IRC channel logs


back to list of logs

<phant0mas>in order to build glibc in commement, I need the headers, which need mig, which need flex
<phant0mas>aaaaa :'(
<civodul>flex can be built with %bootstrap-inputs, i think
<phant0mas>flex-boot0 it is then
<civodul>ACTION sympathizes
<phant0mas>I should really do a rebase of wip-hurd on core-updates and push it to wip-hurd
<phant0mas>then it'll be easier for you to review wip-hurd
<phant0mas>and possible merge them sometime in the (near?) future :-)
<civodul>yes, that would be nice
<mark_weaver>civodul: so, I was thinking for the next core-updates cycle, we could improve the timestamp handling in patch-and-repack: instead of making all timestamps zero, instead preserve timestamps of all files that aren't modified, and all modified files would have a timestamp 1 second newer than the newest mtime in the original tarball.
<mark_weaver>or something along those lines
<mark_weaver>I dunno, maybe it's more complexity than we need
<civodul>yeah it sounds complex
<civodul>honestly i didn't expect the kind of issues we had with libpsl
<civodul>that's probably rare
<mark_weaver>okay, we'll see how often these problems turn up.
<civodul>anyway, going to bed, so the next shift can take over :-)
<mark_weaver>okay, good night!
<rekado>I'm trying to package NTK (a fork of FLTK with graphics rendering through cairo), but they don't provide tarballs.
<rekado>So I use git-fetch to get a particular tag, but the hash of the checkout appears to change each time I try to build it.
<rekado>The repository is git://
<mark_weaver>rekado: that shouldn't be possible with a git repository. are you using a hash commit id (as opposed to a branch name)?
<mark_weaver>the commit hash is a hash of not only the entire directory contents but also the commit history. it should not be possible to change anything in there without changing the hash.
<rekado>I'm using a tag reference.
<mark_weaver>rekado: that suggests that the tag is being updated
<mark_weaver>what tag?
<rekado>I'll try with just the commit id.
<rekado>commit id works.
<rekado>oh, I see...
<mark_weaver>looks like tag v1.3.0 has been stable for 22 months or so
<rekado>yes, it's my fault.
<rekado>I supplied the wrong tag name.
<mark_weaver>ah, okay
<rekado>I guess I should have breakfast.
<mark_weaver>rekado_: thank you for fixing git in core-updates!
<rekado->mark_weaver: I don't think I did.
<mark_weaver>rekado-: oh, sorry, I got mixed up. it was bavier1 :)
<mark_weaver>bavier1: thank you for fixing git in core-updates :)
<mark_weaver>bah, I've hit a code generation bug in gcc 4.9.3 on arm/neon.
<mark_weaver>it miscompiles openssl, leading to bus errors :-(
<mark_weaver>it is probably an instance of the bug described here:
<mark_weaver>although I have a simpler test case
<civodul>Hello Guix!
<civodul>grrr that ath9k issue is annoying
<civodul>anyone using it here?
<civodul>apparently our firmware is okay, but linux 4.1 isn't, somehow
<zacts>hi guix
<rekado->ghostscript-9.16/Resource/IdiomSet/Pscript5Idiom --- this file is part of Ghostscript 9.16.
<rekado->What is the difference between GNU Ghostscript and AGPL Ghostscript?
<civodul>it's not completely clear
<civodul>maybe you can package the other one, and have it used by lilypond?
<civodul>so it seems that we need this thing:
<mthl>civodul: That looks like a lot of fun!
<rekado->bah, ghostscript bundles a lot of libraries and doesn't build when I remove them. More work than I expected.
<civodul>mthl: definitely
<civodul>my guess is that linux 4.0 worked fine without that, but 4.1 no longer does
<rekado->erm, gnu-ghostscript *also* comes with Resource/IdiomSet/Pscript5Idiom, but the file is not installed.
<civodul>uh uh
<rekado->same with AGPL Ghostscript --- the files are not installed.
<civodul>think of all the joy once you've fixed this :-)
<civodul>maybe lilypond is making an incorrect assumption
<rekado->when built with "-disable-compile-inits", the resource files are installed.
<rekado->unfortunately, this doesn't fix anything.
<rekado->maybe it's a font issue after all?
<civodul>hey, remi`bd
<civodul>remi`bd: i just stumbled upon this:
<civodul>it's nice that people are talking about this!
<civodul>i don't know who this "Lynx" is
<civodul>now you'll have to get there ;-)
<remi`bd>“you broke the internet; we’ll make us a GNU one” isn’t that Christian’s moto?
<civodul>well, not only his
<civodul>wait, we're not on the "project map", that's unacceptable!
<mark_weaver>bah, well, I can't make any more progress testing armhf now, because I can't get the netpbm source code
<mark_weaver>their home page is sourceforge, which is having extended downtime, and their subversion repo (which does not have a sourceforge URL) is also down.
<mark_weaver>civodul: hydra is unable to download the libpsl source code:
<quux>i downloaded guix sd and it's a xz file
<quux>how can i make it something virtual box will use?
<isd>quux: did you try decompressing it (unxz)? iirc, it's just a raw disk image
<quux>yeah i used
<quux>xzcat guixsd-usb-install-0.8.2.x86_64-linux.xz > guixsd-usb-install-0.8.2.x86_64-linux
<quux>but im a bit stuck now
<quux>First I tried dd'ing that to a USB to boot it and it did boot but that computer overheated so I can't use it
<isd>quux: it's been a long time since I've used virtualbox. There should be a way to hook up a disk image to the USB bus somewhere.
<quux>oh i got it using
<quux>vboxmanage to convert into a vdi format
<civodul>mark_weaver: oops
<mark_weaver>^^ bug report I just filed about a GCC code generation bug on ARM that causes a Bus Error in OpenSSL when targetting NEON.
<civodul>nice that you were able to isolate it!
<civodul>i've never tried using "Delta" to obtain reduced test cases, but reportedly it works well
<mark_weaver>ah, good to know. I just used my usual stone tools :)
<civodul>anyway, it looks like a perfect bug report
<civodul>i'd be happy if i were "them" ;-)
<mark_weaver>I hope so. there are some similar bug reports that have been sitting unfixed for a couple of years; hopefully the minimal test case will help with that.
<mark_weaver>anyway, I have to go afk now. happy hacking!
***jgay is now known as GrumpyGNU
***francis7 is now known as GoodGNU
***GrumpyGNU is now known as jgay
***GoodGNU is now known as francis7
***francis7 is now known as IGNUThatMuchAlre
***IGNUThatMuchAlre is now known as IGNUItAlready
***IGNUItAlready is now known as francis7