IRC channel logs

2019-10-22.log

back to list of logs

<jlicht>how hard would it be to have a netboot-compatible guix system image?
***18WAAOVH8 is now known as cap
***cap is now known as 18WAAOVH8
***18WAAOVH8 is now known as cap
***cap is now known as 18WAAOVH8
***18WAAOVH8 is now known as cap
***cap is now known as 18WAAOVH8
***18WAAOVH8 is now known as cap
***cap is now known as 18WAAOVH8
***18WAAOVH8 is now known as cap
***cap is now known as 18WAAOVH8
***18WAAOVH8 is now known as cap
***cap is now known as 18WAAOVH8
***18WAAOVH8 is now known as cap
***cap is now known as 18WAAOVH8
<peanutbutterandc>How do I figure out the sha256 of a certain commit of a git repo? `guix download url_to_the_commit` did give me a sha256, but the build fails because of hash sum mismatch. Any ideas?
***18WAAOVH8 is now known as cap
***cap is now known as 18WAAOVH8
***18WAAOVH8 is now known as cap
***cap is now known as 18WAAOVH8
***18WAAOVH8 is now known as cap
***cap is now known as 18WAAOVH8
***ChanServ sets mode: +o nckx
***18WAAOVH8 was kicked by nckx (Please fix the script changing your nick every 10m)
<montokapro>When applying a patch to a existing package with only partial functionality, should we append "guix-1" or some variant onto the version field?
<sneek>montokapro, you have 1 message.
<sneek>montokapro, marusich says: re: https://pastebin.com/p0b9xKLB it failed because your guix-daemon wasn't using MIPS emulation - enable it with qemu-binfmt-service-type: https://guix.gnu.org/manual/en/html_node/Submitting-Patches.html#Submitting-Patches
<nckx>montokapro: No.
<montokapro>Thank you :)
<nckx>:)
<kts>Install icecat via guix via suse. Where to launch? Should I need to create a .desktop file, if so where?
<bdju> https://paste.debian.net/plain/1109038 getting these errors trying to use tesseract-ocr. can anyone assist me?
<bdju>I don't think it told me to export anything at install time. I guessed at what to put in .profile but the errors didn't change
<bdju>kts: if installed in the user profile the binary should be at ~/.guix-profile/bin/icecat
<atw>apteryx: didn't follow through on what I said yesterday, I'll do it tomorrow
<sneek>Welcome back atw, you have 2 messages.
<sneek>atw, nckx says: Registration requirement was to discourage the you-know-who after the you-know-what earlier this month. I've disabled it again.
<sneek>atw, nckx says: (It had already been disabled for the past year or so without incident.)
<atw>oh ok, it got taken care of. Thanks nckx! Though I fail to understand the insinuation
<nckx>atw: There still room under that there rock? 🙂
<nckx>atw: The blog post…
<atw>oh, right.
***catonano_ is now known as catonano
<nckx>atw: So ‘follow through’ above was related?
<atw>yes, I was going to propose removing the registration requirement again when Ludo and others would be around
<nckx>Thank you.
<kts>icecat, fonts all tofus, installed google-noto-fonts,
***ChanServ sets mode: +b nobuddy!*@*
***nobuddy was kicked by ChanServ (Banned: Testing :))
<kts>icecat, Gtk-Message: 00:54:37.882: Failed to load module "unity-gtk-module"
<kts>Gtk-Message: 00:54:37.882: Failed to load module "canberra-gtk-module"
<kts>font_face status is: file not found
<kts>icecat
<kts>guix on suse
<kts>00:54:39.621: Could not load a pixbuf from icon theme.
***ChanServ sets mode: -o nckx
***ChanServ sets mode: -b nobuddy!*@*
<htsr>Hi! Is it possible to replace the shepherd package with a channel?
<atw>kts: have you done fc-cache -f yet? (https://guix.gnu.org/manual/en/html_node/Application-Setup.html )
<kts>atw: Gtk-Message: 00:54:39.804: Failed to load module "unity-gtk-module"
<kts>Why do I not have "$HOME/.config/guix/current/bin:$PATH"?
<kts>Even after invoking "guix package pull".
<kts>Sorry. I have that. "$HOME/.config/guix/(~)current/bin:$PATH" and not "$HOME/.config/guix/current/bin:$PATH".
<kts>Is that normal?
<atw>sorry kts, I don't follow. What is that string?
<kts>Not "$HOME/.config/guix/current/bin:$PATH" as described in the manual but "$HOME/.config/guix/(~)current/bin:$PATH" with ~ before current.
<kts>The $PATH environment to export.
<htsr>you mean this is your $PATH variable?
