IRC channel logs

2025-07-27.log

back to list of logs

<oriansj>gtker: well M2-Mesoplanet -E should output approximately the exact same thing as gcc -E ; so you might be able to leverage that to find issues with M2-Mesoplanet's output.
<stikonas>oriansj: well, the main issue is now that with cat change (which fixes builder-hex0) we can't pick libc-full.M1
<stikonas>hmm, perhaps we could restrict that feature only to the cases when libc-full is used
<stikonas>and use libc-full unconditionally
<matrix_bridge><gtker> stikonas: We can't use libc-full unconditionally because it depends on some things from the C files being there in order to initialize some globals, at least as far as I remember
<oriansj>well the only difference between the full and minimal is if stdio.h is used
<stikonas>gtker: oh I see
<stikonas>oriansj: yes, but we need some logic to determine that
<stikonas> https://github.com/oriansj/M2-Mesoplanet/commit/1389e4ead3b888615327937dc74c6d6323bcc1cc#diff-1fffe4cbc72425230e08ab39a5b191a9529fea4e89807c96a130f27e16ed92c2 removed that
<stikonas>see comment: /* Need to know if stdio has been used to choose correct libc. */
<stikonas>but maybe "reimplementing" catm inside M2-Mesoplanet is not too bad
<stikonas>we already coppied kaem spawning logic into M2-mesoplanet
<oriansj>or just make full the default and have a flag for changing it to something else
<stikonas>gtker: can we not unconditionally include stdio.h?
<stikonas>though it's not completely standard...
<matrix_bridge><gtker> I guess we could, although it will bloat up binaries that don't use stdio.h
<oriansj>or just track what files are being cat'd and see if the name of any of them match stdio.h
<stikonas>(perhaps not the whole stdio.h then but only the parts needed by libc-full)
<stikonas>yeah, that's another option