Seems LGTM. Just when you put all signing keys and account info to mac keychain, you could add those also into slave_pack.py

@Mugur Marculescu (mugurm) Could you test https://www.dropbox.com/s/uhzejcprowqcns2/blender-2.81-ndof-macos.dmg?dl=0 if that works. It is signed package.

In T67686#741013, @Brecht Van Lommel (brecht) wrote:

I'm guessing it's failing to dlopen the library due to more restricted permission for code signed apps.

We may have to either enabled this:
https://developer.apple.com/documentation/bundleresources/entitlements/com_apple_security_cs_allow-dyld-environment-variables

Or use an absolute path here:

module = dlopen("3DconnexionClient.framework/3DconnexionClient", RTLD_LAZY | RTLD_LOCAL);

Wait for buildbot to be able to build new one, it should have the new one.

Check python console and if the version is 3.7.4, your build is up to date.

Python is now updated to 3.7.4, check if it works now.

Roger.

Figure out: do issues in the daily builds risk affecting the release build? Do we use a separate certificate?

We use same build configuration on RC as with buildbot.
Try what happens with clean configuration (--factory-startup), RC3 and daily.

@Brecht Van Lommel (brecht) Yes it was.
@Tjerk Otten (tjerkotten) Does it work now on RC2 ?

I got crash just by enabling rendered.

Committed rBe60a01483ff0

...committed updated script.

added codesigning and notarization loop.

--source and --dmg mandantory
applescript variables set in the script itself, not sedded into place.

Start blender from terminal with --debug-all flag and post the output here.

What do you get when you put

>>> bpy.app.binary_path

into python console?

No instant crash anymore, track stays still on dopesheet, though.
That is gone if doing select all (even there is no tracks available).
Crash occurs now if there is track on dopesheet, but not on movie clip.
So:

In T61902#628317, @Brecht Van Lommel (brecht) wrote:

It should not be necessary to manually set WITH_BOOST in any case, that would indicate a bug in the build system that needs to be fixed.

What exactly you'd like to edit? Did you first run make update under source tree before cmake/cmake gui?

This happens when you have movie editor->tracking -> dopesheet open. clip_draw_dopesheet_main crashes after deleted track which it thinks it can draw.

In D4257#96827, @Stefan Werner (swerner) wrote:

Is it just me or did this break builds with Xcode? Builds using make work fine for me, but when using Xcode projects, datatoc can't launch because it can't find libomp.dylib.

Exception Type:        EXC_CRASH (SIGABRT)
Exception Codes:       0x0000000000000000, 0x0000000000000000
Exception Note:        EXC_CORPSE_NOTIFY

Termination Reason:    DYLD, [0x1] Library missing

Application Specific Information:
dyld: launch, loading dependent libraries

Dyld Error Message:
  Library not loaded: @executable_path/../Resources/lib/libomp.dylib
  Referenced from: /Users/USER/*/datatoc
  Reason: image not found

Added needed build_environment changes and using mkdir + cp instead of ditto.

It's iMac with 5k monitor. @wuxiaowei (cjxw87) indeed you should check the activity monitor if it's CPU or GPU thingie.

Confirmed. Short output from debug, so pasting it here rather than adding file =)

This seems to happen once in a while on some systems, but I can't repro. Debug build is next possible thing to get more info:
https://www.dropbox.com/s/hk0xxsloy6xeu7z/blender-2.8-ccc8d2ab427-macOS-10.9.tar.gz?dl=0

I read correctly, understood wrong. I'll build OSD sample app soon and will upload it. =)

For switching Intel vs AMD/Nvidia (on dual GPU systems) you can use gfxCardStatus But as said on irc, we're updating libraries, and that OpenSubdiv problem should be solved.

Nothing wrong with texture, if Volume is disconnected, the background texture is shown normally.

Confirmed. Unpacking file to disk and using that gives purple. Got crash while testing out, I'll investigate it further.

Slightly confusing, Yosemite and 10.13.6 version. Is this really High Sierra 10.13.6?
What computer? iMac 2011 ?

Committed as rB92217a81b59a, we'll work on from that.

We'll split these commits...

Land this, I'll make adjustments if needed.

macOS 10.9 libs here.

• macOS support for crashpad

Committed from git.

+1 for bump up to latest.

You need to use English for bug reporting.

@Campbell Barton (campbellbarton) @Brecht Van Lommel (brecht) how about landing this? Or shall we split blender side from build_environment first and land all changes to blender when svn libs are all updated?

Just went through the whole cmake and seems that there is many errors on newest xcode. Mostly on libraries that need updating anyway.
But for this diff, needs just to take the patch away and remove POSTFIX settings to work on macOS. Working on it, will update this diff later on.

@Omid (Khanman) Could you give your system info also. (which mac do you have?)

This might be one that was fixed already, @Daniel (danthedeckie) could you try the latest build from buildbot to be sure?

@Clément Foucault (fclem) tested, works. marking as resolved.

Most likely OpenGL issue yeah, but unable to repro here. --factory-startup --debug-gpu first to get crashlog. Nothing more to add.

This is really hackintosh problem, not blender problem. Because in the end, it's non-supported platform, closing ticket.

Now save it to other location, and upload that file (with images packed) to this ticket, thanks!

@Mike Santiago (suicvne) Could you pack images to the .blend and post it again, just to make sure it doesn't happen on while some image processing. (long shot, but still).

Unable to repro, but as I see it, there is no textures packed into .blend file so might be that we're skipping something.
Crashlog says that Threads 1 to 8 were doing same thing and 1 crashed. (IlmThread)

This is indeed Intel-only thing. I can crash blender on GPU_batch_draw_range_ex when enabling only Intel GPU on my MacBookPro.

Let's keep this incomplete until debug build crashlog.

As said on IRC, build with CMAKE_BUILD_TYPE=Debug so we get possible assert also.

Debug build gives:

Confirmed.

In D3398#79562, @Ray molenkamp (LazyDodo) wrote:

why not just update D2915?