<kts>"$HOME/.config/guix/current/bin:$PATH" doesn't exist. "$HOME/.config/guix/(~)current/bin:$PATH" exists.
<kts>~current
<kts>~current is there.
<kts>With additional ~. Why? "guix pull" created that dir.
<kts>Should I export that, anyway?
<kts>htsr: atw Excuse me. Do you have "$HOME/.config/guix/current/bin:$PATH"?
<atw>I'm afraid I couldn't say whether or not that's normal, going to bow out of this :/
<kts>Wait. Do you have that?
<kts>"$HOME/.config/guix/current/bin:$PATH"
<kts>Or this, "$HOME/.config/guix/(~)current/bin:$PATH"?
<kts>atw: Done fc-cache -f. Am in the middle of "guix package -u".
<atw>my PATH on Guix System looks ...:/home/alex/.config/guix/current/bin:... , but if you say that guix pull put the tilde in there, I am hesitant to say that it is wrong
<kts>I love scheme. atw
<kts>mit-scheme
<htsr>kts: so inside ~/home/.config/guix/ you have a directory named (~)current?
<kts>htsr: yep
<kts>htsr: Just ~current, not literal (~)current.
<peanutbutterandc>Hey everyone; can anyone please give me the correct way to get the sha256 of a GIT repo's commit? I tried (sha256 (base32 url_to_commit)) but it didn't work. I am packaging a particular branch of a certain git repository for personal use.
<htsr>kts: I can't help you :/
<atw>peanutbutterandc: what I usually do is clone the repo of the thing I'm packaging locally, cd into it, and guix hash -xr
<nckx>peanutbutterandc: guix hash -rx (See info (guix)Invoking guix hash) in a *clean* checkout.
<nckx>atw: :)
<peanutbutterandc>atw, nckx - That makes a lot of sense. Thank you very much! I will try it again now. (:
<atw>jinx, nckx :P
<kts>htsr: It doesn't really matter. Thanks a lot for all the replying.
<atw>kts: the original issue is still the tofu?
<kts>atw: Yep. Am in the process of "guix package -u" and "guix package --install gs-fonts font-dejavu font-gnu-freefont-ttf".
<nckx>peanutbutterandc: The word ‘branch’ spiders my tingle. Do you really mean to track a moving branch? Because that's not really supported.
<nckx>s/branch/commit/ and all's well tho'.
<peanutbutterandc>nckx - I just plan on using the current latest commit of a feature branch
<peanutbutterandc>Any sensible ideas on what I should set the `version` to? I don't like the commit sha. Perhaps just a string that says 'latestcommit' or something?
<atw>kts: ok, I'd be curious to know if that works, keeping in mind that refreshing the font cache with fc-cache -f should be done after installing fonts
<kts>atw: https://p.teknik.io/6NUAB Perhaps, can some of those be ignored?
<apteryx>atw: seems all is well, thanks to nckx :-)
<nckx>peanutbutterandc: In Guix proper we use <latest real release || 0.0.0>-<Guix revision>-<truncated git hash>, all taken care of by the (git-version …) helper. Examples abound in gnu/packages/*.scm.
<kts>I was dumb to invoke "guix gc" just to free some 500 mb and satisfy my autism considering / is 70 % free.
<nckx>It doesn't matter what you use privately, it could be 0, 1, 2… or names of your favourite sausages.
<nckx>peanutbutterandc: ‘Guix revision’ is just an increasing integer for sorting purposes, since git hashes don't sort.
<nckx>apteryx: Check your inboxen :)
<atw>kts: that icecat startup looks roughly like mine. Is the tofu gone?
<nckx>kts: It's as good a reason as any. Also note that ‘guix gc’ alone won't actually delete any GC roots; ‘guix gc -d …’ will and might clear significantly more space on longer-lived systems.
*nckx often runs guix gc just for that squeaky clean feel.
*atw is the Oscar the Grouch of guix gc
<kts>hehe
<apteryx>nckx: ack :-)
<kts>Excuse me. What DEs are available under the "GNU Guix System 1.0.1"?
<nckx>kts: Er, GNOME, XFCE
<kts>Er?
<nckx>Not really a DE person so those are all I know.
<atw> https://guix.gnu.org/manual/en/html_node/Limitations.html#Limitations might help
<nckx>‘Er’, as in American ‘uh’. 🙂
<kts>Considering WMs, is EXWM available?
<atw>pretty sure
<nckx>kts: Yes!
<peanutbutterandc>nckx, I see. Thank you. I will take a look at the examples too.
<apteryx>nckx: I've tried for example: /QUERY ChanServ AKICK #guix LIST, but ChanServ says I'm "not authorized to perform this operation". Ideas?
