IRC channel logs

2019-08-25.log

back to list of logs

<joshuaBPMan>I'm thinking about purchasing a Lenovo T500. They are super cheap on ebay!
<joshuaBPMan>And you can install a second hard drive!?
<joshuaBPMan>How about RAID on a laptop! That's awesome!
<mcmx>I don't think you can, but yea, I got this thing for under 100 bucks
<mcmx>kmicu: it's from 2008, like I said
<joshuaBPMan>I'm ok with that.
<mcmx>one thing though, is the builtin wifi card requires proprietary software I think, had to use my usb wifi dongle when I was trying to install Guix System
***amfl_ is now known as amfl
<Minall>quiliro: Lo siento que me desconecté de esa manera...
<Minall>quiliro: Cerré emacs por accidente
<jgibbons[m]>I'm working on packaging CV Assistant. Is there a quick way to customize the install directory with qmake?
<jgibbons[m]>It looks like the .pro configures qmake, and it has a hard-coded install directory.
<jgibbons[m]>the Makefile has an INSTALL_ROOT variable, but I don't like it including the usr directory. Not good for guix.
<csprng>jgibbons: does something like this work? (invoke "qmake" (string-append "PREFIX=" (assoc-ref outputs "out")))
<csprng>alternatively: (substitute* "Makefile" (("/usr/local") (assoc-ref outputs "out")))
<efraim>I suggest PREFIX= it's what we end up using on most qt packages
<ngz>Is there a reference somewhere about the filters one can use to search through ci.guix.gnu.org?
<rekado>ngz: there isn’t, but I can tell you.
<rekado>ngz: ‘system:’ and ‘spec:’ work, and you can use ‘^’ and ‘$’ for matching the beginning and end of a string in the text query.
<rekado>this is implemented in the query->bind-arguments procedure in src/cuirass/database.scm
<ngz>rekado: ok, thank you. Would it be useful to add some hints about it in the first page, much like what you did in issues.guix.info?
<kmicu>mcmx: I didn’t have your previous messages in backlog. In that case you can install manually and it will work for sure.
<efraim>the crate importer is working less than normal today :/
<ngz>efraim: It noticed yesterday that it couldn't import some packages from crates.io, too. Not sure it is related.
<ngz>E.g., impossible to import "humantime" or "rayon".
<ngz>I admit I don't know how to debug efficiently Guile (and therefore Guix) code. I'm spoiled with Elisp, which is very straightforward to debug.
<pen14641>Hi, I am trying to run a disk image built with 'guix system disk-image' using virt-manager. I get the error message "Unable to read from '/sys/fs/cgroup/unified/machine/cgroup.controllers': No such file or directory". There is no directory /sys/fs/cgroup/unified/machine, but there is a file /sys/fs/cgroup/unified/cgroup.controllers. I don't know if virt-manager is trying to access the file at the wrong
<pen14641>location..?
<pen14641>Or well, the error message is "Unable to complete install: " and then that error message
<jgibbons[m]>csprng: changing the prefix did not work. I got it working by changing CVAssistant.pro. Only problem now is the icon won't show in gnome, otherwise it's good.
<jgibbons[m]>And my laptop's wireless card isn't working now :(
<efraim>jgibbons[m]: depending on which qt modules you used you'll need to wrap the binary to see the icons
<ecbrown>i always find it useful to have a benchmarking application on the various systems that i go to: would like to invite everyone to checkout `sysbench' on my channel https://github.com/ecbrown/my-guix
<bandali>neat
*ecbrown contemplates whether the next task is to go "--with-courage" (hurd) or to try to roll a system where gz, etc. is replaced with pigz, etc.
<jgibbons[m]>efraim: it uses qt5. It's only the desktop icon that isn't showing up, so I think it's a problem with the .desktop file.
<rekado>ngz: I would be happy if the filter/search options would be mentioned somewhere sensible in the UI of ci.guix.gnu.org.
<rekado>ngz: I had to think for too long about where to put the information, so I punted on it.
<jgibbons[m]>Is the hurd available to try with guix on a vm?
<rekado>jgibbons[m]: sadly no.
