IRC channel logs


back to list of logs

<ksjogo1>Did anything change in the recent icecat version regarding AA? The menu still looks fine, but AA is gone from the websites.
<ng0>oh.. the person who packaged it should knwo
<ng0>so when I only need the python-2 versions of new packages I will add soon, should I add the python3 versions (if available) for completeness?
<paroneayea>submitted my autossh patch to the list
<paroneayea>er, package
<ng0>duplicity check: is anyone working on the following packages (python ones): bleach, flask-wtf, flask-multistatic, kitchen, munch, pyclamd, pygit2, openid, openid-cla, openid-teams, straight.plugin==1.4.0-post-1, trollius-redis, python-fedora, python-jenkins
<ng0>well 22:44 UTC might not be the best time to ask though.. but maybe someone who does work on one or more of them will reply later. I read the log online. It's planed to work on these dependencies in the order I just listed them.
<ng0>if anyone wants to pick any of the packages and work on them, please inform me, it's very welcome :)
<ng0>those are dependencies for pagore
<ng0>as far as I see it, pagore is python-2.7.2, so you also would need to package the python-2 variants
<thomassgn>how do I set domainname? I see hostname mentioned, but nothing about domain? computer in question is a server.
<thomassgn>should I just set (host-name "hostname.domainname.tld") ?
<lfam>Working on the Go 1.7.1 failures on core-updates:
<lfam>I have Go 1.4.3 working
<lfam>Anyone here have any ideas about that failure? I can provide the patches I'm building with (I already fixed some other test failures)
<paroneayea>davexunit: btw, not sure if you saw that the "guix ops" thread reignited on list a bit, in case you have some feedback
<paroneayea>but no wories if you have no time :)
<apteryx>Hello! Which web browsers do you use?
<apteryx>I'm using IceCat right now but for it stalls.
<gry>how does it stall?
<apteryx>It's like the Window stops being drawn. If I scroll nothing happens. CPU reacts to my scrolling but the rendering appears stuck.
<apteryx>Maybe some issue with Ratpoison?
<apteryx>I disabled both SpyBlock & LibreJS to test. So far so good.
<apteryx>I think I found the issue. It's when I watch a video with the HTML5 player. Everything becomes slow after that, even after I close the page with the videoplayer.
<reepca>apteryx: I experienced something a bit like that on my laptop, but I just assumed it was the flash drive. I really need to go about putting guixsd on the hard drive, but I'm not sure how to get dual booting working...
<apteryx>Guix will want you to install grub to the MBR iirc. It'll probably overwrite whatever was there, so you will have to tell later where is the other partition you would like to boot from (grub configuration file).
<apteryx>I can only recommend a full backup of your stuff in case something goes wrong. I'm glad I did; when I resized my partition I wasn't careful enough and ended making it too small (parted didn't warn me about that). It's probably corrupted since I installed Guix next to it, and even had the stupidity to fsck it a little ;). Backup FTW!
<reepca>So I would have to tell GRUB (on the guixsd side, since the GRUB install from the other system would be overwritten) about the other operating system on the hard drive? Would I do that with the bootloader field of my config.scm? Do I just need to tell it the partition name (/dev/sdaX)?
<apteryx>I isolated the problem. It's probably something to do with Ratpoison, or the lack of a compositor maybe. The problem is triggered when I press the fullscreen icon of the IceCat video player. Nothing happens (the screen is not maximized), and IceCat becomes unresponsive.
<apteryx>I'm sure there would be a neat way to define those in config.scm, but I'm quite new to this so hopefully someone more knowledgeable about the declarative config part can chime in.
<reepca>I guess it's probably All There In The Manual™ somewhere
<apteryx>The manual is quite good! I'm like 30% through it maybe.
<apteryx>info guix
<apteryx>And when you'll be in the installer it's setup on the 2nd virtual terminal (tty), Alt-F2. It' nice to refer to it as the system installs itself.
<apteryx>Well you probably saw that since you installed it in a VM ^^
<reepca>I didn't install it in a VM, I installed it straight onto a flash drive from my desktop using guix system init
<apteryx>OK! Nice. It seems the authors went to lenghts to make it installable in many ways.