<nckx>apteryx: I probably just have to throw more flags at ya. Let me check.
<peanutbutterandc>nckx, atw: It seems to be `guix hash -rx .`
<nckx>Oops indeedy.
***ChanServ sets mode: +b *!*@dsl-149-171.b2b2c.ca
***apteryx was kicked by ChanServ (User is banned from this channel)
<nckx>lol
***ChanServ sets mode: +o nckx
***nckx sets mode: -b apteryx!*@*
<peanutbutterandc>So, it seems that I am supposed to pass dependencies as (input ...). So, do I first skim through `guix package --search=dep` and find all the relevant ones and then add them to my .scm package definition?
<kts>nckx: Can I know your WM?
***nckx sets mode: -b *!*@dsl-149-171.b2b2c.ca
<nckx>kts: i3
<nckx>apteryx: o/
<apteryx>oi!
<kts>nckx: ranger or mc?
<nckx>TBH I just don't feel the need for a file browser.
<apteryx>this is a good ritual for new mods -- get banned for some time to get a taste of it ;-)
<nckx>I've used ranger… twice or so the past year.
<nckx>apteryx: I hoped you've learnt your lesson!!!
<apteryx>haha, thanks for the assistance nckx
<nckx>apteryx: (See msgs.)
<kts>Are there non-free guix repo
<kts>?
<kts>There may not be virtualbox. Right?
<nckx>kts: I honestly don't know, but we don't support (or advocate for) non-free software in official Guix channels.
<peanutbutterandc>What is the standard procedure to add (input ...)-s? Anyone?
<nckx>peanutbutterandc: More or less what you described: find out the guix package name, then add ("that" ,that) to inputs.
***ChanServ sets mode: -o nckx
<peanutbutterandc>nckx, I see. Also, another question: seeing that a guix package can have multiple outputs, it seems I could somehow make a single package of all the branches that I am interested in. However, i seems that, at least in this one example: https://guix.gnu.org/blog/2018/a-packaging-tutorial-for-guix/ (the extended example), the inputs are defined only once. Now, if the branches are using separate extra dependencies, how does things work? Or, could you
<peanutbutterandc>guide me to the relevant section in the manual please? I don't remember seeing anything about this there.
<peanutbutterandc>Another question just out of curiosity: once HURD is done, could a user switch from linux to hurd just so on a GUIXSD machine? Like we switch from real-time kernel to normal kernel currently (or between kernel versions).
<nckx>peanutbutterandc: Last question: that would be the intention (modifying details like ‘initrd-modules’ of course, but even that could be abstracted). You already can switch kernels quite easily. If you define a real-time kernel package variant you can simply add ‘(kernel linux-libre-real-time)’, reconfigure, reboot, and everything will work.
<peanutbutterandc>nckx, I see. Guix-SD will start shipping HURD versions like Debian/HURD, before a stable release of HURD, I presume?
<nckx>peanutbutterandc: First question: Outputs are only that, separate output directories. They are usually split only during of after the ‘install’ phase, they share the same build environment and inputs. So you'd simply add *all* inputs. However, outputs might not be the answer here. Sounds like what you want is multiple different -branch-foo packages that inherit from master or whatever.
<nckx>peanutbutterandc: I can't say: we don't have a HURD volunteer at the moment. ☹
<nckx>Guix won't start shipping anything, any time, unless that changes.
<peanutbutterandc>nckx, Ah. That certainly sounds a lot more elegant! I'm really new to this: both guix and guile. But I can already see that this is something the average user could really benefit from. I can package my own stuffs and then share with my friends with just a GitHub account, and a bit of guile. This is super cool!
<peanutbutterandc>nckx, I see. I wish I could help. But that would probably be a dozen books/reference-manuals away, at least.
<nckx>In fact, Guix/HURD is broken at the moment. A few people have had a poke at fixing it but I don't think there's been any meaningful movement.
<nckx>peanutbutterandc: I have to go now, but look into (inherit …). It's probably what you want here.
<peanutbutterandc>nckx, I see. Thank you very much for your time and the explanations. I will look into it. (:
<nckx>peanutbutterandc: ‘Average user’: We're still far from that, IMO, too many rough edges, but I share you optimism and thank you for the positive energy 🙂
<nckx>o/
<peanutbutterandc>nckx, :) [Well I consider me to be an average user. And people like you - the contributors - to be the above-average users. So yeah. Haha :D ] :)
<kts>US, Alt, Intl Layout or US, Alt, Dead Keys Layout?
