IRC channel logs


back to list of logs

<quiliro>i still have my guix system reconfigure /etc/config.scm error
<quiliro>after guix pull
<quiliro>on root
<lfam>quiliro: What is the error?
<quiliro>lfam: not sure.... a lot of text
<lfam>quiliro: Can you put it on sprunge like you did yesterday? And also your config.scm?
<quiliro>lfam: the last is
<quiliro>substitute: guix/ui.scm:1209:6: In procedure run-guix-command:
<quiliro>substitute: guix/ui.scm:1209:6: Bad Read-Header-Line header: #<eof>
<quiliro>guix system: error: build failed: substituter `substitute' died unexpectedly
<mekeor>please stop flooding the channel. paste it somewhere online, please.
***mildred1 is now known as mildred
<quiliro>how many lines is flooding?
<mekeor>>= 3
<quiliro>mekeor: ok... will avoid it
<quiliro>guix system reconfigure /etc/config.scm --fallback 2>&1 | curl -F 'sprunge=<-'
<quiliro>is correct?
<quiliro>so this line is flooding because it is the 3d?
<quiliro>lfam: is that command correct?
<lfam>quiliro: Try it :)
<quiliro>what does the 2>&1 do?
<lfam>It redirects the 'standard error' output stream to the 'standard output' stream. If you don't do that, the errors will not go into the pipe to curl
<quiliro>what manual teaches that?
<lfam>I'm not sure... it's general to all Unix shells. So if you are using Bash, I would read the Bash manual
<lfam>quiliro: Can you file a bug report with these errors? I'm not sure what is wrong, so a bug report will make it easier for other people to notice your problem
<quiliro>lfam: i tryed yesterday but claws mail would not deliver some emails
<quiliro>and i lost all the text
<mekeor>quiliro: you could also use the web interface of your mail-provider and just paste the two sprunge-links from above with some explanation
<jahboite>guix pull (on foreign distro) doesn't seem to update root's guix package; so is it correct to say that one must "# guix package -u guix" to do that, after a pull? and then one would ordinarily update an init script to point to the new guix-daemon?
<lfam>jahboite: That's correct. However, rather than upating your init script, you should have it execute the path '/var/guix/profiles/per-user/root/guix-profile/bin/guix-daemon'. That is a symlink to root's guix package's guix-daemon
<jahboite>lfam: understood. thank you.
<quiliro>guix pull error message report sent
<quiliro>rekado_: you said i could install substitutes exclusively in a certain way?
<quiliro>i do not want to compile because my machine is old
<quiliro>i would like a permanent binary repo
<quiliro>be back after reboot
<lfam>Anyone handy with flex? doxygen is failing to build on the core-updates branch, like this:
<lfam>I tried updating doxygen to the latest and disabling parallel building, but no change
<lfam>A flex update looks promising:
<lfam>Although it basically requires a full rebuild...
<Apteryx>Is there a way to make the text rendering more readable when the "skia" backend is in use in IceCat?
<Apteryx>This was recommended to make IceCat stop crashing so often, but it also makes the text look really rough ('d' looks like 'c').
<Apteryx>'n' looks like 'r', etc.
<rekado_>Apteryx: it worked for me after changing the font, IIRC.
<rekado_>some fonts look terrible when skia is enabled
***snape` is now known as snape
<ng0>hi, disclaimer for the two dovecot plugins I've just sent: I don't run dovecot currently and I've sent the second one from memory, as I'm currently stuck with half-broken systems.. this one is still moving my mail around so I can't build at the same time. The plugins have been build before.
<ng0>but they are just .so files in the end, going by the instructions of upstream :)
<jmd>ng0: It sounds not so much as a disclaimer. More like a mea culpa.
<ng0>yeah :)
<ng0>i made the mistake to buy a laptop for cheap which was used by someone who didn't use it the way I use computers, or even long enough to discover the glitch. first purchase from a private person and not a company who refurbished them :/
<ng0>got my eyes on another one, but I'll wait some months
<jmd>What's the glitch?
<ng0>after a certain time of a certain cpu load, the screen starts flickering and eventually stays at one color. can be worked around by just going into standby for a whort moment. it was cheap enough to not justify looking into repairing this
<ng0>it doesn't look like harmut will review the neo2 "kbd" package addition any time soon, maybe harmut is busy with something else currently. Can someone look into it? For me the way I submitted it worked, and you only need to test that at the tty you now have part of the neo2 layout, you don't need to test not all of it, it has 5 or 6 layers
<ng0>with test I meant review
<jmd>There was talk about us using a tracker for patch reviews. Was any decision made?
<ng0>yes, guix-patches uses debbugs
<jmd>Oh ok. I didn't know we had a guix-patches list now.