<reepca>looks like I should be adding menu-entries, which are of data type menu-entry, to my grub-configuration field. And a menu-entry needs a label, a linux kernel, and an initrd... now how do I find out what those values are?
<reepca>Yeah, the problem is just that now I wish there was a "guix system clone", since I already downloaded some stuff and spent some time getting stuff set up (slowly and painfully) on the flash drive system
<apteryx>reepca: If you put the effort in declaring all the packages you want in your config.scm it should be relatively painless to mirror your current setup.
<apteryx>There might even be a way to reuse your current /gnu/store to speedup the process.
<reepca>is there a way for each user to have their own config.scm, or do I just have to specify the packages available to all of them by default in the system's config.scm?
<apteryx>Sorry I was trying to find in the manual.
<apteryx>From what I understand the config.scm is the *system*'s declaration.
<apteryx>Which means it will prepare whatever package you request into /gnu/store & any configurations you defined, such as your grub config.
<apteryx>What is in the store and what is "installed" is different in Guix.
<apteryx>Packages under /gnu/store would be invisible for a new user.
<reepca>so the closest thing to carrying over what my user has installed is to list it in config.scm so it gets copied over into the store (hopefully), then manually look through and install what is missing for my user?
<apteryx>(If I'm not mistaken).
<reepca>makes me wish the user-account data type had a field for listing the default installed programs or something like that
<apteryx>reepca: Actually I think the packages defined in your config.scm *will* be installed for any user. I don't see how it would work otherwise. I think what confused me is that when I do a "guix package -I" I only see the packages which I manually installed. Still a noob, don't take anything I say as the truth :).
<reepca>"The packages field lists packages that will be globally visible on the system, for all user accounts - i.e., in every user's PATH environment variable - in addition to the per-user profiles."
<reepca>so I guess I can make my user's packages available to him, but I can't prevent them from being available to anyone else... if, for whatever reason, I wanted to do that.
<apteryx>What about those "user profiles", maybe you could define the packages installed only for your account in one of those?
<apteryx>And keep the config.scm as what it seems to be intended for, i.e., the system's baseline configuration.
<reepca>I'm pretty sure the "user profiles" referred to are the .guix-profile directories in the home folders, though I could be wrong
<apteryx>Right. I see a manifest file which seems to be the list of the packages "visible" for my user.
<reepca>huh, I guess I never really looked closely. I need to get some sleep, sorry, but you've given me some good leads on accomplishing what I'm trying
<apteryx>I read in the manual somewhere that you can define a list of packages in a scheme file, which can be passed to guix package to be installed.
<apteryx>You could maintain such a file for your personal packages.
<apteryx>reepca: OK. I'm very new to this, but I'm sure after we've had more time to read the manual it will be clearer!
<apteryx>Good night.
<efraim>csanchezdll: the b9b commit didn't work for me, was I supposed to apply the patches from the mailing list first?
***jonsger1 is now known as jonsger
<csanchezdll>efraim: no, bub b9b is the first bad
<csanchezdll>try its parent
<rekado>having played with d3 and the graph for ghc-pandoc for a while now I think a force-directed graph is not the right visualisation for dependencies
<rekado>it's one big hairball
<rekado>maybe a chord diagram would be more instructive
<efraim>csanchezdll: It built in about 20 minutes, I got the same result as you, with 9b9, the expat patch, being the first commit where it fails to build the bootstrap binaries
<csanchezdll>efraim: I think I found the problem
<csanchezdll>I am writing to the list
<csanchezdll>but it is getting complex with GUIX internals, so I think I will wait for input from Ludo or other hardcore GUIX-er before trying to fix it
<csanchezdll>in the meanwhile, it probably builds if you try with --no-grafts
<csanchezdll>(I haven't tried yet myself)
<efraim>I just checked out my aarch branch, reverted my make-boot0 hack, reverted the expat fix, and called guix build -n to see what the list looks like
<efraim>all of this has me drawing on paper, trying to figure out what git-bisect might look like if it were really git-trisect and there was a slave node for additional computation
<csanchezdll>efraim: sent. I am learning guile along the way so I am no 100% sure of what is happening
<csanchezdll>but I am pretty sure it is something like what I say on the mail
<efraim>grafts has always been a bit of magic for me
<csanchezdll>yup, also for me...