<peanutbutterandc>Can anybody please help me figure out why I am getting this error """/root/chordpro.scm:19:16: In procedure inputs:
<peanutbutterandc>error: perl-cairo: unbound variable
<peanutbutterandc>"""
<kts>US, Altgr, Dead Keys, to be exect.
<peanutbutterandc>Here's the associated expression from the .scm file:
<peanutbutterandc> (inputs
<peanutbutterandc> `(("perl-cairo" ,perl-cairo)
<peanutbutterandc> ("perl-pango" ,perl-pango)))
<peanutbutterandc>It is as recommended; and still I am getting the error for some reason
<lfam>peanutbutterandc: Do you import the module that contains perl-cairo in chordpro.scm? That would be 'gnu/packages/gtk.scm' according to `guix show perl-cairo`
<peanutbutterandc>lfam: I see. How would I import the module? (use-modules (guix gnu/packages/gtk.sm))?
<peanutbutterandc>Also, what kind of packages do I have to import the modules for and what kind do I need not? This is my first time packaging guix and I'm pretty much a n00b
<lfam>peanutbutterandc: You will need to import every package you refer to
<lfam>Start the file with something like this:
<lfam>(define-module (root chordpro) #:use-module (gnu packages gtk)
<lfam>)
<lfam>And you will probably need to import some other things too. Check out this as an example: https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/dvtm.scm
<peanutbutterandc>lfam, I see. Thank you very much!
<kts>Does "alt-intl" layout include dead keys? "us" "intl"?
<peanutbutterandc>While trying to build a perl app, I am getting this error (in the .drv file): Can't locate App/Packager.pm in @INC (you may need to install the App::Packager module) any ideas?
<peanutbutterandc>I am a perl n00b too (in addition to being a guix n00b)
<nckx>peanutbutterandc: Guix naming conventions would make that package ‘perl-app-packager’, which we can see does not exist. So you need to package it [too]. You can use the output of ‘guix import cpan App::Packager’ inside a (define-public perl-app-packager …) to do so.
<nckx>(As a starting point.)
<nckx>kts: The Web says yes…
<nckx>Only one way to find out.
<peanutbutterandc>nckx: That made me laugh. It's time to hit the books (reference manuals) then. Again. And maybe then I should submit it upstream, too. Whoa. Things are escalating a little too fast. I will probably be spamming this channel with my n00b questions again.
<nckx>That's… good and what we want? :)
<peanutbutterandc>nckx, I hope by the end of all this, I can hit the texinfo manual (or maybe just plain old github markdown) and write a n00bs guide to packaging stuffs with guix.
<peanutbutterandc>:)
<nckx>That would be great.
<peanutbutterandc>`guix import cpan App:Packager` is throwing the following error for me, strangely enough: https://paste.debian.net/1109089/
<nckx>Hm.
<nckx>There's a typo there (: instead of ::) but that shouldn't cause such an error.
<peanutbutterandc>Oh no, I did type :: in the command line (I just checked again)
<peanutbutterandc>What does guix import do, exactly? o.O [I did skim through the reference about this too but didn't quite get it. Maybe I should re-read again.]
<nckx>peanutbutterandc: Sorry to break your flow but I have to leave #guix again. Could you send a bug report with full error output to bug-guix@gnu.org? This should obviously not happen, and the command works here: http://paste.debian.net/1109090/ (so you can at least get started packaging).
<nckx>peanutbutterandc: It queries an supported upstream repository (see --help for a list) and tries to extract as much info as possible for you. The quality of the result varies accordingly. Some repositories specify dependencies well, others don't. Looks like CPAN doesn't have licence info for this package, so it's missing (#f). Etc.
*nckx → AFK.
<peanutbutterandc>nckx, Okay! I will first try with a clean container (I'm running guix in alpine-based docker container for now: too n00b to use it on my host already). I haven't `guix pull`ed either, yet; so I'll have to do that too. Thank you for your help, again.
<peanutbutterandc>nckx, I tried to reproduce again but couldn't. It works now.
<sirmacik>hey (:
<sirmacik>anybody hit the problem with docker/docker-compose about runc?
<roptat>hi guix!
<roptat>if any of you is on mobile: can you try https://guix.gnu.org/manual/devel/en/html_node/index.html and tell me if it looks right?
<kdtsh>roptat: looks pretty good!
<sneek>Welcome back kdtsh, you have 1 message.
<sneek>kdtsh, nckx says: It's exactly the kind of weird, unreproducible (here anyway) bug that softly whispers ‘corruption’ in one's earhole, so I don't blame you. The only man bug I ever reported is at <http://issues.guix.gnu.org/issue/30785>; probably unrelated but has some nice debugging tips later downthread. If you're interested. Don't blame you if you just want to reïnstall.
