14:55 cmarcelo: is there a way to see the actual image from a failed trace in gitlab ci? e.g. https://mesa.pages.freedesktop.org/-/mesa/-/jobs/74961702/artifacts/results/summary/results/trace@gl-intel-amly@gputest@pixmark-julia-fp64-v2.trace.html
14:56 cmarcelo: seems some images referred by the page are not found, so maybe there's a bug with copying or linking them...?
19:39 linkmauve: Hi, I’m writing a Vulkan driver, but it seems the loader overrides my objects’ VK_LOADER_DATA even if it hasn’t been initialized to 0x01CDC0DE, making me wonder why my memory got corrupted because I didn’t notice I had to do that.
19:39 linkmauve: Is that a bug in the loader, that it should reject the object as invalid if it isn’t initialized properly?
19:41 linkmauve: static inline void loader_set_dispatch(void *obj, const void *data) { *((const void **)obj) = data; }
19:41 linkmauve: It would be better to abort() if the obj isn’t defined to the right value, or at least assert() when compiled in debug mode.