Intermittent `Tex.FromCubemapEquirectangular` failure about stereokit HOT 5 OPEN

Here are some logcat entries filtered by StereoKit
17:58:53.802 StereoKit [sk_gpu] skg_tex_get_mip_contents_arr error: 1281
17:58:53.802 StereoKit 000: 0x0x77647a451c log_write
17:58:53.802 StereoKit 001: 0x0x77647a4be4 log_errf
17:58:53.802 StereoKit 002: 0x0x7764830fb4
17:58:53.802 StereoKit 003: 0x0x77647f0714
17:58:53.803 StereoKit 004: 0x0x77647ad5e4
17:58:53.803 StereoKit 005: 0x0x77647f05f0 tex_get_data_mip
17:58:53.803 StereoKit 006: 0x0x77647f0428
17:58:53.803 StereoKit 007: 0x0x77647ad5e4
17:58:53.803 StereoKit 008: 0x0x77647ef804
17:58:53.803 StereoKit 009: 0x0x77647ad0f8
17:58:53.803 StereoKit 010: 0x0x77647ad1c4
17:58:53.803 StereoKit 011: 0x0x776484ca68
17:58:53.803 StereoKit 012: 0x0x77647aad1c sk_step
17:58:53.803 StereoKit Couldn't get texture contents!

from stereokit.

I've tried the new version. The following error now fills the logcat with no other instructions between it and the application hangs:
11:04:43.641 StereoKit [sk_gpu] skg_tex_get_mip_contents_arr: eating a gl error from somewhere else: 1281
11:04:43.641 StereoKit [sk_gpu] skg_tex_get_mip_contents_arr: eating a gl error from somewhere else: 1281
11:04:43.641 StereoKit [sk_gpu] skg_tex_get_mip_contents_arr: eating a gl error from somewhere else: 1281
11:04:43.641 StereoKit [sk_gpu] skg_tex_get_mip_contents_arr: eating a gl error from somewhere else: 1281
...

Hmm. This may help me identify the origin of my issue. Will update soon,

from stereokit.

That did help move me forward and identify the original problem. I generate icons from SDF's and that has to be done on a separate thread due to the processing time. The problem was caused by some of the things I was doing on that thread.

It seems that Tex.SetColors and Material.Copy were the issue. I guess these result in GL calls and any GL calls from a separate thread risk upsetting things.

Thanks.

from stereokit.

Oh, yeah, GL doesn't handle threading well at all! Tex.SetColors should be okay though? I do have threaded tests for that. Material.Copy may currently be problematic though, I haven't reviewed that one for threading issues yet.

SK.ExecuteOnMain can sometimes be useful in these cases as well.

from stereokit.

I'd assumed it was SetColors because when I paused the code that's where it was, so it must have been something in that area that was failing and causing a tight loop. I've restructured a lot now so I'm not sure what it would have been.

Thanks for the tip on ExecuteOnMain.

from stereokit.