<roptat>ok, if it works well, I'll port the change to the manual for the 1.0.1 when I can :)
<leoprikler>quick question: how do I mount NTFS partitions as rw on Guix?
<nckx>leoprikler: ntfs-3g /dev/foo /bar
<nckx>Same as other distributions I've used.
<leoprikler>except that guix doesn't come with it out of the box
<leoprikler>I've solved it with an ad-hoc environment for now
<leoprikler>would GFVS/Nautilus pick it up, if ntfs-3g was installed inside the OS's (packages ...)
<nckx>leoprikler: Hm, not a Gnome user, no idea. (And I've never had ntfs-3g out of the box, either…)
<nckx>ntfs-3g provides mount.ntfs, so it definitely *should*.
<efraim>the font-awesome zip source keeps on being regenerated, even from their new website
<nckx>Such web.
<efraim>I've fixed it locally to use git-fetch, I'll push it in a bit
<lprndn>Hello everybody!
<raghavgururajan>Hello Guix!
<raghavgururajan>I was wondering if there all packages in master branch are considered stable? That is, if a user periodically runs `guix pull`, `guix system reconfigure` and `guix upgrade`; will the system prone to break?
<leoprikler>raghavgururajan: it should not break, but errors can happen
<roptat>guix is more of a rolling release
<leoprikler>if they appear, should report them and perhaps work around them with ---roll-back
<htsr>Hi! How can I run a modified version of shepherd? If I create a new shepherd package in a channel, will it work?
<leoprikler>s/---/--/
<raghavgururajan>leoprikler I see. Is this something that can also be encountered on other LTS-based distros as well?
<roptat>but we package only "stable" versions from upstream, not beta versions
<roptat>htsr, you'll have to have it used somehow in your config.scm
<raghavgururajan>roptat I see. But for that, the upstream should be follow LTS model right? For example, for youtube-dl, I don't think there is stable and unstable variants.
<roptat>right
<htsr>roptat: by modifying %base-packages?
<roptat>htsr, I would modify the %base-services rather
<roptat>to make sure the shepherd service is run with your modified shepherd
<roptat>see gnu/services/shepherd.scm
<roptat>it seems you can't easily pass another package
<roptat>to shepherd-root-service-type
<raghavgururajan>roptat What about Guix itself? How is it released?
<roptat>oh, the shepherd-root-service-type is not part of the base services, it's automatically instantiated, so I think you can add a similar service, with your modified shepherd
<roptat>raghavgururajan, when we feel like it :)
<raghavgururajan>roptat I mean, is every release considered stable? That is, are updates and modifications are internally tested before releasing?
<roptat>not really, we make sure tests pass and we take a week or so to make sure everything is not completely broken
<roptat>that the installer works, substitutes are available, ...
<roptat>we use a gc root on the build farm to make sure they are always available, too
<raghavgururajan>roptat I see. Thanks!
<htsr>roptat: i just want to change the reboot action of shepherd...
<htsr>i've looked at shepherd.scm, is it possible to change extensions?
<htsr>because if I can change (service-extension boot-service-type shepherd-boot-gexp) to my shepherd-boot-gexp it's perfect
<jlicht>hey guix
<jboy>I'm attempting a "borg-like" install of GuixSD on a virtual server. Everything seems to work -- I am able to init the system and ssh into it -- but then suddenly things fall apart. I've documented what I did here: https://gist.github.com/jboynyc/ef312a686749de693dddc4f3a657a919
<jboy>does this look familiar to anybody? What might be the reason that files are no suddenly no longer found?
***emyles`` is now known as emyles
<leoprikler>jboy: "guix system init" overwrites stuff, that already exists
<jboy>right, that's the intention leoprikler
<leoprikler>of course
<leoprikler>what does "which guix" return?
<jboy>I can't run which anymore either. I did however boot the server from a grml recovery disk, and I'm able to see that the /gnu/store looks intact, with guix in /gnu/store/2170pnw...-guix-1.0.1-9..../bin/guix
<leoprikler>ah, but 2170... is not 23q7...
<leoprikler>perhaps you can run "/gnu/store/2170pnw...-guix-1.0.1-9..../bin/guix pull" from the running distro?
<htsr>roptat: i don't understand, if i add a new service with the same name it will work?
<htsr>and where do you see it's automatically instantiated?
