Affinity Photo Customer Beta (1.10.0.247)

[Florida]

OK I did the control launch thing and selected every checkbox. Then ran the exact same scenario - same beachballs, two layers pasted with no content. Clearly not related to default settings but must be hardware or OS related then.

One difference I see from you is that I running Big Sur 11.4

[R C-R]

4 minutes ago, Florida said:

One difference I see from you is that I running Big Sur 11.4

Maybe someone else running Big Sur will text with your files & see if they get the same results you do....

[Florida]

Would be nice to fix this, if I can't work with this size image, I might have to use Photoshop (I'd rather use AP)

[Florida]

I tried the same experiment loading only the big image and one smaller one, and then copying/pasting - still a beach ball.

[jdvoracek]

11 minutes ago, Florida said:

I tried the same experiment loading only the big image and one smaller one, and then copying/pasting - still a beach ball.

What do you mean by a "beach ball"? I followed your directions, and my final document looks identical to that posted by @R C-R.

Regards, - - john

[Florida]

The macOS Beachball cursor that shows up when the app is chugging. Basically the OS shows this cursor when it detects the app is stuck. There is no reason why it should have any issues copying these images other than it must be related to hardware, there is nothing oddball about a 2017 iMac 5K and this never happened in versions prior to 1.9.

Note I also have 8 spin reports in Console from today from AP having this issue, if some programmer at Affinity would find them useful. I've attached one to this message. MacOS generates these when it detects an app is stuck. Only meaningful to someone with access to source code. The heaviest stack trace function is this:

void Raster::PixelProcessor<Raster::Horizontal>::Process<Raster::Red8, Raster::Green8, Raster::Blue8, Raster::Alpha8, Raster::X5, Raster::Index8, Raster::X2, Raster::X3, Raster::X4, Raster::X5, Raster::X1, Raster::X2, Raster::X3, Raster::X4, Raster::X5, Raster::Bitmap<Raster::Red8, Raster::Green8, Raster::Blue8, Raster::Alpha8, Raster::X5>, Raster::Bitmap<Raster::Index8, Raster::X2, Raster::X3, Raster::X4, Raster::X5>, Raster::EncodePalette, Raster::IdentityMask, Raster::IdentitySampler, Raster::IdentityTransform, Raster::Safe, Raster::NormalBlend>(Raster::Bitmap<Raster::Red8, Raster::Green8, Raster::Blue8, Raster::Alpha8, Raster::X5> const*, Raster::Bitmap<Raster::Index8, Raster::X2, Raster::X3, Raster::X4, Raster::X5>*, Raster::EncodePalette&, Kernel::RectT<int>, bool, Raster::IdentityMask::Params const*, Raster::IdentityTransform::Params const*, Raster::IdentitySampler::Params const*, Raster::NormalBlend::Params const*) 

Your iMac is still way different from mine. It clearly related to something about this hardware, there is nothing else different now.

spinreport.txt.zip

[jdvoracek]

47 minutes ago, Florida said:

Your iMac is still way different from mine. It clearly related to something about this hardware, there is nothing else different now.

Thanks for the beach ball definition. A while back a subset of us with MacBook Pro's running the same AF and MacOS but with some hardware variances lacked the custom input controls for the textural procedure live filter. Serif eventually found and fixed the issue in Beta but it's still present in the official 1.9.3 so sometimes Beta is a good thing. Sorry for your issue.

Regards - - john

[R C-R]

1 hour ago, jdvoracek said:

What do you mean by a "beach ball"?

If you Google "macOS spinning beachball" or the like, you will get tons of hits. Officially, it is called the Spinning Wait Cursor. It indicates that the app is busy & cannot accept new user input. It does not necessarily mean the app is hung -- often it will disappear a few seconds after it appears -- but if does not disappear after waiting a while, you can assume the app is no longer responding.

[Old Bruce]

I have always thought of the 'Beachball' as a child's toy multicoloured Pinwheel, it spins furiously but does no actual work.

[Florida]

It's not just copy paste that does this oddness, I get the same with any operation that creates a new layer, like rasterize. It often locks up the Mac entirely, then you wind up with an empty layer. Hard to get anywhere. Note that if the OS determines the app has frozen the Window Server (the thing that runs the UI) for too long it force shuts everything down and you have to re login and start over. I try to stay away from those 🙂

[R C-R]

2 hours ago, Florida said:

Note that if the OS determines the app has frozen the Window Server (the thing that runs the UI) for too long it force shuts everything down and you have to re login and start over.

??? The OS is fully capable of terminating an app without requiring a user to log out & back in again.

[Florida]

WindowServer is what drives the UI in MacOS. If something freezes that for long enough all windowed apps have to be killed in order to restart. Only happened to me once at least, but it started with a freeze in AP.

