IRC channel logs

2021-03-11.log

back to list of logs

<fossy>stikonas: I guess so
<stikonas>ok, wrote Makefile
<stikonas>wasn't too hard...
<stikonas>slightly ugly thing with sed is that we don't have tarball...
<stikonas>only dirty source checkout with object files
<Hagfish>it's weird that a problem with sed should end up affecting things, and even weirder if the problem is due to mes
<Hagfish>i know sed scripts can get arbitrarily complicated, but i didn't think that many features of it were actually used in the bootstrapping process
<stikonas>Hagfish: it's not too weird...
<stikonas>I think mes has a lot of stubs that break some things
<stikonas>we didn't observe it too much when we were doing simple things like simple string substitution
<stikonas>but autotools use all those other tools much more heavily
<stikonas>and there are also some minor bugs
<stikonas>in mes libc...
<stikonas>I think pder already sent a couple of patches upstream
<Hagfish>stikonas: that's great info, thank you
<Hagfish>can you give an example of some deep feature of sed that autotools uses?
<stikonas>well, the thing I hit was autoheader script
<stikonas>let me see if I can find source
<Hagfish>thanks
<Hagfish>i see that LFS suggests running a sed test suite
<Hagfish> http://www.linuxfromscratch.org/lfs/view/9.0/chapter05/sed.html
<Hagfish>i know it's not sed's job to find problems in mes, but it would be cool if that worked
<stikonas> https://git.savannah.gnu.org/cgit/autoconf.git/tree/autoheader.in?h=autoconf-2-13-rc1#n125
<stikonas>this sed script
<Hagfish>yeah, wow, that is a little spicy
<pabs3>is there a way to contact rain1 when they aren't here? (want to ask about enabling the API on the bootstrappable wiki)
***janneke_ is now known as janneke`
***mephista is now known as spicy_icecream
***janneke` is now known as janneke
***adds68_ is now known as adds68
<pder>stikonas: nice find on sed. would you be able to push what you have as far as a new makefile?
<stikonas>pder: I can probably go as far as rebuilding binutils, but might not be today...
<stikonas>need to go grocery shopping this evening
<stikonas>maybe I'll be able to push at least sed
<stikonas>and do libtool/binutils tomorrow
<pder>I wanted to try rebuilding or upgrading gawk
<stikonas>I also have a fix for autoconf-2.13...
<stikonas>ok, I'll try to push sed PR today then
<stikonas>so that you are unblocked on gawk...
<stikonas>pder: you'll probably need newer autoconf/automake?
<stikonas>that will be a bit annoying though
<pder>for gawk?
<stikonas>yes
<stikonas>at least for 3.1.7
<stikonas>current gawk is fine
<stikonas>although, maybe it will be fixed if you just rebuild
<stikonas>especially with fixed sed...
<pder>Ill try that first
<pder>Also I can look at upgrading tar
<pder>do you know if that autoheader problem has anything to do with bash, or was it just fixed with sed?
<stikonas>pder: fixed with sed
<stikonas>no need to rebuild bash early
<stikonas>pder: if possible tar > 1.14 please
<pder>ok
<stikonas>pder: https://github.com/fosslinux/live-bootstrap/pull/61
<pder>stikonas: thanks. Currently I have the latest master commit 0d84a1c2dcdff1481dd1caf43 and it is failing with autoconf-2.13
<pder>oh it looks like you addressed it
<stikonas>pder: it's somehow failing intermittently
<stikonas>not sure why
<stikonas>we have been building autoconf 2.13 for some time
<pder>this is the first time ive seen that error
<stikonas>I saw it twice
***warren_ is now known as warren
<stikonas>pder: we can probably get rid of tcc-mes and mes libc too now
<stikonas>we are not really using it...
<stikonas>even though I initially left them
<stikonas>at this point I don't see where it might be useful anymore
<pder>I agree that they could be removed. At some point it might be investigate what is broken in mes libc and fix it.
<pder>s/investigate/good to investigate/