<htsr>and because shepherd-root-service-type use (extend append) i can't modify the list of extensions, so right now it's nearly impossible to start a script at reboot in guix :/
<roptat>htsr, I don't see it, so I suppose it's automatically instantiated
<roptat>if a service of the same name already exist, it won't be automatically instantiated
<roptat>htsr, actually, it's not automatically instantiated :/
<roptat>I found it in operating-system-default-essential-services
<roptat>so it doesn't seem like you can modify the shepherd service as you want, unless you change the sources for guix
<roptat>jboy, this kind of thing happened to me once, my disk was actually corrupted, and the guix binary replaced by some Irish error message from glibc ^^'
<jboy>roptat: But this is the fourth or fifth time in a row I installed the system like this, and every time the same thing happens.
<roptat>right, it also happened when installing guix on an SD card :/
<roptat>I tried three times, and the fs was corrupted three times, I don't really understand why
<jboy>leoprikler: I'm trying to chroot from the grml recovery disk to see what happens
<roptat>but I suppose you're not installing on an SD card :)
<roptat>I guess you could run fsck from the recovery?
<jboy>I can try that roptat
<jboy>If I try to run the bash from /mnt/gnu/store I get "No such file or directory" even though the file clearly exists in the fs hierarchy :\
<jboy>fsck tells me it's clean
<roptat>mh...
<roptat>can you check permission?
<roptat>oh, and the output of ldd too
<roptat>maybe one of the shared libraries (or the dynamic linker) is broken
<roptat>or missing
<leoprikler>I'm not sure, how well guix works from chroot -- my only experiments so far were no success stories.
<apteryx_>civodul: Editing multiple lines at the Geiser REPL without M-j has been implemented! I just tried it, it suites my needs very well: https://gitlab.com/jaor/geiser/issues/282
***apteryx_ is now known as apteryx
<bluekeys>Hi guix. I've installed emacs-all-the-icons. But when I try (insert (all-the-icons-for-file "foo.js")) I don't get a beautiful icon.
<bluekeys>I can see the fonts are in .local/share/fonts
<bluekeys>fc-list doesn't show the fonts as available
<bdju>looks like the irssi package is a couple versions behind
<jboy>roptat: I'm not sure I know how to check all that, so I think I will write to the list.
<civodul>heya apteryx, that sounds like good news!
<htsr>roptat: thanks
<bluekeys>Bye guix.
<kts>Is there xfs?
<bdju>I think just ext4 and btrfs are available
<kts>The unavailability of aria2c is inconsequential.
<leoprikler>According to "guix build aria2 --dry-run", it would build "/gnu/store/zsbx19dc0hgkwpr22dp758ydbl1pn2yk-aria2-1.35.0.drv"
<Snai1>hi! i'm trying to configure skeleton files with the config.scm but when `guix system vm config.scm` execute i get "build of /gnu/store/...-skel.drv failed" and i can't find any documentation about
<lprndn>Snai1: can you share your config.scm?
<Snai1>yes https://gitlab.com/Aleci/AlecisLaptop/blob/master/Guix/Aleci.scm
<lprndn>Snai1: What is config-injection? Otherwise the use of skeletons seems to be alright so my guess would be either guix doesn't find the files or a permission problem... :/ (Are ***BASHRC*** and others returning strings?)
<Snai1>lprndn: its a parser that replaces a tag (like ***BASHRC***) with the content of the file bounded with that tag. (the special character like \ and " are automatically escaped before the injection. I did that because I don't like to work with other configuration in the same file.
<lprndn>Snai1:What if you replace ,(plain-file ...) with (local-file "path/to/your/file") ?
<lprndn>* ,(local-file "...")
<lprndn>(also maybe try with sudo? in case its a permission thing...)
<Snai1>lprndn: no local-file procedure runs only at runtime, i want to keep my configuration in my config.scm without see them basically. that is not a problem of privileges because i already run that as a root
<lprndn>Snai1: Ok, sorry, I don't know. :/ Hope you'll find a solution
<Snai1>lprndn: thank you! (the fact is that everything works if i keep just the bashrc and i3 configurations, the problems appear with the .emacs file)
<roptat>Snai1, when it fails building, it should give you a log location, that you can open with e.g. bzless
<roptat>maybe you'll find more useful information in it?
<Snai1>roptat: you're right i'll check it out, thanks
<dongcarl>civodul: Did you understand why mhw wanted to avoid `-idirafter` here? https://issues.guix.gnu.org/issue/30756#12
<dongcarl>It seems like the perfect solution to me...
<kts>Why doesn't linux-libre support my ethernet?
<leoprikler>perhaps you're using a very weird device?
<leoprikler>can you do "lspci | grep Ethernet"?
<civodul>hi dongcarl!
<dongcarl>Hello!
