an example of a UI interaction which is slow on purpose, to emphasize how much work the OS is doing for you.

TaxAct and TurboTax, for example, both operate on (in the typical case) kilobyte scale data requiring trivial math. They also make both saving the data and calculating taxes take 5+ wall-clock seconds when they actually require milliseconds and nanoseconds respectively. This is largely because (non-technical) users don't trust your computer did the math right if the answer comes back instantly. (I also suspect there is an element of "Wait if it is so easy to calculate my taxes why am I paying you to do that.")

TazeTSchnitzel3• 10 years ago

This reminds me of how some PlayStation game's developers complained that they were required by Sony to add a false multi-second delay to their game's save routine, so the user would have enough time to read the mandatory warning not to turn off the console. Their game in fact saved so quickly there needn't be a warning at all.

matheusmoreira2• 10 years ago

This is really interesting. I read that warning every time I save and I never noticed the delay. For some reason, it doesn't bother me like the archive utility animations. I guess it's because saving the game is a such ritualized action.

I wonder what other requirements Sony imposed on developers. Maybe they have their own video game-oriented human interface guidelines. Do you have a source for this information? I found plenty of technical documents and SDKs but nothing about interfaces.

_v7gu• 10 years ago

I'm never actually sure if I actually saved in the PC port of Valkyria Chronicles. I think we've been trained to wait a while to save since the 8 bit consoles so something feels a bit wrong when things instantly save. Undertale makes amazing use of this, though.

8ytecoder• 10 years ago

No kidding. In one of my previous companies we explicitly added a delay in our financial webapp. We saw a significant improvement in bounce rate after that.

seanp2k2• 10 years ago

Hahaha, I remember coding some physics problem solvers in Visual Basic back in high school, and I added a "calculating" progress bar and messed with the timing a lot to make it still pleasant to use but seem like it was really crunching through your problem. "Crunching" is even funny and a throwback to the days of spinning rust + lots of random seeks.

pvg8• 10 years ago

That's highly unlikely. Archive Utility is just slow.

eric_h• 10 years ago

I think you're probably right. I'd imagine that most of the code in Archive Utility.app is very old, and there's very little business reason to spend any engineering time on improving it beyond doing just enough work to make sure it works with API changes for later OS releases.

There are third party GUI compression utilities that are faster that you'd find if you actually care about how long it takes to decompress your archives, and the CLI utilities are all there out of the box for people who don't need a GUI.

mwfunk• 10 years ago

What is highly unlikely about that? That seems more plausible than, zlib is magically 1000x slower when used by Archive Utility.

coldtea3• 10 years ago

It's highly unlikely that it's 1000 times slower than the command line utility to decompress, with which they share the same libs anyway...

radley2• 10 years ago

No, it's about showing it long enough so you know what the dialog is and what it's doing.

If a dialog popped up for the brief amount of time it actually takes to unzip, it might freak out users: "huh? WTF was that??"

arnarbi2• 10 years ago

Then just don't show a dialog.

yoz-y• 10 years ago

Maybe a new type of dialog could help? Rather than showing this blinking things, just show a notification of style "24 archives extracted" (kind of like what forklift does)

kalleboo2• 10 years ago

The Finder already handles this differently for file copy operations - you don't get a file copy progress dialog if it takes less time than you need to react (just an audio confirmation). It seems like Archive Utility didn't get that extra attention though.

aplummer• 10 years ago

I have been made to do this, I called the function wasteMyTime() and all it did was waste the user's time for a few seconds with a spinner. Still a little bitter about it.

homero• 10 years ago

Yes I've added delays in many web apps, even loader gifs to show fake hard work

hawkice2• 10 years ago

I've been asked to blank out a screen in a single page web application and re-render after 250ms when people switch 'folders'. My boss thought it was a familiar user interface pattern to draw attention to the page content. This is... not the silliest thing I have done professionally.

noblethrasher• 10 years ago

Don Norman talks about that example of intentional delays in some tax software here: https://vimeo.com/96714148 (approx 52:30).

It sounds like he and his team came up with the idea in that instance.

interpol_p• 10 years ago

I don't think that's the reason. It's more of an edge case due to some minimum animation timing.

