OpenMW 0.45.0

[Capostrophic]

For instances in interior cells their interior cell is shown. For instances in exterior cells nothing is shown but a blue background. That is pretty confusing.

[Zini]

I can not reproduce this issue. Is it all exterior cells? Or only specific ones?

[Zini]

However what I do see is a bunch of bad_alloc exceptions that seem to be related to body parts.

Everybody, hold their horses. We may have another regression here.

[Capostrophic]

Is it all exterior cells?

Any cell that has #N N ID format in the Instances table. I see this and nothing more.

Screenshot_20181030_142221.png (94.41 KiB) Viewed 3207 times

Edit: yes, Instances table specifically, Cells table is fine, or I would have noticed this when I was trying to fix instance dragging.

On an unrelated note, leveled lists can be "previewed" since they're technically normal objects, but they don't have models so you also see an empty scene - with a error about the non-existent model logged.

Is there any specific way to catch the body part errors?

[Zini]

Edit: yes, Instances table specifically, Cells table is fine, or I would have noticed this when I was trying to fix instance dragging.

Okay, now you are making sense. I can reproduce that. Note that the title of the scene subview is "Scene: sys:.default (empty)", meaning it does not show any cells. This of course will result in the scene being empty. A new bug, it seems.

Is there any specific way to catch the body part errors?

You mean like getting a stack backtrace?

gdb ./openmw-cs
catch throw
run

[Ace (SWE)]

Windows RCs;

32-bit
64-bit

[Zini]

Stop! We still have a regression!

[Capostrophic]

Which I still don't understand how to reproduce.

[Ace (SWE)]

Stop! We still have a regression!

Sorry, my bad.
Been building these basically offline due to net issues, missed the posts about the regression.

I'll do a round of proper RC1's (RC1.5s?) once the regression's been found.

[Capostrophic]

Since psi29a is going to be excessively merge-happy these days, what is decided about the "contributor adds the changelog entry" policy of 0.45.0?

IMO changelog micromanaging was extremely messy during these 5 months and if returning to the "legacy" changelog managing style is considered personally I vote for that.