<bavier>almost done packaging clamav
<davexunit>rekado: wow!
<rekado>chord diagrams for the dependencies of ghc-pandoc and r, respectively.
<ng0>am I the only one who tried to reconfigure a system today? even a make clean did not fix the broken slim
<rekado>hover on the segments to only show chords for this package
<bavier>ng0: oh no
<davexunit>rekado: this is awesome
<ng0>ok :)
<bavier>ng0: what's the issue specifically? I built several vms before pushing my changes to slim
<ng0> <- this is all i got
<ng0>"malformed patch"
<bavier>ugh, sorry about that. I did modify one of the patches, but emacs usually adjusts the line numbers properly
<ng0>no problem
<bavier>I can't fix it myself right now
<rekado>I just need to clean up the JavaScript code and figure out what to do about the d3 dependency (do we bundle the JS file or load it from Otherwise it’s ready as a graph backend.
<rekado>generated with: ./pre-inst-env guix graph --backend=d3js r > graph2.html
<bavier>correcting the chunk line numbers would fix it though
<davexunit>rekado: I think bundling is the only real option so that we provide full corresponding source
<ng0>bavier: i haven't really looked at it as I was busy with other stuff. setting up a system, dentist appointment, annoying pain in tooth going on until wednesday, etc :)
<bavier>ng0: it's been hectic on my end too, lately. I probably shouldn't push commits in a hurry like that :(
<ng0>someone would've noticed my emails eventually.. <24 hours is okay :)
<ng0>it's not like you broke grub
<ng0>easy fix
<apteryx>Would someone know how fullscreen features are handle by tiling WM such as ratpoison?
<apteryx>(if they are handled at all)
<ng0>awesome-wm, default config: [meta-4] [f]
<apteryx>Suppose I watch a youtube video in a ratpoison frame using my right half of the screen; I press the fullscreen icon... What is supposed to happen?
<ng0>i don't know ratpoison
<bavier>apteryx: on ratpoison, e.g. "fullscreen" icecat will occupy only the portion of the screen ratpoison gives it.
<apteryx>ng0: OK, good to know. I'll be trying awesome later.
<ng0>awesome-wm will fullscreen the one your mouse is in
<bavier>i.e. it'll stay in the frame it's in
<apteryx>bavier: OK. It seems to be broken for me. In IceCat for example, attempting to make videos fullscreen make icecat unresponsive, I have to kill it.
<ng0>and when you're in this fullscreen (or even just a fullscreen video without this) you can press the key combo to reach another screen and you get there
<bavier>apteryx: I see, I haven't tried fullscreen video from icecat before
<ng0>it works for me here
<apteryx>ng0: OK. So this works in awesome-wm, and probably should work in ratpoison (but only inside the frame the window is), but not for me, using icecat at least. Will try another app.
<apteryx>Is there a canonical way to install gcc so that it can build C? Right now I can an "gcc: error trying to exec 'as': execvp: No such file or directory" error.
<bavier>apteryx: best to install the 'gcc-toolchain' package
<apteryx>OK, it's passing that now. Thanks :)
<apteryx>Hmm... Is there an equivalent of ldconfig that I can use to see if the libX11 library I just installed is "found"?
<apteryx>Ah ha! I had to use -lX11 rather than -lx11.
<apteryx>The libs headers are installed under .guix-profile/include
<apteryx>And are found because of this env var: C_INCLUDE_PATH=/home/maxim/.guix-profile/include
<apteryx>woohoo! Compiled sloppy.c & ran it. Now I have mouse focus on ratpoison's frames. RSI saver.
<apteryx>Now suppose I want to make this sloppy binary available to my path, what is preferable between copying the binary to my .guix-profile/bin vs creating my own, user owned bin location, say ~/.bin and adding this to my PATH?