<civodul>dongcarl: i think mhw was saying that our problems come from the fact that libc doesn't come last
<civodul>and that therefore we should move libc first in the search path
<civodul>but i forgot the details so i'd have to page all that in ;-)
<civodul>maybe we can give the -idirafter hack a try, if you feel it's worth it?
<dongcarl>civodul: Right, I was having the same problem when compiling with mingw-w64 as well, it somehow gets added at the top of the search paths, and include_next is not happy with that
<civodul>oh yes, i see
<dongcarl>I'm thinking `-idirafter` is the solution no? Perhaps I'm missing something more fundamental?
*dongcarl looking at how I can patch `gcc` to have an env var for `-idirafter`
<kts>rtl810xE
<civodul>leaked password :-)
<civodul>dongcarl: it might be the solution, dunno
<dongcarl>civodul: Okay, I'll respond to that thread then... Perhaps mhw had something else in mind
<civodul>cool, thanks for investigating
<civodul>i'm sure we can do better than what we have now!
<dongcarl>civodul: True!
<brendyyn>So, since there is no support on wayland/sway yet for something like xmodmap, im looking for a way to set my custom xkb layout. i could modify xkeyboard-layouts, but then that would require rebuilding hundreds of packages. setxkbmap would only change the layout for xwayland programs. has anyone done this on Sway?
<brendyyn>i will read the chat log if anyone comments while im offline
***ajames` is now known as ajames
<dongcarl>sneek later tell mbakke Here's my hacky fix for the nsis cross-compilation: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=37870, I believe we might be able to use `-idirafter` to resolve this in a cleaner way, and `-idirafter` may also be a fix for #30756, will post there!
<sneek>Okay.
<nckx>sneek: later tell kts: The rtl810xE is (in theory) supported by the r8169 driver but appears to require proprietary firmware which Guix doesn't and won't provide.
<sneek>Will do.
<efraim>and that's for ethernet⸘?
*efraim was trying for a ‽
<nckx>'pears so.
<nckx>efraim: How'd you compose that?
<efraim>compose key and some combination of ! and ?
<efraim>while holding down the compose key
<efraim>same way I'd do ™ or ⓒ
<nckx>I was wondering *which* combination 🙂
<nckx>I can only ‽.
<kts>nckx: Yet the Icecat tofu exists.
<sneek>kts, you have 1 message.
<sneek>kts, nckx says: The rtl810xE is (in theory) supported by the r8169 driver but appears to require proprietary firmware which Guix doesn't and won't provide.
<peanutbutterandc>So, I have a file that I want to `guix package --build-from-file=FILE` from, in the same veins as the first my-hello.scm example in https://guix.gnu.org/blog/2018/a-packaging-tutorial-for-guix/, but it defines other packages in the same file as well; but only installs a single one when given the command. Any ideas?
<peanutbutterandc>I guess the question is - can one not define multiple packages: actually, just a single package and it's dependencies, in a .scm file and use it to --install-from-file with?
<peanutbutterandc>s/--build-from-file/--install-from-file/
<efraim>can you return a list or only a single package?
<dongcarl>peanutbutterandc: I'm trying to understand what you're asking... Do you want to build multiple packages or just one?
<peanutbutterandc>dongcarl, efraim - This is what I currently have: https://paste.debian.net/1109206/ It is an attempt to package `chordpro`; but since it has dependencies that aren't in guix (but in cspan), I have the definitions for the dependencies in there as well
<peanutbutterandc>But running `guix package --install-from-file=chordpro.scm` does not install chordpro
<dongcarl>peanutbutterandc: Ah! You gotta move the "my-chordpro" package to the bottom
<dongcarl>and also... `(define perl-image-info (package ......))` for the dependencies
<peanutbutterandc>dongcarl, Will try that again (I'm pretty sure I tried it already to no avail). Does the file name also have to be my-chordpro.scm? Sorry I don't understand the second part
<dongcarl>Lemme send you an updated scm...
<peanutbutterandc>That would be great. Thank you
<dongcarl>peanutbutterandc: Here's how it'd look like https://paste.sr.ht/%7Edongcarl/6465d0dfa065e04349888866b5b501f4b0d3186a
<dongcarl>the thing is... `(package ...)` EVALUATES to a package... whereas `(define foo bar` defines the symbol foo to refer to bar...
<dongcarl>And the `--from-file` flags only builds/installs whatever the file evaluates to, which is the last `(package ...)` in your case
<dongcarl>So in your original file, all the other `(package ...)` calls were basically thrown away
<dongcarl>(all the other `(package ..)` calls except the last one)
<peanutbutterandc>dongcarl, That makes a lot of sense. Thank you very much!
