IRC channel logs

2022-10-23.log

back to list of logs

<mihi>It also seems that tar-1.24 package does not work (or not properly override older tar package) when placed in repo-preseeded. As I get "unrecognized option `--no-same-owner'", which resolves itself when removing that one package.
<mihi>Anyway, I guess I can announce <https://github.com/schierlm/undead-bootstrap/> here now (~1 week early)
<mihi>logs for the failing run when tar package is present: <https://github.com/schierlm/undead-bootstrap/actions/runs/3304698834>
<mihi>but when having everything in packages except heirloom-devtools and tar, CI bootstrap takes only 5 minutes, which is acceptable speed for me :)
<fossy>stikonas[m]: preseed is slightly broken, but unrelated to sources changes i believe
<stikonas>fossy: hmm, not good :(. But hopefully not too hard to fix
<fossy>no, trivial fix
<fossy>i have a commit ready, i'll push it in a moment
<stikonas>sure, push it...
<stikonas>we are getting quite a few different modes, hard to test everything...
<fossy>sorry for not seeing zlib PR. math final was occupying my time right around there
<stikonas>earlier this week I fixed --external-sources mode...
<fossy>yeah. hoping CI will be ready soonish
<fossy>i saw that
<stikonas>no problem, math final is more important
<fossy>gotta say, recently i've noticed how far python has come since even the early days of py3
<fossy>older python has so much weird behaviour
<stikonas>oh, what in particular?
<stikonas>and I'm now back home, so hopefully can finish stage0-uefi soonish
<fossy>searching for python stdlib is quite strange. it first checks if it's in the build directory, then checks where it might expect to be in terms of lib/pythonx.y, except *from the current directory*, then it goes to predefined paths (prefix and exec_prefix given as configure options)
<fossy>setting PYTHONHOME breaks building python & makes it unable to load dynamic libraries
<fossy>and lib-dynload/ subdirectory is not covered in all cases
<stikonas>hmm, somehow building python sounds more complicated than I initially hoped for...
<fossy>it is a little complicated, but not overly so (not as bad as eg autogen). it's just that there's strange behaviours that are hard to debug
<stikonas>uefi stuff is also hard to debug...
<stikonas>I didn't set up gdb (though I think it might be possible to do that)
<stikonas>and without gdb it's a bit tricky
<fossy>as in for stack dumps or stepping through?
<stikonas>I often relied on just using exit status to print debug info
<stikonas>stepping through
<fossy>i see
<stikonas>most of it is sorted now
<oriansj>mihi: nice