In D5287#225467, @sera wrote:

CPU skinned? Is this what is commonly known as gpu skinning but the work done on the cpu?

Yes.

git diff instead of svn.

In rP28109#62140, @Dunedan wrote:

As the pot-files didn't change due to this commit, neither did the strings in Transifex change. In Transifex the string looked fine before the commit as well.

Yeah, I meant the strings before the commit. If they look the same there shouldn't they have the same problem in the game?

In rP28109#62134, @Dunedan wrote:

That's the only known location where it was broken. I can imagine that it only happens when using markForTranslation() as that's the only usage of markForTranslation() which contained a non-breaking space.

I just opened transifex and I see strings from markForTranslation and translate in the same way (both with highlighted nbsp). So can the issue happen for other usages (inside translate or xml)?

In D5287#225409, @phosit wrote:

I can't apply this diff.

Seems like svn can’t handle renaming.

In rP28109#62119, @vladislavbelov wrote:

@Dunedan What's the point of the change only for the tutorial map?

@Dunedan What's the point of the change only for the tutorial map?

After its commit https://trac.wildfiregames.com/wiki/PortA26ToA27 should be updated accordingly.

In D5281#225155, @phosit wrote:

GetUsage is mostly used with IBuffer::Usage::DYNAMIC. Maybe the IsDynamic should be kept to deduplicated thous. That would make this diff a lot smaller.

I was thinking about that but didn't decide.

In D3807#225134, @phosit wrote:

I'm not sure what you want to ask. A naive answer: the continuation is executed _after_ the promise settles.

So we should guarantee that RunJobs will be called at the right moment and from a proper thread.

In D5276#225132, @phosit wrote:

Yes a27 isn't released yet...
Or do you think thous functions should be removed in a27?

In D5281#225145, @Stan wrote:

Maybe you could add a new type like storage_usage_t or something.

A more narrow type is preferred indeed. But it requires additional steps (mentioned in the inline comment).

In D3807#225078, @phosit wrote:

It's not possible to run promise-continuation concurrently. (promises as such don't "run")

But this should be commited afther a27 is released to let modders addopt.

Is that still relevant?

So it's possible to use promises in the simulation code. Is it possible if they're running concurrently?

How the diff avoids using promises in the simulation code?

I'll commit it soon.

In D5244#223224, @Stan wrote:

So what happens here exactly when it cannot find the shader file?

This functions are deprecated for over a week now. I think it's time to remove thous functions XD

It's better to mention related changes explicitly.

In D5257#223695, @Stan wrote:

Any reason the renderer and the postproc manager would have different sizes ?

Yeah, because we have multiple render targets with different resolutions depending on a pass: window resolution (UI and post effects), scene resolution (scaled by postprocmanager), reflection/refraction resolution, shadow map resolution.

The patch isn't related to the OOS reason. Building a unit preview is just a trigger for the OOS. So this change does seem like a cleanup after the previous one. But maybe we want to support building units, then we must have elexis's patch (I'd prefer to commit it anyway, maybe after some changes/tests).

In D5242#223156, @sera wrote:

You could render in a different window and copy back the frame as well as forward the events. Worth it? Probably not.

Also might be a performance loss.

In D5244#223158, @sera wrote:

I suggest to add a shaders not found error message and maybe give a hint as for the possible cause and solution, otherwise LGTM apart from the obvious linter warning.

In D5242#223121, @sera wrote:

About atlas, wx itself uses gl but that doesn't mean you can't use vulkan for the viewport, anyway mostly a non issue and certainly low prio.

Vulkan can't access the viewport without interacting with wx.

Is it possible to make a test?

LGTM.

@Papnaas you can abandon the patch, for that you need to choose Add Action > Revision Action > Abandon Revision.

LGTM.

unsigned int doesn't fit static assertions (see the diff build results):

1>source\network\fsm.h(173): error C2607: static assertion failed (compiling source file ..\..\..\source\network\FSM.cpp)
1>source\network\fsm.h(173): error C2607: static assertion failed (compiling source file ..\..\..\source\network\NetClient.cpp)
1>source\network\fsm.h(173): error C2607: static assertion failed (compiling source file ..\..\..\source\network\NetClientTurnManager.cpp)
1>source\network\fsm.h(173): error C2607: static assertion failed (compiling source file ..\..\..\source\network\NetSession.cpp)
1>source\network\fsm.h(173): error C2607: static assertion failed (compiling source file ..\..\..\source\network\NetFileTransfer.cpp)
1>source\network\fsm.h(173): error C2607: static assertion failed (compiling source file ..\..\..\source\network\NetServer.cpp)
1>source\network\fsm.h(173): error C2607: static assertion failed (compiling source file ..\..\..\source\network\NetServerTurnManager.cpp)

elexis found the problem (I'll try to make a patch): this.phenotype = pickRandom(this.GetPossiblePhenotypes()); in Identity. So the current patch is needed only if it's useful for the gameplay.

We're trying to avoid patching third parties. Ideally a PR should be created in the third party repo and then our library might be updated.

@phosit LGTM, you might commit it.

In D5231#222734, @wowgetoffyourcellphone wrote:

@vladislavbelov does it address your concerns now?

The texture seems ok to me. Maybe I'd rename it to hud_panel (because it's reused for all panels), but it's optional.

In D5206#222595, @real_tabasco_sauce wrote:

So, those issues being: graphic error upon click, errors for training while reforms tech completes (fixed with current diff), and the OOS?
I really don't know what would cause this graphics error, particularly when something is clicked. @nifa made was responsible for some changes to the onager and particularly for designing the packed onager. I would think it is because of a missing spirv shader, but that fact it only happens upon click makes me doubt that.

see https://trac.wildfiregames.com/ticket/6888#

I'd like to note that #6888 contains 3 different issues (4 maybe). OOS is unrelated to the current diff.

Combines the textures of all 3 panels into one texture sheet (per @vladislavbelov 's ancient request).

It's best when textures are reused or tightly packed. According to the preview it has much more of redundant space. Is it possible to reuse the same texture for all three panels?

Fixes notes.

Fixes notes for options by @Stan and @sera.

In D5218#222359, @phosit wrote:

Likely we mean something different. I suggested a change for the two places I thought about.

I thought you're talking about all shaders code :)

In D5218#222345, @phosit wrote:

There is still a lot duplication.

I agree and I'd like to fix it. But unfortunately I don't have time for everything, first of all I'd like to get some performance for the next release.