José Fonseca

I finally looked into this. The frames seem to be rendering, but most are renderered super dark. I didn't spot any smoking gun. Maybe the game is messing with the...

Thanks for the additional info. So if I understood correctly, rendering with DXVK was also bad as seen when replaying with apitrace. If the game, is not fiddling with screen...

> That hack was removed two months ago [doitsujin/dxvk#1982](https://github.com/doitsujin/dxvk/pull/1982) IIUC, the `d3d9.apitraceMode` has a similar effect, ie, even though it doesn't excplitly ignore `D3DLOCK_READONLY`, writes done to buffers are visible...

This is already possible on Linux. But it's not the default. it's not yet implemented on Windows (or macOS). I agree it would be useful to though.

It's not easy to do this sort of stuff (polymorphism) just from the spec, but it can be done. See https://github.com/apitrace/apitrace/blob/15de87464417ffb072fea4f7f0ea2cec119cc3fc/specs/gltypes.py#L305 for example, on how the index buffers are handled....

That's worrying. This certainly looks like it should _not_ be automatically false. libbacktrace is a thirdparty lib. Perhaps it was fixed upstream. Please confirm what's FreeBSD compiler and version.

This is a known issue and a long standing tech debt...

This might be related to issue #409. I think the problem is that `DYLD_LIBRARY_PATH` mechanism is not very reliable on Mac OS X, particularly with Java apps. I've started prototyping...

No, you need to checkout the `dyld-interpose` branch, build, then run `apitrace trace ...` as usual. It will setup `DYLD_INTERPOSE` correctly for you.

That's odd. No idea what could be happening. Are you sure that app is really loading OpenGL? I tried to repro this with lwjgl-2.9.3 but it just crashes with macOS...