<bavier>apteryx: your .guix-profile is a symlink into the store, so should be considered read-only
<bavier>apteryx: you could also make a guix package for it ;)
<apteryx>bavier: Hehe, that'd be an interesting experiment. Maybe the binary should just be built along with the ratpoison package we have, since the sloppy.c is bundled with it (but not built).
<bavier>has anyone looked at patching our linux-libre for CVE-2016-5195?
<ng0>wasn't this implied with the upstream release of the version which was announced after that cve?
<ng0>it was the cow cve, right?
<bavier>right. idk, I'm not subscribed, and I don't see any announcement in the archives
<ng0>upstream linux said something.8.3 fixed this, so I assumed something.8.3 of linux-libre applied this.. I mightbe wrong
<bavier>ok, makes sense
<paroneayea>I wonder what the global dns attacks look like to people who have no idea what DNS even is
<paroneayea>bavier: since the patch to slim doesn't work right now I'm finding I can't do a system reconfigure. Ok with you if I try to either fix the patch to apply or comment it out for now?
<bavier>paroneayea: yeah, sorry about that. fixing the patch would be great. just the line numbers need fixing
<bavier>I trusted emacs too much when I edited the patches in place...
<paroneayea>bavier: it happens
<paroneayea>bavier: cool, fixed it.
<paroneayea>fix pushed.
<paroneayea>ng0: you should be able to reconfigure now fwiw
<zloster>bavier: the 4.8.3 kernel release contains the commit that fixes CVE-2016-5195. This is the changelog: And in Guix.git there is this commit: which bumps the libre-kernel version from 4.8.2 to 4.8.3. So 'guix pull' and 'guix system reconfigure' should do the work.
<zloster>I have question about i3 window manager - it seems that i3 dmenu functionality (either the right click or the Alt+D combo) is not working after fresh install of GuixSD 0.11. Also the i3 status bar was not working but after some upgrades it started working and shows the various indicators
<ng0>you need the i3menu something i think.. separately
<ng0>paroneayea: fixed, yep
<zloster>ng0: how to install this i3menu - I don't see such thing in the packages at
<zloster>ng0: there are i3-wm 4.12 and i3status 2.10
<ng0>i don't know... last time i used i3wm was on gentoo and also there 5 years ago
<ng0>oh, dmenu functionality
<ng0>did you install dmenu?
<ng0>on second note.. don't listen to me. I have no idea about i3wm on guix :/
<ng0>this is not a first problem.. i've read this before
<ng0>bavier: i like the "Logging in" message.. makes it more clear to new users that slim did not freeze
<zloster>ng0: any pointers to where you have read it? I'm not able to find
<ng0>i rebooted
<ng0>what do you mean?
<zloster>ng0: about the i3 dmenu problem
<zloster>you said that you've read this before
<ng0>what is "it" then? I just said I think I read someone else talk about it either on one of the mailing lists or here on irc.
<zloster>ng0: i was asking if you can give me to that place with the description of the problem. But I think that I don't need it: there is a dmenu package in Guix. I'll check if it is installed and if not will install it to see if it helps
<ng0>ok.. like i said, I have no recollection of i3wm and what it is now, especially how it behaves on guix
<ng0>i use awesome
<ng0>there's dmenu, interrobang and some others I can use for a menu in awesome. well depending on the definition of menu. right now I'm so distracted by painkillers I don't even know for sure what dmenu does. I just launch applications via text, not a graphics/dialogue menu
<zloster_>нг0: installing dmenu package helped and now the $mod+d starts the menu. Thank you
<zloster_>ng0: thank you
<paroneayea>why do you taunt me so
<wingo>we need a nice guix environment --container CLI that maps /usr/bin/env
<bavier>ng0: great!
<wingo>probably not much work but yarrrr
<paroneayea>wingo: that would be nice
<wingo>i think there are existing incantations that can work
<bavier>paroneayea: thanks for the quick fix
<wingo>but --container is automatically --pure so you need to have all the dependencies of what you need at hand
<paroneayea>bavier: yup
<paroneayea>ACTION currently looking at building a new version of coreboot for laptop
<paroneayea>#!/usr/bin/env everywhere :(
<ng0>here, there, everywhere :(
<paroneayea>wingo: I'm going to propose this to the list
<ng0>it would be nice to not have to deal with this, yes
<bavier>could maybe make the "patch-shebangs" function into a cli utility
<wingo>i think there is agreement that some sort of container /usr/bin/env thing is the right so'n
<ng0>is there really an improvement in speed when you build/search/write with SSD instead of HDD?
