***scs is now known as Guest20543
<stikonas>pder: by the way, maybe early sed 4.0.7 can be dropped in favour of 4.0.9? <Hagfish>i'm glad my connection came back in time for me to read about automake. great news! <stikonas>well, building autotools is not particularly exciting <stikonas>but it means that we are getting closer to gcc <Hagfish>i can't imagine what twists and turns gcc is going to add to the journey (other than noting that it's stubbornly not reproducible, as distros have found) <Hagfish>so yeah, getting gcc built as part of the bootstrap sequence is going to be a much more exciting milestone <stikonas>well, we'll see whether we can keep reproducibility... <stikonas>if not, we can still finish bootstrap without reproducible builds <Hagfish>yeah, there might be some interesting discoveries (or at least hacks) to be found in the process <stikonas>but I thought guix can build gcc reproducibly <Hagfish>it might be that the environment for doing the bootstrap is sufficiently constrained that the unreproducibility won't be a practical problem <fossy>The one package that reproducibility is probably not a feasible goal for is gcc <stikonas>well, there are two things, gcc itself and packages built with gcc... <stikonas>I think we need to get to autocont 2.59 first <stikonas>I don't expect much difficulty with those... <Hagfish>does that mean we'll need much more recent versions later on? (for more modern gcc?) <stikonas>but then possibly just build each newer one with the previous version <stikonas>and i think autotools became more compatible later <stikonas>it's that period around 20001 where there were many incompatibilities <Hagfish>it would be nice if it was just a phase that free software development went through, before learning their lesson, but i think new languages/ecosystems are still making bootstrapping mistakes <stikonas>we probably just neeed a few versions in total 2.12, 2.13, 2.52 and maybe one new version <stikonas>it's probably just a matter of autoconfs maturing <Hagfish> This is a small advance compared to 1.2. It adds support for assembly, and preliminary support for Java. <Hagfish> Perl 5.004_04 is out, but fixes to support Perl 4 are still regularly submitted whenever Automake breaks it. <stikonas>which already incorporates APIVERSIONing <Hagfish>"These two releases, Automake 1.4 and Autoconf 2.13 make a duo that will be used together for years." ***xwvvvvwx- is now known as xwvvvvwx
<fossy><stikonas> well, there are two things, gcc itself and packages built with gcc... <fossy>packages built with GCC shouldn't be too hard... ***scs is now known as Guest17776
***remexre_ is now known as remexre
***scs is now known as Guest61540
***scs is now known as Guest67472
***scs is now known as Guest59056
***pgreco_ is now known as pgreco
***jelly1 is now known as jelle
***scs is now known as Guest16313
<stikonas>fossy: automake stuff needs some reworking, I'll hopefully do it this evening <stikonas>that aclocal.m4 file was actually used in stage 1 (both automake 1.4 and 1.6) <stikonas>need to manually bootstrap aclocal first (with sed) ***scs is now known as Guest63903
<pder>stikonas: I will see if sed 4.0.9 can be built early on ***stikonas_ is now known as stikonas
***vup is now known as __vupbot
***__vupbot is now known as vup
***scs is now known as Guest93585
***scs is now known as Guest99582
***scs is now known as Guest99737
***scs is now known as Guest31785
<fossy>I was almost 100% sure that it isnt <fossy>cause in stage1 we only use configure <fossy>and aclocal.m4 is regened when we RM it afaict? <fossy>unless im wrong... I did start going down the bootstrap aclocal path but I didnt think it was nessecary <stikonas>like I said yesterday, aclocal is always before configure <stikonas>it shouldn't be too hard to create aclocal.in->aclocal with sed <stikonas>it's just setting some instalation paths, etc... <stikonas>fossy: so I'm thinking of moving automake 1.6 much earlier (before 1.4) <fossy>stikonas: how does stage1 even work then <fossy>I realised why I didnt run into my ci error <fossy>I think I was on the wrong branch oops <stikonas[m]>But I only yesterday realised that aclocal is used there ***scs is now known as Guest97525