Basically if the OS watchdog fails to see a checkin from WindowServer within some timeframe, it figures that it's dead and starts the recovery process. It's like a kernel panic except it only affects apps with windows, the OS itself can recover.

[R C-R]

2 hours ago, Florida said:

WindowServer is what drives the UI in MacOS. If something freezes that for long enough all windowed apps have to be killed in order to restart.

But that "something" can't be a user domain application process, nor can a user domain process cause the OS 'watchdog' to fail. 

If an app stops responding -- you get a never-ending beachball whenever the pointer is over its window -- you can still use the Force Quit option or Activity Monitor to force it to quit. There is no need to log out of the current user domain account or to kill any other app(s) besides any that stop responding.

Whatever happened to you that caused WindowServer to freeze, it was not caused directly by AP or any other app you were running at that time.

[Florida]

AP clearly caused the issue as I see it over and over again, generally only for a few seconds at a time; that time the UI of the entire Mac froze and did not return until I was face to face with the login screen, and after logging in all the apps were restarting.

When the mouse cursor freezes in place and does not move, the UI is not working. Most of the time it comes back. Often you just see a beachball.

 I don't think necessarily that AP is to blame but it is triggering the issue, which could be in some other system or kernel extension or driver. Window Server is a process likely any other. I see the OS generate "Affinity Photo Beta_2021-07-08-094949_Andrews-iMac.wakeups_resource.diag" files every day. Clearly MacOS sees the issue.

[Florida]

I restarted the Mac and with no other apps running at all, the problem does not occur. So it is environmental. Maybe the VM requirements of AP require much swapping of memory. I only have 16MB of ram you have 32MB. Perhaps AP requires more ram for larger images than the OS (or AP requesting memory) can deal with. I have seem desktop and server systems lock up trying to keep up with crazy swapping before.

Ordering more RAM.

[R C-R]

1 minute ago, Florida said:

I restarted the Mac and with no other apps running at all, the problem does not occur. So it is environmental. Maybe the VM requirements of AP require much swapping of memory. I only have 16MB of ram.

I doubt your problem is caused by insufficient RAM or swaps with the backing store. On my Mac, even if I set this beta's RAM limit to something absurdly small like 3 GB & load the app up with lots of high resolution files, all I get are some short-term lags or ~10 second max beachballs when I perform certain operations on jumbo sized raster layers (like 18K x 18K px), & even with that most operations remain fluid & lag-free.

I think it is much more likely that some other app that you previously were running was causing a problem, perhaps one running in the background you were not aware was still running.

[mdriftmeyer]

The spinning beach ball is a non-modal state to allow the system to manage processes dealing with cache, swapping, writing and more and thus whenever one hits a wall with physical memory and core activities it will surface. This goes back to our NeXTSTEP operating system us NeXT folks developed and brought to Apple. And first rule of thumb in any large image processing, video scrubbing, audio mixing, etc., you max out your system RAM and get the fastest drives you can. 

[R C-R]

27 minutes ago, mdriftmeyer said:

And first rule of thumb in any large image processing, video scrubbing, audio mixing, etc., you max out your system RAM and get the fastest drives you can. 

A good rule of thumb but do consider that one of the goals for the Affinity apps has from the beginning been high memory efficiency. So among other things, unlike many similar editing apps it does not load everything into memory at once or keep everything in RAM continuously. You can see the effects of this on Macs in Activity Monitor -- so-called "memory pressure" can remain very low even with massive high resolution files open.

[fde101]

23 hours ago, R C-R said:

Whatever happened to you that caused WindowServer to freeze, it was not caused directly by AP or any other app you were running at that time.

It can happen if the app performs a series of actions that triggers a bug in the WindowServer.

[R C-R]

1 hour ago, fde101 said:

It can happen if the app performs a series of actions that triggers a bug in the WindowServer.

If there is a bug in the WindowServer that can be triggered by any action or series of actions in an app, that is an OS level bug. There is nothing the app can do to avoid triggering that other than not allowing that action or series of actions to be performed by the user.

[fde101]

9 hours ago, R C-R said:

If there is a bug in the WindowServer that can be triggered by any action or series of actions in an app, that is an OS level bug. There is nothing the app can do to avoid triggering that other than not allowing that action or series of actions to be performed by the user.

True, but if the series of actions performed by the app are sufficiently unusual that other apps do not commonly perform the specific steps in the specific sequence needed to trigger said bug then it may explain why it looks like the app is causing the problem.  The bug should be fixed in the OS either way if this is happening.

It may also be possible that this could be caused by a driver (or possibly a system extension?) in use by the WindowServer.