<ecbrown>jgibbons[m]: i'm going to try to build one
<rekado>jgibbons[m]: I also found a couple of problems with Guix on the Hurd.
<rekado>our statically linked binaries segfault
*ecbrown listens intently
<rekado>‘tar’, for example, just crashes, which makes using Guix on Hurd impossible right now.
<rekado>it’s possible that this is just a problem with my bootstrap binaries, of course, but nothing has changed in that regard recently
<rekado>so I suspect the issue still exists
<ngz>rekado: As in issues.guix.info (i.e., right below the search bar, which is centered full-width) is not an acceptable location?
<rekado>jgibbons[m], ecbrown, if you want to play with this I recommend getting a Debian GNU/Hurd VM image and building Guix on it.
<ecbrown>my plan was to try to build it with debian hurd 2017, with the commit right before guile 2.2
<rekado>(that’s what I did, anyway)
<jgibbons[m]>Is the Hurd on any completely free distros?
<rekado>remember to resize the VM image before doing any serious work.
<ecbrown>naively, i tried with debian hurd 2019, and am having a hell of a time with guile-gnutls
<ngz>Or, a [?] right to the search bar, which opens a pop-up with the information (less invasive)
<rekado>resizing the file system cannot easily be done later, so better do this before you start.
<ecbrown>i have 100G provisioned now
<vagrantc>ecbrown: yeah, guile-gnutls doesn't exist in recent versions of debian
<rekado>ngz: on issues.guix.gnu.org we have that really big input field, so there’s space.
<rekado>ngz: but on ci.guix.gnu.org we don’t have that.
<rekado>so I think a little help link in or next to the search bar would be a good place
<ngz>Sure, but the first solution implied to move the search bar
<rekado>moving it where?
<ngz>Though I agree the second solution, or related, would be better
<rekado>oh, just on the main page?
<ngz>(moving right above the tabulated output)
<rekado>I see.
<rekado>I guess that would work, too.
<rekado>I’d have to see what it looks like
<rekado>my fear is that moving the search input out of the top bar would make it seem more cluttered.
<rekado>don’t know.
<ngz>Ultimately, it would be nice to have both issues.guix.info and ci.guix.gnu.org use the same interface
<rekado>we’re still far from Github style clutter (with these many bars at different levels)
<rekado>ngz: I already tried to make them more similar.
<Minall>Hello guix!
<ngz>of course. I'm juste talking about the search help
<ngz>just*
<Minall>quiliro: Kiel vi fartas... Lo siento que estoy tarde, tuve que salir..
<rekado>but that was just really a bunch of superficial changes
<rekado>I’d be happy to see some convergence between the two.
<rekado>ngz: if you want to work on Cuirass: more power to you! :) Please keep me on Cc for Cuirass patches, and I’ll try to review them.
<ngz>Therefore, my vote, for what it is worth, would be to have both use [?] to the right.
<rekado>ngz: sounds good.
<rekado>for issues.* I think it still makes sense to have a big search input on the main page, though, because there isn’t really anything else to do there but to search for issues.
<ngz>rekado: I'm not sure I would be able to do that =/ I'm just suggesting stuff for the time being.
<rekado>ngz: suggestions like yours are half the work, really. Often times I don’t really know what to focus on, so hearing a good plan is often enough to solve a real problem quickly.
*rekado is currently stuck on texlive-union again
<rekado>TeX is really confusing.
<rekado>I solved one problem (regenerating font maps), but just the existence of the correct font maps breaks a completely unrelated feature (searching for *.cls files).
<rekado>I’ve been thinking about this for a few days now, but no idea has struck me yet.
<Minall>quiliro: Debo de salir de nuevo, disculpe, no pense que tendría que salir hoy..
<Minall>Ya vuelvo...
<xavierm02>Where can I find the thing that iters over phases of a build system to run them?
<ngz>xavierm02: See gnu-build in gnu-build-system.scm for example.
<ngz>I think most build systems ultimately call gnu-build with a possibly augmented #:phases argument.
<ngz>s/augmented/altered
<xavierm02>ngz: Thanks!
<xavierm02>gnz: I couldn't find where phases are iterated over in the gnu-build file. Also, it seems to call build-expression->derivation but when I look up the definition of build-expression->derivation, there's a comment saying it's depreciated.
