Just an update on this:

I think this is somewhat a known issue, see e.g.

Will actually mark this as incomplete until we either have confirmation it works in new builds (or problem still persists...)

Cannot reproduce when using rendered viewport (both Eevee and Cycles)

Another thing I noticed when having an animated custom property is that even other keyframes might not be evaluated anymore (and this is even true for 2.79 with new depsgraph)...:

@Antonio Vazquez (antoniov): asked in IRC if I could also have a look, and here are my findings (sorry for the lengthy detour, this might all be totally obvious for others...)

Could you please share a bit more information here:

@Thomas Gropp (togro) : have you tried upgrading your nvidia driver?

could you share the .blend file here?

Please:

• Give us your exact GPU (including drivers) versions, as requested in the template!
• Ensure both your OS and drivers are fully up-to-date (and use official GPU drivers, not those provided by windows or tablet/laptop maker).
• Try to disable any running antivirus.
• Ensure you have no python executable available in your %path% envvar.
• Try to re-download and re-install Blender.
• Try the latest build from our builder.blender.org.
• Try to start Blender in factory settings (--factory-startup commandline option) (this will ensure whether this is a userpref or addon issue or not).
• Try to tweak your GPU driver settings (e.g. try different values between 'performance' and 'quality' if you have such slider, etc.).
• Launch Blender from the command line with -d option and attach as text file here any error printed out in the console (do not paste it directly in comment).

Before digging deeper myself, I would like to ask @Lucio Rossi (luciorossi) if this is expected behaviour?

Can confirm in 2.79b (official release), but not in 2.8 nor 2.79 master.

Pretty sure this is a known limitation of the Unwrapping algorithm (ngons - especially non-flat ngons), so you are probably better of using other methods like Smart UV project.

Ah, good to hear, closing then (feel free to reopen if issues come up again)

@tin truong (a0s): so better remove the environment variable again (to avoid future issue)

@Brecht Van Lommel (brecht): sorry about that [wont do again]

Hm, last idea for now (regarding userprefs crash):
Have you tried deleting the UserPrefs [found here]?
Or started with the --factory-startup option?

Confirmed (in both 2.79 and 2.8), will have a look...

So nuking userpref.blend should do it?

review comments

reporting devices correctly here [but there i obviously something wrong with the AWS cards...]

address review comments by @Brecht Van Lommel (brecht)

oki, waiting for @Sergey Sharybin (sergey) then.

quoting @Brecht Van Lommel (brecht) from D3807:

Almost certain this is not the right fix, but dropping this here as I couldnt find out where mvert, mpoly, mloop should be filled again otherwise.
This might be a known limitation even (given some code comments in this area that this is still kinda WIP)

Hm, glad we have a solution, even though the root of the problem is still not found...
Anyways, closing then.

Cannot reproduce.

just a shot in the dark, but are you possibly using a tablet?
There was T54493 and the solution there was:

Not a windows guy, but I think in a command prompt
SET CYCLES_OPENCL_TEST=NONE should do it, you can check with
ECHO %CYCLES_OPENCL_TEST%

Is that including the crash? (please make it crash)

Thanx for that (keep forgetting crashlog is disabled for windows)
Are there any (error) messages when launching blender from the commandline with the -d option?

Thanx for the report, these crashes are always hard to track down, Please:

Have you tried uninstalling, then installing again?

commited rBAa24896c2dbbb

Please don't triage priorities yourself, someone will look at it...

Can confirm whenever multires is used and smooth() is involved (Smooth Brush, Mask Smooth) -- cannot reproduce using anything else...

D3800 actually seems to fix & survive (limited) tests (but not totally sure if other problems might arise...)

Ah, yes, there's a difference between F12 and Ctrl+F12...

Thanx for the report.
It is very well possible something is wrong here (I can sort of reproduce already), but can you share a simple .blend file with short instructions to reproduce?
(this way not every developer looking into this has to set up an own file [time consuming process], also we are all on the same page then....)

Please follow our submission template and guidelines and make a complete, valid bug report, with required info, precise description of the issue, precise steps to reproduce it, small and simple .blend and/or other files to do so if needed, etc.

I cannot reproduce.

Hm, I cannot reproduce the crash (linux).
What I can reproduce is the behaviour of T57125 (element selection drawing not in sync between instances) -- but not the crash...

You are right, this is not a per-viewplayer option, but instead a scene render setting.

Also cannot reproduce.
Marking as incomplete until we have a crash log...

Seems like a duplicate of T56063

Think we can close this then?

@Brecht Van Lommel (brecht), @Dalai Felinto (dfelinto) : I have argN working fine [no leaks for the workspace add menu], but am still looking at the nested menu case [collections], expect update tomorrow...

Commited rB1fad887d9e22da97009f28288cc6f21cc32a6c37

Commited rBae3ac8062f4786938fcbd3cbe1b0151a54818821

Cannot reproduce, working fine here...

Cannot reproduce here, someone on a MAC here could check?

File renders fine here in a bit less than 10sec.

There were some similar reports and 'usual' response there was:

like I said in T57238, similar pie is already included in blender default, not sure if we better remove it from the addon then?
(or keep if someone is not using Blender default keymap?)

Can confirm this will spawn an empty pie atm, will have a look...

Looks like something went wrong with the graphics driver.
Have you tried updating the graphics driver?

Glad to hear, closing then...

Confirmed, will have a look later...

Thanx for the tinker.obj, but obj.mtl is missing...
Without that, I cant reproduce in 2.8 or master...

Can confirm this (although I am not sure this is solvable easily)
`
Subdivide smooth depends on vertex normals. These are mostly fine when having meshes with faces but are more "problematic" in edge-only meshes.
(you can inspect these with the Normals Overlay)

Unless I am misunderstanding something here, I am seeing the exact same behaviour in 2.79b / 2.78c as well.
Could you show us this is working in 2.7x?

D3784 could do it, but I am unsure if this is a good solution.
(after all this is a cornercase and having a non-selectable object still the active object is rare).
(also not sure if we should do that for editmode gizmos as well?)