16:07:08drmeisterHe confirms we should upgrade to 2.0.7
16:07:59Bike"I'm going to try removing the -flto-thin argument and try linking again" "Warning: execute-link-fasl worked after removing -flto from the argument list --- FIGURE OUT WHAT IS GOING WRONG WITH THAT ARGUMENT!!!"
17:59:30karloszit builds fine when i have Bikes version of sicl
17:59:39karloszbut when i swap it out with my own it gives that assertion
18:00:04karloszBike: how did you test the pull request i sent?
18:01:03BikeOh, I might have had to edit tools_for_build/cleavir_file_list.py
18:01:45Bikedrmeister: i made some changes, and now building asdf and serve event seems to hang, but if i do it through the workbench there's no problem. do you know what the difference could be?
19:42:12Bikegathers a lot of rest arguments, but i didn't change anything there
19:43:20Bikei can load cleavir in bclasp, and functions compiled with it seem to work, even if they do weird things like have &optional and &key that i didn't think about before
19:49:21Bikei'll just destroy things a bit at a time. like a neurosurgeon. no problem
19:53:17karloszinteresting, the commit that upgrades to llvm6 builds successfully and gets past linking on arch
19:53:31karloszso there must be something in between that breaks
20:35:13karloszBike: do you have a more full backtrace for the unitialized data problem you ran into with the closure conversion fix? i couldnt succeed getting the latest clasp to build
20:36:02Bikeno. i can get you one if you want it. a backtrace might nto be sufficient since it's (probably) a miscompile, though.
20:37:10karloszyeah, some kind of backtrace or at least vague problem area/place would help
20:41:46Bikei pretty much just took my regular configuration and changed the sicl to your branch.
20:42:55karloszokay, sounds good. i also have partial inlining and closure conversion as the first two steps in my passes, so it should help narrow down where the possible problems could be from, considering it builds for me