@Fabio Prinzis (SonicBlue) @Craig Dolphin (craigdolphin) the issue happens right away with a debug build with asan (memory address sanitizer). In a release (final user) build it won't necessarily crash right away, but it will lead to impredictable behaviour and apparent "random" crashes. There is a chance the original reported issue is something else, but either way this has to be solved first before re-testing vertex slide itself.

All good then, closing now.

@Sergey Sharybin (sergey) wants to look at this one? It has the one-line-repro the way you like it ;)

I can get a crash the moment I open your file:

@Gabby (gabdar) your blender_pythonOutput.log says a lot. I trust you are using the latest RC download build for Mac, right? (we had it uploaded twice I believe)
@Sergey Sharybin (sergey) did you try with the official RC1 dmg?

@Brecht Van Lommel (brecht) / @Sergey Sharybin (sergey) is this by design?

I have it working even with the RC1, but anwyays, problem is gone

In T66861#720735, @dono dono (dono) wrote:

Thank you Dalai and Robert. Indeed it is difficult for me to know if it is a bug or not. Thanks to you and all the devs for this amazing 2.8 and all your fantastic work on it! All the best from France.

In T66861#720734, @Robert Guetzkow (rjg) wrote:

@Dalai Felinto (dfelinto) I agree this is not a bug, but imho bad UX. Seems like something minor that could be done for 2.81 and added to the UI paper cuts.

What I'm trying to say is, this is a design issue, not a bug per-se. We are open for patches and design proposals, but just pointing to a (known) weak design doesn't fit in with the other bug reports (it is no one's fault really, it is often hard to tell when something is a bug or not). That said I appreciate the attention by @dono dono (dono) on trying to spot these issues in Blender.

The buttons set the new default "mode" for new typed text. Like it or not that is the current expected behaviour (not a big fan either).
We could remove the buttons from object mode, that would at least make things less misleading.

I cannot reproduce in Linux. Did you run factory startup? The Add custom property button should be in the object panel.
Can you share a screenshot of the expected place for the Add property?

It is working as expected. To have them change the currently selected text in edit mode use the shortcuts (Ctrl+B, Ctrl+I, Ctrl+U).

For the records, if I change the view angle a bit this works.

Not enough information to reproduce the issue (no Blender version, no .blend file, ...).
My guess is that you went to the Compositing workspace and didn't click on "Use Nodes".

Fixed.

Confirmed indeed.

Please follow the bug report guidelines. We need a sample .blend and clear steps on how to reproduce the issue.

It is probably a GPU problem: https://code.blender.org/2019/04/supported-gpus-in-blender-2-80/
That said we need more information, like version of blender and crash log.

I'm not sure we actually want truncate to work for any alignment that is not top/top-baseline. That said D5246 fix the most common cases.

Let's now poke our fanciest Mac dev - @Sergey Sharybin (sergey)

I cannot reproduce here (on Linux). Could you run Blender from command-line and see if there is any errors there?

Hi @jeremy joe (blenderjoe) please provide a simple .blend file with specific instructions on how to reproduce the issue. Without that there is not much we can do.

It is indeed strange that we call BKE_base_eval_flags instead of BKE_object_eval_eval_base_flags. That said calling DEG_id_tag_update((ID *)&scene->id, ID_RECALC_BASE_FLAGS); is not enough so I would say the patch should be good as it is.

Basically layer_collection_sync calls BKE_base_eval_flags right away while iterating over the bases.
However when a parent/sibling collection is to influence the collection flag of an object that exists in more than one collection, it is too late since we deselect the object in BKE_base_eval_flags right away.

This is a legit bug, I will look at it.

It was not my intention to have a different logic for tool and non-tool. The way I see it they would work the exactly same way.

@Ingmar Franz (Uranus) cage baking was fixed recently, and from my tests the edgesplit modifier is effectively being removed.
From your file the baking result is still a bit different when enabling/disabling the sharp edge setting but that is because this affects the low poly mesh (though not its cage).

@Bastien Montagne (mont29) weren't library overrides supposed to be disabled for now?

Strangely confirmed with an old file I had around:

Please attach a sample file

Are you using a tablet or a mouse for the drawing? Which model/driver version?

Please attach a simple (small) PSD file here. Not many developers have a way of creating the faulty file.
Also can you test if the issue also happens in 2.79?

Germans, unite: @Jacques Lucke (JacquesLucke) @Sebastian Koenig (sebastian_k) could you confirm this?
In theory the shortcut is hardcoded to the key position under the Esc in the top left corner of the keyboard. But perhaps it is not working in some systems.

Could you try again? I just tried again here and it worked fine. Maybe it was a temporary downtime in the server.
That said that button should indeed bring you to the developer.blender.org website.

Patch on D5210.

@Campbell Barton (campbellbarton) by the way, CTX_wm_window_set() (introduced on rBd9b676852102) seems to have broken BKE_reports. This patch fixes only the crash, but not getting a BKE_report makes things too enigmatic.

Attached T66484 with a proposal to handle base point.

@MACHIN3 (MACHIN3) can you test with 2.79? To determine if this was introduced in 2.80 or was ever there? (the best sample files in this case are the ones that were created in 2.79 in fact).

The report lacks the steps to reproduce the issue from scratch. What I see from the file is that you have objects that are not linked to any of the scene collections.

I have an old addon (which is outdated) that was handling text selection with no problems:
https://developer.blender.org/diffusion/BAC/browse/master/text_editor_hastebin.py

@Marco Alamia (Macro) from what I can read, the last comment by @Brecht Van Lommel (brecht) wasn't addressed yet (code is still bisecting).

All good now, thanks Sergey!

@Campbell Barton (campbellbarton) I originally noted this on https://developer.blender.org/T66126#706986 but since that bug was fixed (and this one was not), it is reported here now.

@Sergey Sharybin (sergey): me_cage should have the same totloop and totpoly as me_low. We even check for that when users set the cage manually.

We have the same issue in 2.79 it seems.
The fix is easy though, and I think we should tackle it both ways (outliner to take instancing type into consideration, and make instance real to set group/collection pointer to null).

So what is happening here is: