IRC channel logs

<Googulator>given the nebulous release date of 0.26.1 (which may or may not include the patch for regenerating Nyacc's files), we might need a 0.25-lb1 or 0.26-lb1

<stikonas>Googulator: no, there isn't

<stikonas>and I'm not evne aware of some private patch

<stikonas>well, 0.26 or 0.26.1 makes it easier to run guile modules there

<stikonas>but I don't think it solves the problem anyway

<stikonas>because the current guile method was depending on ppsyntax file

<stikonas>(which would be pregenerated file in mes)

<stikonas>otherwise unused in mescc and we delete it

<Googulator>seems like it was Timothy Sample who achieved... something... without psyntax-pp: https://github.com/schierlm/guile-psyntax-bootstrapping/issues/1#issuecomment-1899297115

<Googulator>and I guess the issues he ran into are fixed in 0.26, hence the reference to 0.26 being a requirement for properly bootstrapping Nyacc

<stikonas>oh, yeah, now I seem to remember this comment

<stikonas>anyway, I don't know any more than in this ticket

<Googulator>he did release https://gist.github.com/samplet/78a2e92d47dd37b17886b26dffa14ffd - but I'm not sure if this + double-execution fix on top of 0.26 is sufficient

<Googulator>if it is, I'm inclined to roll a 0.26-lb1 release

<stikonas>well, ask him I guess...

<stikonas>and you can ask jannekefor 0.26.1

<Googulator>syntax-rules doesn't depend on psyntax-pp, right?

<Googulator>only syntax-case

<stikonas>no idea :(

<stikonas>it's scheme

<stikonas>but most likely you are right

<oriansj>yeah the mes.c to tcc step is kinda mesccy but nothing should be private in live-bootstrap

<oriansj>and this https://the-dam.org/docs/explanations/Plan9ListenOnLinux.html gave me false hope; because seeing guix on plan 9 would have been a sweet port

<pabs3>sam_: personally I would like to see autotools move towards generating multiple tarballs, 1) VCS archive 2) autotools cruft 3) submodules 4) prebuilt docs etc. then users can select the ones they want and combine them themselves

<pabs3>ACTION not on the list though

<pabs3>Googulator: ^

<matrix_bridge><Andrius Štikonas> well, VCS archive + submodules is a useful choice of real source archive

<matrix_bridge><Andrius Štikonas> autotools cruft and prebuilt docs is just helping others not to install more dependencies

<pabs3>ACTION would separate out the submodules so that people who like multiple copies of code can have that and people who don't (distros) can have that

<stikonas>yeah, but generally submodules do need to be at specific commit

<stikonas>e.g. even the same gnulib

<stikonas>different tools use different gnulib versions and there is no source compatibility

<mihi>Googulator & stikonas: samplet's patch is for nyacc, not for mes, and I doubt anybody wants to fork the old nyacc version mes uses to make a new release. So i'd keep the patch out-of-tree (either in mes or in live-bootstrap). And whether syntax-rules (and defmacro) requires psyntax or not depends on how macro support is implemented. In the way how mes do it, they don't need psyntax, in the way Guile does it, they

<mihi>do (which made bootstrapping psyntax on Guile slightly harder)

<Googulator>stikonas: nyacc-1.00.2-lb1 time it is, then

<Googulator>janneke: is a bugfix release of mes-0.26 planned in the near future, to fix the double-compilation bug?

<janneke>Googulator: certainly!

<Googulator>because at this point, the nyacc pregen files are basically *the* blocker for live-bootstrap v1.0 release, and the fix for that depends on 0.26

<Googulator>(although I would like to also prioritize openssl-3.0 now that 1.1.1 has gone EOL)

<stikonas>GoogulatorMobile: yeah it's hard to patch before patch program

<stikonas>so separate snapshot is easier

<Googulator>fossy: what's your opinion on the OpenSSL 3 upgrade? It's currently marked optional for 1.0, but the release we currently use (1.1.1) is EOL

IRC channel logs

2024-04-12.log