<ng0>seems like many people don't. I think the patchwork url could be exchanged wit hthe debbugs instance. but there was more than one announcement emai labout it.. I don't know how to communicate it better
<slyfox>updating guix/HACKING would be a good first step
<ng0>whoever can change the topic here should do it. and one of our maintainer people should tell the people to remove the patchworks for guix
<ng0>i think it's alraedy in the docs
<ng0>the online manual is just at the state of when 0.12 was released
<slyfox>i've just synced and it still lists -devel
<ng0>ah.. I'll send a patch for HACKING. thanks :)
<jmd>ng0: But make sure that you send it to the right place :)
<slyfox>having a new guix release would be also nice :)
<ng0>jmd: I use guix-patches.. only for older patch discussions I use -devel
<snape>wingo: ng0: FYI the dovecot service has been broken for about 3 months, I did a patch to fix it, I'll send it tonight
<ng0>ah.. nice
<wingo>when you send the mail, if you know what broke and how it broke, that would be good so we can avoid this happening again
<snape>I know what broke, it is commit 5305ed200 :)
<ng0>there's an linux-libre grsec TAILS variant preview coming out today:
<ng0>oh, and on the weekend I had a chat with Jookia, and I have some action item to forward to Guix. it's indirectly-directly affecting us, about the progress with aspell and the other spells.
<ng0>it makes more sense to write an email about it rather than chatting here about it :)
<ZombieChicken>It's sad; emacs is supposedly such a great editor yet it can't apparently handle modelines sanely and, when it's given one, it barfs and does nothing with it
<ZombieChicken>and unfortunately the community is so shit there is no way to get any help
<jmd>ZombieChicken: I normally find #emacs of great help.
<ZombieChicken>jmd: All I've ever found there were trolls
<ZombieChicken>and extremely rude people
<ng0>so heads doesn't use systemd. but they don't say what they use instead, guess reading the code is the only answer
<ZombieChicken>Anyone tried getting nonroot X working with elogind?
<ng0>Okay, it uses whatever devuan uses.
<erliphant>I'd like to "freeze" a set of guix versions so that a bunch of developers on different machines see the same view of the world. Is there a recommended way to do this?
<erliphant>I'd like to snapshot the guix package meta information rather than exporting a profile.
<ng0>I'd like to choose a logo for plymouth to test it with. Would the one we use for the grub background be okay if the format works for plymouth?
<rekado_>erliphant: you could have a centrally managed guix git checkout at a fixed version
<erliphant>@rekado - that's something I've considered. I tried just taking the package definitions and sticking them in by GUIX_PACKAGE_PATH but all sorts of stuff seemed to break
<erliphant>@rekado - how would I convince guix to pull from there?
<rekado_>you cannot have conflicting definitions of modules on the GUIX_PACKAGE_PATH
<rekado_>packages are defined in modules and they must be unique
<erliphant>@rekado - ok that's useful. I thought I might get away with path precendence at load time
<rekado_>this is why it wouldn’t work to put a copy of the “gnu” directory in GUIX_PACKAGE_PATH
<rekado_>you could link ~/.config/guix/latest to the shared checkout.
<rekado_>or you could create /usr/local/bin/guix that uses a particular Guix checkout in some global directory.
<ng0>wingo: is there any chance you have an up to date copy of the shallow-branch patch for git still around? I had to rewrite half of it and other than the copyright I could attach it to your name.
<ng0>*I could not attach
<erliphant>@rekado - that's interesting. Where does guix store it's local package metadata. Is it inside the profile? Or does it go somewhere in /var?
<erliphant>How is an instance of the guix binary linked to it's associated metadata?
<rekado_>Guix is a set of Guile scripts that use the rest of Guix as a Guile library
<rekado_>every package is bound to a value in the library
<ng0>wingo: what I have so far is not functional as I keep bumping into issues with other services I'm writing, so the shallow-branch patch has only been roughly updated. I'm pushing it to my checkout now
<rekado_>the majority of Guix consists of package values.
<wingo>hi, i don't care about the copyright, and i don't have an up-to-date patch; just the one that went to the ML
<ng0>ah, ok
<erliphant>@rekado - I understand. So it's just library code inside that particular version of guix.
<rekado_>erliphant: there is no package metadata. There’s just package values.
<rekado_>when you clone the Guix repository you get all of that.
<erliphant>@rekado - that's exactly what I need. Thanks.
<rekado_>you can link ~/.config/guix/latest to the directory of the checkout to use the Guile modules from there.
<rekado_>Guix is configured to first check if ~/.config/guix/latest is set
<erliphant>@rekado - guix pull an upgraded guix and libs and installs them.
<erliphant>@rekado - thanks - extremely helpful as always.
<rekado_>so even if you start out with an older version of Guix (e.g. some installation in a global directory, or in your profile): if ~/.config/guix/latest is set the latest modules will be used.
<rekado_>“guix pull” downloads the latest version to the store and links ~/.config/guix/latest to point at it.
<erliphant>@rekado - is there are risk that guix/latest will be incompatible with your older guix?
<rekado_>I don’t think so.
<rekado_>the “entry” script has not changed in a long time, and all it does is dispatch to the actual scripts in the effective version of Guix.
