15:26 PaulePanter: Hi. Any ideas regarding https://github.com/rstudio/rstudio/issues/3781 ?
15:26 PaulePanter: One of our users hit the same issue.
15:27 PaulePanter: https://paste.debian.net/1097605/
15:27 PaulePanter: Linux 4.19.57
15:58 PaulePanter: https://bugreports.qt.io/browse/QTBUG-41242 seems to be the Qt bug report.
16:03 PaulePanter: Mesa 19.1.1
16:42 RSpliet: PaulePanter: "Note btw, that in 5.13, even WebGL should work as we there have a non-threaded OpenGL mode." is mentioned as a solution. So this is just another manifestation of nouveau being broken in multithreaded contexts. One of the devs has a 95% solution to this problem, but as with a lot of GPU stuff the devil is in the details
16:44 RSpliet: I'm not sure whether his solution in its current state would introduce any regressions as a result of one of those details biting him in the back. Until regressions are demonstrably ruled out, it's not merge-ready.
16:45 PaulePanter: RSpliet: Thanks. Are those patches available for testing?
16:46 RSpliet: Not at the moment. The dev in question is currently absent in this channel, I don't want to risk bugging him on his free day or sth ;-)
16:47 RSpliet: But last time I checked, he reported promising results and known issues.
16:47 RSpliet: bbl
16:49 PaulePanter: Thanks.
20:19 danvet:still looking for a tester with pre-nv50 for https://patchwork.freedesktop.org/series/65489/
20:19 danvet: skeggsb, machine still busted?
23:11 ReinUsesLisp: hi, F2F (and probably the others) have rounding operations and "half" selection operations
23:11 ReinUsesLisp: e.g. F2F.F16.F16 Rd, Ra.H1 ;
23:11 ReinUsesLisp: and F2F.F16.F16.RM Rd, Ra ;
23:15 ReinUsesLisp: I wanted to add these but I'm not sure where h1 should go in envydis declarations