<paroneayea>wingo: posted to the list. thanks for the idea.
<bavier>depends on the access patterns
<paroneayea>bavier: it's not a great idea if working with upstream code
<paroneayea>consider running patch-shebangs from a git checkout of something you're trying to get a patch into
<paroneayea>bam, suddenly you dirty up your checkout with a bunch of replaced paths.
<bavier>paroneayea: yeah, that'd be messy
<ng0>i will spend some more time cycling between gentoo and guixsd.. I need to understand and apply some things for hardening in general for guix, uclibc-ng etc.
<ng0>it'd be nice to have some cooperation etc as there's some knowledge in the hardening project, but i'll just try to dig into it on my own for this is purely out of self interest
<lfam>What a pain in the ass Go is:
<bavier>I didn't know closing a bug made it immune to further comment
<lfam>Socially, it's immune to further comment. So I have to go through the annoying steps of opening a new bug
<lfam>A new bug report for the exact same bug
<lfam>I'm just ranting, I know I don't have a real complaint :)
<reepca>It seems to be related to that bug, since it happens when applying the fix, but it's not the bug itself manifesting once more, is it?
<lfam>It's the exact same bug, but manifesting for a different version of the compiler. The fix they suggested for the current version does not work for the old bootstrap compiler version
<lfam>That's what I was trying to communicate
<lfam>I wonder how long they will support the bootstrap compiler. Probably not very long if they don't test bug fixes against it
<bavier>the patch fix doesn't touch the missing file though, right?
<reepca>if fmt.h has anything to do with printing time/date I would imagine it would
<bavier>sure, go does dependency tracking I suppose
<roptat>hi, I'm trying to use gpg to generate a key, but it fails with "gpg: agent_genkey failed: No pinentry" although I installed pinentry and pinentry-tty
<lfam>I assume the code used in the patch requires a newer Go
<lfam>roptat: Put this in your ~/.gnupg/gpg-agent.conf: 1 pinentry-program /home/leo/.guix-profile/bin/pinentry
<lfam>Hm, not the '1' character
<lfam>Something like that, anyways :)
<lfam>Now there are two Golang bug tickets with the same name:
<ng0>sometimes you just have to accept that upstream is not always doing "sane" decision etc..
<lfam>Perhaps my original question was unclear. I have to swallow my pride and be nice when communicating with upstream maintainers :)
<reepca>I guess we could always ask around at #go-nuts (#golang redirects there)?
<lfam>Go IRC is hit or miss. I prefer to use their bug report system. That way, other people can see the bug report and (hopefully) the resolution.
<lfam>I already found some answers in their bug database for other problems.
<lfam>What about upstream's question here?
<lfam>"Is stat expected to return complete information in your environment or do you fake the modtimes?"
<roptat>lfam, it worked \\o/
<lfam>Cool :)
<bavier>lfam: they might be wondering if we use something like libfaketime, like debian sometimes does
<lfam>My understanding is that we don't do anything special to time in the build chroot
<bavier>no, we don't
<lfam>I was building on btrfs, retrying on ext4
<lfam>I wish I had some more ideas for them in that bug report. It's going to be annoying if they can't reproduce it at all.
<lfam>Hm, my x200s can't do `guix build $(guix refresh -l python-pytest python2-pytest)` without becoming totally unresponsive. Too bad
<lfam>When I save the failed Go build environment and stat the file in question, it's mtime is 1969-12-31 19:00:00.000000000 -0500
<lfam>But the atime and ctime are current
<lfam>I think the problem is caused by unpacking and repacking the tarball when applying the patch. When I remove the patch and use a substitution instead, the mtime remains post-epoch and the test passes
<lfam>So, I'll do that
<lfam>Is it expected that applying patches will set file mtimes to epoch?
<lfam>Yes, it happens at guix/packages.scm:570, if I understand correctly