<xavierm02>ngz: ^
<ngz>xavierm02: At the very end of "gnu-build-system.scm", you have (define* (gnu-build #:key ...))
<ngz>In this gnu-build function, you can see (every (match-lambda ((name . proc) ...)) phases)
<ngz>This every is iterating over the argument `phases'
<ngz>I'm not sure the `every' is necessary, btw, since every call to (apply proc args) is expected to either return #t or raise an exception.
<xavierm02>Oh. I was in build-system/gnu.scm. Thanks again :-)
<ngz>Backward-compatibility stuff, probably. Maybe it would be better to add a `raise' inside the `unless' and use `for-each' instead.
<mcmx>are you guys planning on making python -> python3 any time soon?
<mcmx>like arch linux
<ngz>BTW, is there some documentation somewhere about how to debug Guix internals, ideally through Emacs?
<mcmx>kmicu: I don't have enough skillz to make an encrypted MBR partition table
<rekado>mcmx: I don’t know what you mean. The ‘python’ package installs ‘python3’.
<rekado>If you want the ‘python’ executable to be ‘python3’ you need to install the ‘python-wrapper’ package.
<rekado>Upstream doesn’t install ’python3’ as ‘python’, so we don’t either.
<quiliro>i get the following error every time I enter a new chatroom on Emacs ERC:
<quiliro>Error enabling Flyspell mode:
<quiliro>(Searching for program No existe el fichero o el directorio ispell)
<quiliro>Starting new Ispell process ispell with default dictionary...
<quiliro>No existe el fichero o el directorio = The file or directory does not exist
<rekado>you probably don’t have ‘ispell’
<nly>flyspell requires you to install one of several ispell programs(hunspell, ispell ..)
<rekado>you can install ’aspell’
<rekado>and then configure flyspell to use aspell
<vagrantc>do builds under guix have a writeable /tmp dir? diffoscope 121 is failing to build as an updated test tries to write out a file in the tests directory
<rekado>or you coudl disable flyspell on ERC.
<rekado>vagrantc: /tmp is writable
<rekado>vagrantc: is this from a git checkout?
<vagrantc>rekado: yes
<rekado>by default files taken from a git checkout are not writable, so you need to use make-file-writable on the directory first.
<vagrantc>could switch back to tarballs from pypi, but those aren;t consistantly released
<vagrantc>i could make that specific file/directory writeable?
<rekado>we use make-file-writable in a couple of packages.
<rekado>yes
<vagrantc>thanks!
<vagrantc>that's much more elegant than the hacks i was thinking of :)
<nly>i cannot ./bootstrap on guix checkout
<quiliro>rekado: ty :-)
<nly>i mean ./bootstrap results in an error
<rekado>nly: what error?
<nly>sorry, I'm in bed now. i'll ask again tomorrow
<nly>and provide a log.
<vagrantc>does it make sense to reference the upstream issue that required writeable access to a tests file in a comment?
<vagrantc>trying to briefly summarize the issue that lead to diffoscope needing the writeable file
<rekado>vagrantc: I think that’s really a Guix quirk, not an upstream problem.
<rekado>it would be sufficient to state in a comment that the tests directory needs to be writeable.
<vagrantc>ok ... so far just a single file needs to be writeable
<vagrantc>thanks
<vagrantc>if another file requires write access at some point, i can just make the whole directory writeable, i gues
***MinceR_ is now known as MinceR
<raghavgururajan>jgibbons[m] Just read your email. Thank you so much for your hard work. Have the package been added to master branch?
<bluekeys>Hi guix. Sometimes mye headphone socket doesn't work. Where should I start looking to troubleshoot it?
<rekado>bluekeys: check in alsamixer that it isn’t muted.
<quiliro>bluekeys: first check for hardware
<bluekeys>I don't seem to have alsamixer, do I have to have it?
<nckx>bluekeys: Install the alsa-utils package. You only have to have it if you want to use it :)
*civodul fixed Ghostscript: https://issues.guix.gnu.org/issue/34877 \o/
*vagrantc is eager to see qemu 4.x from core-updates
<Minall>Hello guix!
<Minall>quiliro: Kiel vi fartas!