<dongcarl>:-)
<peanutbutterandc>This is strange... I get the following error:
<peanutbutterandc>allocate_stack failed: Cannot allocate memory
<peanutbutterandc>Warning: Unwind-only `stack-overflow' exception; skipping pre-unwind handler.
<peanutbutterandc>allocate_stack failed: Cannot allocate memory
<peanutbutterandc>Warning: Unwind-only `stack-overflow' exception; skipping pre-unwind handler.
<peanutbutterandc>I am running guix inside a container on my foreign distro
<peanutbutterandc>Yeah... I'm still getting the same error... even with just `guix build -f FILE.scm`
<dongcarl>peanutbutterandc: Not sure about that error sorry :-( If you could upload full logs maybe someone will know
<peanutbutterandc>dongcarl, Could you perhaps please try running the build on your guix and seeing if the error persists? That would mean something is wrong with the definitions and I'll have to try and work it out..
<peanutbutterandc>Also, it just failed... silently, apart from the message above. Without giving me the location for the logs...
<dongcarl>peanutbutterandc: take a look at "perl-image-info"
<dongcarl>it uses itself as an input...
<dongcarl>So...
<dongcarl>recurse until out of memory
<peanutbutterandc>Ah! I see it too. I'll try removing it and then trying again. Thank you!
<peanutbutterandc>It's showing good signs now - downloading packages. Yay!
<civodul>hmm "weasyprint-library-paths.patch" is missing, which breaks evaluations
<efraim> http://git.savannah.gnu.org/cgit/guix.git/commit/?id=73f0ed8dbfde97341716fd8b156907e4ef800873
<dongcarl>peanutbutterandc: no worries, I'm seeing that some of the tests fail for `my-chordpro-idk` but you can probably either fix them or make the package skip tests
<peanutbutterandc>dongcarl, I see. Thank you. Same here. perhaps I'll report the tests failure to the devs, too. Thank you for your help!
<dongcarl>peanutbutterandc: no worries!
<peanutbutterandc>:)
<civodul>efraim: yeah i've pinged Hartmut, let's see
<civodul>maybe we'll revert if they're afk
<dongcarl>Did `guix environment` lose its little spinny `\|/-` indicator?
<civodul>no, why?
<civodul>the spinner is only when the build process doesn't provide completion info
<civodul>when there's completion info, you get a sweet progress bar :-)
<dongcarl>Huh... I'm getting nothing right now when building my manifest... :-(
<civodul>you got a "building ..." or "downloading ..." line?
<dongcarl>guix environment --manifest=contrib/guix/manifest.scm --max-jobs=(nproc)
<dongcarl>Updating channel 'bitcoin-guix' from Git repository at 'https://github.com/dongcarl/bitcoin-guix.git'...
<dongcarl>Updating channel 'guix' from Git repository at 'https://github.com/dongcarl/guix.git'...
<dongcarl>then... silence...
<civodul>hmm!
<civodul>there was a bug report about duplicate channels
<civodul>but that doesn't seem to be the case here?
<dongcarl>civodul: Right, not the case here... I mean my manifest might be broken, but still the indicator should be there at least..
<civodul>yeah that's weird
<dongcarl>To the channel in general: I found something broken in make-gcc-toolchain and made a minimal reproducer (and attached logs) if anyone has the time: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=37876
<civodul>dongcarl: re https://issues.guix.gnu.org/issue/37876 could you send the tail of the build log?
<dongcarl>civodul: Yes, how many lines you want?
<civodul>dunno, the relevant bits, could be 40 lines or so?
<dongcarl>civodul: Gotcha
<civodul>coolio :-)
<civodul>that makes the bug report self-contained
<dongcarl>civodul: Uhh... The first ` error:` occurs about 3k lines from the bottom of the log...
<dongcarl>:-(
<civodul>ouch :-)
<civodul>alright, let me see
*dongcarl is afk for a bit while I get home from the office
<nckx>Hi #guix! All right to revert the missing weasypatch on master? It's not in the bug report either.
<leoprikler>out of curiosity: what is a weasypatch?
<nckx>civodul: ☝
<nckx>leoprikler: A patch to weasyprint that's missing :)
<leoprikler> https://lists.gnu.org/archive/html/guix-devel/2019-10/msg00530.html
<nckx>leoprikler: ?
<leoprikler>The patch has already been demanded from what I can see. Let's wait for it.
<nckx>leoprikler: I'd seen that mail; I wouldn't call it a demand.
<nckx>leoprikler: There's no reason to keep master evaluation broken if they've gone to bed.
<leoprikler>You can still revert the patch if they don't wake up.
<nckx>I already have :) Good night! o/
<leoprikler>Good night.