Hello, thanks for answering. It works perfectly on pc, but I would like it to work on the chrome web browser with android, any ideas? Thank you. Using google translator.
Good morning Robert,
Any way to work on the mobile with chrome?
the only way i can get it to reliably work is like this , but it feels like a cop-out.
More than a week passed without a reply.
Still have the issue with unlinking fonts - incorrect context.
Load BLF fonts when required rather than when selected.
The second example crashes if you paint in some locations. It works sometimes because mirror modifier may correctly reference original vertices.
detecting the colorband is easy, that is working as is. The problem is for the button enable/disable poll UI_context_active_but_get returns NULL. now i can just check if it's null and return 1, but when i do that the button kinda randomly enables/disables it self..
The poll could be modified to detect this particular button.
Same results now. Thank you Sergey!
Fml. I just updated my CPUs Intel Graphics Driver. https://downloadcenter.intel.com/
Eventhough my Windows was seemingly up to date, this did the trick.
Sry for the noise.
Rats, I'm not able to figure out if i'm being polled for the button enable disable , or because the mouse is over a ui element, (even with a hacky way, the button still turns into a light show of randomly enabling/disabling) only way i see to sidestep this is to add yet another operator just for the button? Any guidance you have here would be appreciated.
Thanks very much for your answer Erick, I just open a thread on blender artists. https://blenderartists.org/forum/showthread.php?434552-Parenting-with-automatic-weight-very-slow
Since this commit the eye-dropper button above the color-band no longer works.
Would rather have this as a pass for the existing basic engine.
No bug here i bind the first mesh in less than 2 min and the second in less the 10 min, i was not checking my watch, those are more how i felt ;)
Another file if someone wants to test it, this time the armature is correctly lined with the body (some result, took forever for parenting with automatic weight) :
Confirmed on Arch Linux with a build from today (hash: 4838512e7d4)
According to 533544117bf3 searching on user ID should work, but it doesn't in /u/ (even after reindexing users).
Actually, this bug should have been marked as resolved, unless there are other cases when there is a difference :)
An IMO important thing to improve is to reduce the usage of dictionaries. In certain places, for example in the prepare_user_data() function, a dict is used where a class would be more suitable. This will immediately resolve many ambiguities; there are too many things that are now called "user":
It is strange, it fails the unit tests because of different noise patterns, but appears to converge to the same result. I'll do some more investigating to find out where things take a different path.
For some reason a previous commit overwrote this patch and closed it, finally managed to force the diff back into Phabricator using arc diff --browse --update D2950.
Uploading file here:
Minor style issue.
Everything said in the mentioned task, also applies to 2.8.