<jmd>Does /homeless-shelter exist in the build environment? <civodul>so it was probably an out-of-memory situation, no? <alirio>it allocates memory to execute a file? <civodul>the root file system on the image is a unionfs (FUSE), so i'm not sure exactly how it does thing <civodul>but yeah, it most likely allocate things on the heap somehow <alirio>but outside the daemon it always worked, and I have more than 1GiB free on a idle system; I think its more likely to be a bug in the driver <civodul>yeah, it works well for simple uses, but it's probably not very robust <jmd>Who is interested in libreoffice for guix ? <jmd>Because it takes over 10 hours to build. ***pigoz1 is now known as pigoz
<DusXMT>ah, good old libroffice, heavier than a browser <jxself>Wildebeest could probaly help with that. <jxself>I spend more than that amount of time compiling kernels. <jmd>jxself: But not one kernel ? <jmd>Last time I built a kernel it took less than an hour. <jxself>3 to 4 different versions and multiple variations of each version. <jxself>Something like 15 packages get made or somewhere around there. <jxself>A machine I donated to help build packages for Guix. <jmd>Oh. Great. How do I get on it? <jxself>Hydra submits build jobs to it automatically, so just get the package into the guix git repository. The build farm takes over from there. :) <jmd>But the package has to be ready. It wont do half finished packages. <jmd>I suppose I could push it in a half finished state, but I don't suppose civodul will be happy. <jxself>I'll let civodul speak to that. I wonder if it could go into a branch and then get merged? <davexunit>we should be able to make a wip-libreoffice branch or something <davexunit>and that will rebased as needed, and merged when ready <jmd>Might be an idea. Then others can work on it too, and my poor little box can have a rest. <jmd>jxself: What architecture is Wildebeest ? <civodul>jmd: you can push it to a separate branch, if you want, and ask for help on guix-devel <jmd>ok. Do I have permission to create branches? <jmd>Is this wildebeest box the same one that is hosting www.gnu.org ? <jxself>No, that's just a hostname collision. It's actually wildebeest.jxself.org. <jmd>Well they go around in herds. <jmd>Anyway, I cannot pull from git. Is it down at the moment? <civodul>note that this particular host name is not particularly original in our context ;-) <jmd>What does one call a female wildebeest ? <jxself>Hmm. I don't think I've ever heard of different names. Time to go look... <jxself>Not finding anything, but the gnu project's mascot seems to be the blue wildebeest from the direction of the horns. <davexunit>can you use the provided guix disk image to burn to a cd? <davexunit>haven't tried it, but people ask about in #gnu and such, and I want to have a better answer for why we don't provide ISOs. <jmd>jxself: What's the systematic name ? <jxself>Is what you're thinking of right? :) <jmd>No, I meant the sytematic name of a wildebeest <jmd>Or Connochaetes taurinus <jmd>The next machine can be called connochaetes <jxself>Connochaetes taurinus seems to be the mascot of the gnu project. <civodul>davexunit: the reason for not providing ISOs is that (1) it's more work ;-), and (2) they are inconvenient to many, and (3) one can use UNetBootin or similar to make one <civodul>i wasn't convinced it was worth it at the beginning <civodul>oh, actually UNetBootin is the other way around <jxself>I have to say I'd also like to see an actual ISO image too. <civodul>jxself: oh thanks for the note about UNetBootin <civodul>i'm afraid it won't be for 0.8, unless someone volunteers? <jxself>Doesn't have to be. Just a "would be nice" IMO. <jxself>I picked up a copy of The Little Schemer. It wasn't at all what I expected. <jxself>Seems to just be a whole bunch of examples of what is (or isn't) and then you have to figure out why and make sense of it. <civodul>(i have The Reasoned Schemer, which uses the same style) <civodul>davexunit: could it be that the CVS web-page update hook is broken today, or that '+' is forbidden in file names for the Web repo? <davexunit>I'm a little overwhelmed with work at the moment, but if it continues to be a problem I can take a look sometime <civodul>mark_weaver: should we update both in core-updates, while we're at it? <civodul>but for GnuTLS we may hit the limitation due to "non-recursive" grafting <th3kent``>i tried to add extra modules (pata_acpi.ko, pata-atiixp.ko) to my initrd but my "guix system init ..." command fails with "guix system: error: failed to load operating system file '/mnt/etc/config.scm': (unbound-variable #f "Unboung variable: ~S" (cut) #f)". help. <taylanub>question: patching e.g. bash means rebuilding everything because it's a gnu build system input, right? couldn't one use a "bash-for-compilation" package for this which is rarely ever updated, since even if e.g. security holes are found in bash, they don't matter for compilation? <civodul>th3kent``: oh, you need to add (srfi srfi-26) to the (use-modules ...) form <civodul>taylanub: it could work that way, yes <civodul>that how we handle guile upgrades (but, ahem, they're infrequent these days ;-)) <th3kent>civodul: am making progress! my "guix system ..." command this time fails with 'In unknown file:\\n?: 0 [system* "/gnu/store/...-cpio-2.11/cpio" ...]\\nERROR: In precedure system*: Cannot allocate memory'. i currently have 237mb free r.a.m. (???) <civodul>has it been building lots of things, or just downloading? <th3kent>yes it's an hp nx6125 laptop. before the error i got "In ice-9/boot-0.scm:\\ <th3kent>157: 11 [catch #t #<catch-closure ...>] <davexunit>civodul: are we going to ease the pain of installing an old guix with this release? <civodul>davexunit: you mean installing from an old image? <davexunit>civodul: how installing from 0.7 gives you 0.6 <civodul>davexunit: no; that's not exactly what happens though, and one can always use "guix pull" <th3kent>buit plenty of files, 1577 operations. <civodul>so that's not optimal, but that doesn't render it unusable either <civodul>i should really upload a 0.8ish image for you to test <davexunit>well, it installed a sufficiently old guix that it made the 0.7 config used to install the OS unusable for doing guix system reconfigure <civodul>davexunit: right; but one could still use "guix pull" to get the latest <civodul>(modulo bugs, but there are fewer of them in that area now ;-)) <davexunit>as long as guix pull works, I suppose that's fine. <DusXMT>But I think it doesn't work afaik <civodul>now is a good time to report bugs if you have evidence that it doesn't work <civodul>i'm not aware of any serious bug in "guix pull" at this time <DusXMT>I wonder, in the installer images, how about including a slightly updated ~/.config/guix/latest for each user with definitions that recognise the given guix version? <civodul>there's the possibility of keeping the currently installed Guix, as i mentioned on the list <civodul>i'll push the GnuTLS upgrade in master (62 packages depend on it) <davexunit>civodul: we have an outage! FSF sysadmin and savannah admins working on it <davexunit>civodul: fyi, the great emacs migration is what killed git.sv.gnu.org <jxself>I should start bzrhub in defiance. :) <civodul>i didn't know they had finally migrated, sounds like good news <davexunit>yeah, after the 24.4 release, ESR started the migration. <civodul>oh he really did, after all this chatting? *civodul is being tongue in cheek <davexunit>I'm sure none of us really agree with ESR on a lot of things, but I'm happy to see him do the grunt work for this <civodul>yes, but that also has a cost, socially speaking