00:42 <alyssa> Ugh that was a stupid bug

00:51 <alyssa> Okay, I have LCRA working on all of glmark

00:53 <alyssa> Hrmmmm

00:54 <alyssa> Some shaders are spilling pretty terribly :/

00:55 <alyssa> Admittedly there is no spill cost support yet so that could be why

00:57 <alyssa> Regardless, fix regressions first

01:17 <alyssa> Oh

01:17 <alyssa> I need to actually do 16-bit RA now

01:17 <alyssa> Now's the time

01:17 <alyssa> :V

01:37 <alyssa> Hrm.

01:37 <alyssa> This shall be continued ... next time!

01:38 <alyssa> The constraint system does seem to be the way forward, though

08:14 <urjaman> sigh ... apparently valgrind has some sort of an effect on math (rounding) that causes CGAL (openscad graphics thing) to throw an assert

08:14 <urjaman> Expr: -CGAL_IA_MUL(-1.1, 10.1) != CGAL_IA_MUL(1.1, 10.1)

08:18 <urjaman> (and then it also proceeded to tell me that it is wrong rounding and that i might've forgotten to pass -frounding-math to gcc, and then SIGABRT :P)

08:23 <urjaman> ok apparently openscad folks have know about this (github issue from 2015) for like 4 years but nobody has fixed valgrind to support the rounding modes in the mean time? -.-

08:24 <urjaman> i suppose i'll have to build openscad next (with that check disabled, as they hinted in the issue..)

08:29 <urjaman> (or try to fix valgrind... i mean that sounds harder but also actually more useful, but also like how far am i going to go to chase down these things...)

12:03 <alyssa> Le sigh

22:39 <urjaman> just disabled the check -- i did find a valgrind bug originally from 2006 about the rounding modes lol -- but anyways...

22:39 <urjaman> https://urja.dev/valgrind_openscad.txt this is the rather very verbose output valgrind/memcheck gave me when i closed openscad

22:41 <urjaman> i'll need to check code more before i can say anything about that ... except that it doesnt look like it has anything really to do with occlusion queries