If you were expanding 24x 100MB, or 1GB files, the overhead of animation would be much smaller relative to the total time. And it would be quite an informative GUI showing the progress of each file. While the command line version may sit there for a long time before returning to prompt.

coldpie• 10 years ago

Mercifully, in those tax applications you can often just click the "Next Step" button to skip the useless "loading bars". I laughed out loud when I discovered this.

headcanon3• 10 years ago

I've encountered this in command-line programs as well. Print statements take a non-zero amount of time and depending on the language, can be quite slow, so for an otherwise-fast operation in a loop, if it prints a line every time then it can contribute significantly to the overall speed of the program. Removing the print statements in some cases can speed up the code by an order of magnitude.

dijit• 10 years ago

quiet reminder that writing to file descriptors is blocking.

STDERR and STDOUT are file descriptors.

orf• 10 years ago

Yep, and this is very noticeable if your using windows and select any portion of the terminal. All stdout/stderr is blocked.

I had a call when one of our services stopped responding. Took a while to figure out there was a tiny one-character selection in the power shell.

thaumasiotes3• 10 years ago

Maybe so, but, as pointed out elsewhere in the thread, printing to stdout is much, much, much, much slower than printing to a file. The fact that stdout is a file descriptor can only explain a vanishingly small quantity of the time cost of using it.

anthk2• 10 years ago

Avoid verbose flags on tar.

jmiserez• 10 years ago

rsync as well, especially the progress bars.

vortico• 10 years ago

This will be sort of bluntly stated, but I challenge anyone to show me an example of where animation is a usability improvement over no animation at all. I understand how it may be visually pleasing, but I work on computers, not enjoy it as an art form. I do some UI design too and don't remember the last time I added animation. Can anyone find an exception?

pvg8• 10 years ago

Most transition effects where things are added or removed from a group of objects - without the visual cue, this can happen so quickly it's hard to see where the removal or addition happened. The tab bar in your browser is a good example.

xsmasher• 10 years ago

Many animations are simply stylistic, but others convey information.

Any animation that gives you "almost there" information about a gesture; fro example, holding your finger on the screen in Minecraft PE starts an animated circle. When the circle completes, the action is complete. It provides extra information.

Ditto for the "swipe right to delete" animations in iOS mail; the animation lets you know which side of the "show options vs delete" wall you're on, and how close to the edge.

appleflaxen• 10 years ago

the rotation animation when turning your phone to landscape mode.

ythn• 10 years ago

There's a point where printing progress is actually a huge bottleneck to performance. Try it out in python. Make a for loop that loops a million times and prints "Hello". Run the program once normally, and a second time piping the output to a file. You'll find the first one takes 15 seconds, while the second one is near instantaneous.

goldenkek4• 10 years ago

This is a false equivalence. As long as the progress is set correctly via messaging (queue) to an alternate UI thread, the time lost will be under a millsecond. The naive print loop you provide is obviously IO bound and print is a blocking IO call. Use an async print IO and youll have much better performance. Narf!!

castratikron• 10 years ago

Isn't there some disk caching going on here? Surely the OS is not writing directly to disk immediately after gunzipping. I don't think that's enough to explain the large difference, though.

bluedino• 10 years ago

Reminds of of this Raymond Chen post:

https://blogs.msdn.microsoft.com/oldnewthing/20060220-00/?p=...

A Windows program ran faster if you clicked and held the mouse button on the title bar (as if you were going to move the window) because it would stop the window from re-drawing itself, making whatever loop it was running (and constantly updating the GUI) go faster.

michaelbuckbee• 10 years ago

It's fixed now, but for a while it was significantly (20 to 40%) faster to install npm without the progress bar.

https://github.com/npm/npm/issues/11283

couchand• 10 years ago

Just to clarify, that's 'npm install' not 'install npm'.

somanychoices• 10 years ago

The rationalizations are kind of silly... oh no, the uninformed users might otherwise do or think this or that; actually, this is shit workmanship, that is all it is, and it's up to the people who have the faintest clue to point at it an laugh until it becomes better workmanship. If you use energy you should use attacking it on defending it, you're now part of it.

rebuilder• 10 years ago

Even worse, IMO, is that the default behaviour for Archive Utility is to just unpack the contents of a zip file in a directory under which it's found. Sometimes I just want one file out of a huge zip... Actually, quite often I want to see what's in a zip before uncompressing.

whywhywhywhy2• 10 years ago

Great example of someone justifying engineer driven design over user driven design. Not understanding at all why it was made that way in the first place.

stefs• 10 years ago

"... slow, it’s the GUI interface to the computer ..."

he ... he wrote GUI interface. heh.

The Finder’s GUI tax can be very expensive

Related Stories