<fossy>stikonas: oh, oops, missed that <pder>no worries, I just needed to insert some stuff into the README and thought it might start to get confusing at merge time <siraben>Ben Lynn has pushed some commits to blynn-compiler <siraben>probably will cause conflicts when we rebase <siraben>He indicates that precisely.hs has #include support ***ChanServ sets mode: +o rekado_
***rekado_ is now known as rekado
<fossy>stikonas: i'm just finishing testing the checksums for the tcc one <fossy>then the ls one will be easy <fossy>i love how >50% of the "Parts" take <5% of the time <stikonas>well, we can temporarily comment out blynn-compiler <stikonas>I can try to find at least first cp checksum in the meantime <stikonas>that will not be affected by musl changes <stikonas>fossy: also, speaking of speed, I think logging slows down CI significantly <stikonas>so maybe we need to start thinking about log files once we have bash <stikonas>in CI, those parts after tcc takes just as long as mes <stikonas>fossy: so pass1 checksum for cp is "5a869b028ce076d0a2ee06721956f38128a7d82d68be61e223a3051fbbb59e90 /after/bin/cp" <OriansJ>siraben: now that is exciting news. First glance seems to indicate it shouldn't conflict too much when we rebase <stikonas>fossy: you'll now have to update tcc-musl-pass2.sh when you merge cp PR <stikonas>but should be just trivial copy-paste... <OriansJ>now to try to check to see if PowerPC big endian distros behave differently than little endian distros <OriansJ>or if the MSR flag is process set via the ELF header and the resulting binary behavior is adjusted accordingly <OriansJ>and if it changes the byte order of the instructions or just the data. <OriansJ>and mescc-tools-seed is now officially stage0-posix <pder>Thanks fossy, for merging and finishing up the musl rebuild. I had done something very similar but got hung up on the checksums and the README numbering <stikonas>next one is perl 5.6.2 (can't jump to 5.8.9 directly) <stikonas>hopefully after that we can switch to microperl (so that I don't need to tweak config.h and handwritten Makefile) <stikonas>hmm, perl 5.6 is proving to be difficult... <stikonas>it might be that we'll have to stay with perl5.005_03 fo now <pder>Thanks stikonas for updating the README. I closed that pull request <stikonas>fixed some perl 5.5 checksums now, so hopefully PR #44 will pass now <pder>Maybe we should have a way to easily disable checksum verification during development so it doesnt kill the build <stikonas>maybe better option would be to update checksum files... <stikonas>although, they'll still be in sysa/tmp, but at least you can compy them back to sysa <stikonas>pder: so for perl, I think we'll have to live with 5.5 for now <stikonas>5.6 doesn't build some a couple of reasons (one might be wrong bison version) <stikonas>I couldn't regenerate 5.8 header files though <stikonas>so unless we patch it out, looks like we need to regenerate ./Configure first <stikonas>pder: so for now we can target autoconf 2.50 and automake 1.5 <gforce_d11977>pder: stikonas: fossy: yes, for now the checksumming should always return 0, but printout in LOG <gforce_d11977>fossy: in CI you can just grep the LOG for these err-messages, so we can still check it <stikonas>gforce_d11977, pder, fossy: oh with some patches I think I am able to jump from perl 5_005_03 to microperl of perl 5.8.9 (possibly newer perl's will work too) <stikonas>argh, but bison is still broken in 5.8.9... <Hagfish>it sounds like a game of minesweeper or something, but i'm really hopeful that the issues get easier with later versions of everything <stikonas>so I think perl 5.6 allows jumping directly to perl 5.18 <fossy>gforce_d11977: i firmly disagree <fossy>there is no point in having checksumming if we dont fail if it fails <fossy>the entire point of the checksumming is to ensure your hardware matches what is expected and is not compromised, and that the build is reproducible <fossy>i'll do that this arvo first then cp <stikonas>you can still add sha2 description later... <stikonas>so now readme and kaem.run/run.sh numbers are back in sync <stikonas>anyway, this perl 5.5 is not too bad already, it allows us to run up to automake 1.5 and autoconf 2.50 <stikonas>hopefully sufficient for most of our purposes *stikonas had to check what "arvo" means. That's Australian... <fossy>sometimes my australian slang creeps into my online vocab, lol <stikonas>fossy: I guess I can skip checksumming if package installs only text files? <stikonas>(later once we have automake we can build it with autotools) <fossy>anything just cpd from the tarball dosent need checksumming <stikonas>but for now I just ran sed to replace a few placeholders <stikonas>hmm, I still have to create an empty checksums file... <pder>fossy: for the checksuming, I think there should at least be an easy way to make it not a hard failure, and just log checksum errors during development <pder>anything merged to master obviously has to pass all the checksums. Its just annoying to have a 10+ minute build die in the middle because of something expected