IRC channel logs

2025-08-17.log

back to list of logs

<matrix_bridge><Andrius Štikonas> Jeremiah Orians, gtker : so I've fixed test0004 now in M2-Mesoplanet https://github.com/oriansj/M2-Mesoplanet/pull/21
<matrix_bridge><Andrius Štikonas> perhaps a bit slow since we need to read .o file again but it seems to work
<stikonas>ok, M2-Mesoplanet 1.13.0 is now uploaded: https://github.com/oriansj/M2-Mesoplanet/releases/tag/Release_1.13.0
<stikonas>I think just need to make combined stage0-posix release now...
<stikonas>and it's up here: https://github.com/oriansj/stage0-posix/releases/tag/Release_1.9.0
<stikonas>janneke: we'll need to get newly used defines into mes though...
<janneke>stikonas: great, i just got a 0.27.1 ready that should support 1.12.0 :)
<stikonas>ok, that's good
<stikonas>though 1.13 needs some more defines
<janneke>there's also a patch from you with 1.13.0 defines; but i haven't checked if that's enough to support 1.13
<stikonas>0.27.1 is already out?
<janneke>nope, but almost -- running tests
<stikonas>hmm, not sure if they are all there, let me check
<stikonas>oh I see
<stikonas>do you have commit pushed somewhere?
<stikonas>with latest defines that are already in mes
<stikonas>then I'll check diff between that and M2libc
<janneke>stikonas: /me just pushes staging and version-0.27 branches
<stikonas>ok, I'll check those a bit later today
<stikonas>I suspect we'll need more patches
<stikonas>I guess I'll send a bit new patch on top of 0.27.1
<stikonas>and you can then ignore the patch that I already sent
<janneke>stikonas: that's great
<janneke>i wanted to get 0.27.1 out soon because it doesn't build with gcc-14 and recent nyacc's
<stikonas>yeah, that's fine
<stikonas>you can always make a mini 0.27.2 that just adds support for new M2-Planet/stage0-posix
<stikonas>janneke: by the way, while you are here, how much contact do you have with nyacc developers?
<stikonas>there is this commit on top of v1.00.2 https://github.com/Googulator/nyacc/commit/c301626b5883378f07be68819d88cd4c4897876c that would be nice to upstream if nyacc upstream is willing to take it
<stikonas>ok, sent a mail to bootstrappable mailing list now about stage0-posix 1.9.0
<janneke>i usually mail to Matt Wette <matt.wette@gmail.com>, cc'ing guile-user@gnu.org
<janneke>i'm wondering about the string-every and vector-fold bits
<janneke>why not add them to mes, hmm and we already have vector-fold even
<stikonas>well, as long as they can be done without psyntax.pp
<stikonas>the idea of this patch was to rebuild nyacc pre-generated files with just mes
<janneke>yeah, that's pretty cool
<janneke>F_OK is on the wip-gcc4 branch (which has been waiting for far too long...)
<janneke>and should be rebased some time soon
<stikonas>oh nice, so this patch can be made a bit smaller then
<ekaitz>janneke: what's gcc4 branch? building GCC directly instead of tcc?
<janneke>ekaitz: nope, it's for building gcc4-core without glibc/musl
<janneke>but it's not there yet :(
<ekaitz>well, good things take time
<janneke>hehe
<janneke>i've been bisecting musl function implementations but there is a whole cluster of dependent functions that cannot be bisected further
<janneke>my guess would be the struct FILE* / filedes casting hack, but dunno
<janneke>iwbn to rewrite that, ungetc and buffering come to mind, but that's quite a bit of work and no guarantee of success
<ekaitz>uh yeah
<ekaitz>the FILE* thing is practical but problematic
<stikonas>ok, just tried to diff M2libc and mes M1 defines, this is what I got
<stikonas> https://paste.debian.net/1392095/
<stikonas>mosty additions except for -cmp +cmp_ebx,eax
<stikonas>but I guess in mes we want to keep cmp too
<stikonas>to keep compatibility with older M2-Planet
<stikonas>hmm, actually mes-0.27 doesn't build with new M2-Planet
<stikonas>include/signal.h:190:newline expected at end of macro directive
<stikonas>Might be M2-Planet bug...
<stikonas>so we might need to re-spin...
<stikonas>so it's not happy with:
<stikonas>#endif // !__M2__
<stikonas>#endif //! SYSTEM_LIBC
<stikonas>but adding a newline between these 2 #endif works around it
<janneke>stikonas: thanks, pushed some more stuff to wip
<janneke>also, i found a nitpick problem with 0.27.1, so i can probably include these
<janneke>but that will have to wait for a new round of testing
<stikonas>janneke: you probably want to keep old cmp define too
<stikonas>i.e. DEFINE cmp 39C3
<stikonas>otherwise the older M2-Planet 1.12 won't build
<stikonas>janneke: I've now identified which commit broke M2-Planet -> mes build (well, at least the current problem, there might be more...)
<stikonas>it's the one that removes support for // CONSTANT
<stikonas> https://github.com/oriansj/M2-Planet/commit/ca0eda58fabf252f272e7099c5fe74102b5cc46b
<stikonas>anyway, let's hope that's the only problem...
<stikonas>yeah, I think that's the only problem
<stikonas>with that commit reverted, M2-Planet compiles mes sources into mes.M1
<stikonas>so I think it will be a respin of M2-Planet
<stikonas>and then it will all work with M2-Planet 1.13.1
<stikonas>though I still recommend to add "DEFINE cmp 39C3" to x86 defines
<stikonas>ok, fix now commited to M2-Planet
<stikonas>and released as 1.13.1