*rekado just fixed the build of the old hurd glibc <drp>searching the internet revealed that on this channel a person stated that the pinebook is 85% of the way there <drp>I can see that there's pine64 firmware in guix, does this work for the pinebook as well? Anyone actually using it here? *janneke is trying to eradicate obsolete SUN rpc support -- why does it seem so hard, why does it seem that our software stack still infested with that? <ng0>what's the requirement for Julia's testsuite again? 32 GB RAM, or is 16 enough? <mbakke>janneke: I have a patch for disabling sunrpc in glibc laying around. <mbakke>I needed it for GCC7 (for which I'm stuck on GCC-FINAL). <janneke>mbakke: nice! i'm working on glibc-2.16.0, but it could help to inspire me (and it may even apply) <janneke>i thought i removed it, but apparently gettext-minimal includes netdb.h, which includes rpc/netdb.h -- i just removed that include from glibc-2.16.0... <janneke>mbakke: do things like these look familiar? <janneke>getent.c:705:19: warning: 'struct rpcent' declared inside parameter list [enabled by default] <mbakke>janneke: Oh, right. I haven't built very far without sunrpc support. <mbakke>The Guix CVE checker seems unable to parse the recent Ghostscript entries because they are not tied to a product in the XML. <janneke>mbakke: need to run, but a pointer to your patch is much appreciated -- i'm trying and struggling anyway <janneke>okay...perl-boot0 fails -- major step forward! <janneke>the strange thing is, that perl-boot0 also has bootstrap leaks...oh well <janneke>hmm, bison-boot0 invokes ranlib -D...but binutils-boot0's ranlib does not have -D? <janneke>could it be that my `bootstrap leak' fixes broke this...hmm <janneke>OriansJ: thanks! Now building gcc-final... <janneke>I made some really ugly libtool hacks to build this, composing an email to guix-devel now <janneke>i'm sure this can be fixed, but atm i lack the awarness of what's causing this, why i need such hacks <roptat>how do I set a timeout with fibers? <roptat>I'd like to do something like (perform-operation ...) but if it didn't perform anything after some time, I'd like it to continue and return 'timeout or somthing similar <rekado_>on berlin “herd status” freezes up; nginx cannot be stopped / started; a defunct nginx process lies around. ***rekado_ is now known as rekado
<rekado>civodul: berlin seems stuck, can’t even reboot <rekado>is there a way to get shepherd to reload? *rekado unfortunately has to go afk for the next few hours <rekado>berlin.guixsd.org is currently without nginx *janneke is building guile-final! <janneke>it would be nice to have a big iron i686-linux offload server <rekado>civodul: I started nginx manually, but the configuration isn’t correct. Like before all extra domains are wrong (e.g. issues.guix.info doesn’t redirect properly). <civodul>shepherd showed signs of life at 14:36 (for a reconfigure operation) <civodul>stat("/var/run/nginx/pid", 0x7ffd498a7520) = -1 ENOENT (No such file or directory) <civodul>rt_sigprocmask(SIG_BLOCK, NULL, [], 8) = 0 <civodul>select(4, [3], NULL, NULL, {tv_sec=1, tv_usec=0}) = 0 (Timeout) <civodul>stat("/var/run/nginx/pid", 0x7ffd498a7520) = -1 ENOENT (No such file or directory) <civodul>i wrote something to that file, and then it immediately rebooted (i guess it had reboot messages in its queue) <jlicht>congrats on joining the spending committee, nckx! <civodul>i have to go as well for a few hours :-/ <janneke>`configure: error: Guile-Gcrypt could not be found; please install it.' <janneke>guix environment --ad-hoc guile-gcrypt <janneke>guix environment: error: guile-gcrypt: unknown package <janneke>./pre-inst-env guix environment --ad-hoc guile-gcrypt <janneke>i don't want to guix pull, becase we have decided to stay of version-0.15.0 for now and be able to exchange substitutes *janneke found civodul's Guile-Gcrypt heads-up mail... <snape>janneke: I think you need to roll back your git checkout to a point where (gcrypt hash) wasn't needed, install guile-gcrypt, and update your git checkout <apteryx>hello! To get guix offload to work with a foreign distro, I found I need to add 'guix-authenticate' to the PATH. It is typically found in under /path/to/guix/libexec/. This directory seems absent from the new Guix deployed at ~/.config/guix/current/. Is this normal? <apteryx>Workaround would be to install Guix in my user profile on that machine and add ~/.guix-profile/libexec to the PATH. <nly>How can I view the guix manual in emacs? <atw>I somehow missed the "in emacs" part of the question <ecbrown>nly: export INFOPATH="$HOME/.config/guix/current/share/info:$HOME/.guix-profile/share/info${INFOPATH:+:}$INFOPATH" <ecbrown>you may have to guix package -i guix <ecbrown>(as well as the up-to-date documentation) <ecbrown>you're welcome! just passing on wisdom i learned from #guix <emacsomancer>I'm trying to create a package build, but it's complaining about missing .h files even though they're part of the package itself - do I need to add something in order for 'local' .h files to be visible to the build? *jonsger tries to update LibreOffice to 6.1.0.3... <rekado>something weird is happening on berlin.guixsd.org <rekado>I get proxy timeouts for both cuirass and mumi. <civodul>rekado: i just had the same reaction :-) <civodul>is mumi running at all? there's no Shepherd service <rekado>(it’s running from a git checkout in a screen session; not ideal) <rekado>hmm, I get “upstream timed out (110: Connection timed out) while reading response header from upstream, …” for both cuirass and mumi. <civodul>cuirass' web server seems to have a problem <civodul>one of its fibers is building something: "ok\ntest [run-pass] run-pass/issue-41803.rs" <civodul>oops i "inadvertently" restarted cuirass <civodul>i was gdbing it, and my hypothesis was that cloning/pulling a git repo from one of the fibers was hanging, leading the whole thing to hang <rekado>huh, so: when I remove the “recent activity” section issues.* loads up fine. <rekado>but any debbugs search times out <civodul>maybe debbugs has become slower and thus mumi is slower? <rekado>(or maybe debbugs admins saw many requests from one IP and applied rate limiting?) <civodul>perhaps you should get in touch with help-debbugs <civodul>actually i think there's no admin left for that since a few weeks ago... <rekado>it’s fine when I run it from my IP. <jonsger>building LibreOffice is very...tough... <rekado>hmm, I wonder if it’s just my browser caching failures…? <rekado>I also think mumi should use a fiberized server <roptat>am I right at suspecting guile-fibers is actually not multi-threaded? <civodul>it spawns several kernel threads and runs fibers on top of them <civodul>you can tell 'run-fibers' how many kernel threads you want <roptat>ok... because I have this behavior where a fiber only seems to be run when it received a second message <roptat>(so the first one was sent, but the receiving fiber didn't run anything until it was sent a second message and the first fiber blocked) <roptat>but then, nothing happens between the first and second message, because the first fiber is just waiting for things from the network <roptat>mh... that may be one of those pitfalls, no? <civodul>read the "Pitfalls" section in particular *civodul goes afk for a while <rekado>In my Hurd VM I can now build packages with Guix. <rekado>I can’t use substitutes, though, because it looks like the substitute helper uses a syscall that’s not available on the Hurd. <rekado>hmm, I get “error: failed to run download program '/root/guix/nix/scripts/download': Permission denied” – but I can run the thing manually. <roptat>rekado: I've downloaded the debian/hurd image, what options do you use with qemu?