IRC channel logs

<jxself>civodul: I think you missed my earlier message. The DMD site is up. I had to improve a bit based on what I knew, like referring people to the guix-devel mailing list. Hope that's okay.

<jxself>er; improvise.

<jxself>I can't type...

<jxself>Also, would you be willing to consider GPLv3 for DMD maybe? :)

<civodul>jxself: ooh excellent, thanks for doing it!

<civodul>i thought i'd do it yesterday, but then basically fell asleep

<civodul>and yes for GPLv3 :-)

<jxself>Sleep is an important thing though.

<civodul>indeed

<civodul>jxself: i realize i'm not yet the new dmd maintainer

<civodul>although i emailed ftp-upload pretending to be the maintainer ;-)

<civodul>so we'll have a QEMU image good enough to do Guix hacking in it

<Steap>good

<Steap>will you be merging core-updates in master tonight ?

<civodul>Steap: i think so

<civodul>i'm basically waiting for feedback from hydra

<civodul>it was kinda stuck lately

<Steap>ok

<mark_weaver>civodul: I'm interested in investigating the possibility of creating bootstrap tarballs from outside of Guix, as an alternative bootstrap method. Do you see a reason why this is infeasible?

<mark_weaver>(my motivation is to avoid creating a single point of trust, e.g. the x86 bootstrap tarballs)

<mark_weaver>remember, the bootstrap tarballs don't have to be identical to the ones Guix creates. They need only be sufficient to use Guix to create a more pristine set of bootstrap tarballs.

<mark_weaver>am I missing an important detail?

<civodul>mark_weaver: no, that should be doable

<civodul>the only thing is that you'll have to mimic by hand what's written in the make-bootstrap.scm recipes

*mark_weaver looks

<civodul>we discussed this months ago when Nikita was trying to port to MIPS

<mark_weaver>okay, thanks. I'll take a look at that code.

<mark_weaver>I think this is likely what I'll be working on during the 30th anniversary hackathon in Cambridge MA

<civodul>excellent! :-)

<civodul>that's definitely a useful area

<civodul>BTW make sure to bring others in the hackathon while you're there ;-)

<mark_weaver>civodul: how do I compute the hash of the bootstrap tarballs? what's the theory there?

<civodul>sha1sum foo.tar.gz?

<civodul>what do you mean?

<mark_weaver>the Nix hash

<mark_weaver>or do the bootstrap tarballs not have a hash? I confess I've not looked closely.

<civodul>the hash in the file name is the hash of all the inputs

<civodul>and no, the bootstrap tarballs don't have that

<mark_weaver>ah, okay.

<civodul>well, if you type "guix build bootstrap-tarballs", that returns /nix/store/xxx

<civodul>but that directory just contains plain tarballs

<civodul>and their contents do not depend on the directory name

<civodul>any idea what the right "video=" option is to get a framebuffer device that works with QEMU?

<mark_weaver>well, I guess my (probably premature) question is: once I've hand-created the bootstrap tarballs, how will Guix find them? which /nix/store/xxx directory will it look for them in?

<civodul>mark_weaver: you just need to have bootstrap.scm refer to them

<mark_weaver>okay

<civodul>see http://www.gnu.org/software/guix/manual/guix.html#Bootstrapping

<civodul>under "Building the Bootstrap Binaries"

<mark_weaver>thanks

<civodul>Steap: still 46 packages to build at http://hydra.gnu.org/jobset/gnu/core-updates

<civodul>so i may wait until tomorrow morning to press the trigger

<civodul>there are very few changes in core-updates, so hopefully nothing will go wrong

<Steap>civodul: yeah, I'd just like to test gobject-introspection with the new default Python and the fixed ldd before pushing it

*jxself watches as everything goes wrong

<civodul>:-)

<civodul>Steap: i thought you had tried in core-updates, no?

<Steap>civodul: I think so

<Steap>civodul: I think we still sued Python 2 as the default at the time

<Steap>not sure

<Steap>so Ithough I might as well try in master once core-updates has been merged

<civodul>ok

<civodul>we'll see, then :-)

IRC channel logs

2013-09-25.log