<erliphant>@rekado - interesting.. I think I've got a few things to think about and a possible way forward. Thanks
<ng0>I just rebased my palemoon package branch. the problem here is, it's almost finished I'm just running into runtim/linker path problems (I can't remember what exactly because I haven't touched it in a while). Other than that, it's okay once someone can help me to fix that. What I don't know (yet!) is what needs to be changed so taht the browser can be added to master. I know that the initial "new tab" page must
<ng0>go for the services it promotes on there.
<ng0>and I think due to the palemoon dbinary distribution license we must build it like texlive
<ng0>the license forbids any unauthorized _binaries_ (not sources) to be redistributed
<jmd>ng0: Doesn't that mean it's non-free?
<ng0>I'm not sure if I want to deal with upstream if issues should appear, as this sort of binary distribution restriction makes me think that they will regard every non-standard build as an insult
<ng0>it's on the website, jmd . which is behind cloudflare with tor exit node block
<ng0>the source code license is mpl2.0, gpl2+, lgpl2.1, and others (like firefox)
<ng0>but because of this re-distribution license I only bothered to check months after I started working on it, I'm having doubts
<ng0>if I would have to drop it, it's okay.. as one of the dependencies I had to package apng, which can be used in icecat to unbundle that :)
<ng0>jmd: and
<ng0>I read redist as when we don't use the official branding we could redistribute it. However, I haven't bothered to check if they ship anything other than official branding to choose from
<ng0>but it also applies some non-commercial conditions
<ng0>and the exceptions do not apply as we would alter more than what is needed to make it work (new tab page is problematic for example)
<ng0>and there's 13 which could apply for us
<ng0>I'm not sure if this browser is worth the effort. Okay, I started in september with it, but it requires more modifications then, and the only browser I'd feel like maintaining by following upstream and keeping patches to free it is torbrowser, and I feel like investing more energy into packaging torbrowser. But I'm not a lawyer, I could understand the licensing completely if I wanted to but I don't have the
<ng0>time today.,And maybe someone feels like palemoon is a good idea, so I'll just keep it around in its branch on
<jmd>Sounds like a plan - f its distribution conditions are uncertain and there are other programs which do a similar job.
<ng0>on gentoo it was easy. I could use gentoo foundation as the copyright, and after all I had no clue who did the ebuild I used in my overlay and updated it. there's even the palemoon-overlay. It's all sources, no binaries so they can do this. for guix, there is my email attached to the commit.. I'd rather avoid confrontation with some questionable organization
<jmd>ng0: Gentoo *can* do that. But then they are distributing non-free software.
<ng0>it was not in gentoo portage, that's why I explicity said "overlay" :)
<ng0>maybe not so obvious if one never developed packages for the system though
<jmd>I've never played with gentoo, so I don't know what all those words mean.
<ng0>portage is what in Guix would be master of guix.git, and overlay is what's guix_package_path (roughly) in Guix
<ng0>where I'm now I'd argue that with both pros and cons of each user-package-path variant, GUIX_PACKAGE_PATH is more flexible
<ng0>especially the (inherit) is smart and avoids hardlinking files and version if/else bashism :)
<efraim>cmake failed on aarch64 on core-updates
<quiliro>how can i subscribe to bugs-guix
<thomasd>quiliro: (it's bug-guix)
<thomasd>However, you don't need to subscribe in order to submit. You can just send an a-mail to (if it's your first report, it can take a while before it appears. this is due to spam detection mechanisms, as far as I understand)
<quiliro>thomasd: i am interested in learning by watching how other people's problems are solved
<thomasd>quiliro: ok. There's also help-guix (in you're not already subscribed)!
<quiliro>help-guix i think is better, thomasd
<youri>Hi there, I am working through the guix manual and I am just past the `guix package -i hello` command but I'm suffering from many 'note: source file ... newer than compiled' warnings.
<youri>Also the binary responds really slow: time guix --help 8.28s user 0.04s system 107% cpu 7.757 total
<youri>Could somebody suggest What can I do or what am I doing wrong?
<jamesrichardson>Hello perl users ;) Apparently perl critic in not in guix yet. I've run guix import cpan Perl::Critic which is fine? Now I have to import all the many dependencies. Is there a way to do such recursively?
<jmd>jamesrichardson: Not that I know of
<jamesrichardson>Outside of that, would the project welcome the addition of perl critic and relavent dependencies?
<jmd>relevAnt !!
<jmd>I don't know what perl::critic does, but if it it free software then I don't imagine that anyone would object.
<Apteryx>youri: I guess that you are running Guix from a git checkout?
<youri>from a .tar.zx package
<youri>from an official gnu url
<efraim>false alarm for cmake on aarch64, it built the second time
<Apteryx>youri: Hmm. I don't know then. Usually this happens for me when I do a "git pull" in my guix checkout and the source files & compiled files do not match.
<thomasd>Hi #Guix
***remote_ is now known as remote
<Apteryx>rekado_: Thanks. Seems non-trivial to figure out which font doesn't play well with skia, but I'll remember to try others if I get too annoyed about it.
<Apteryx>Any idea why Guile will keep displaying "note: source file ... newer than $HOME/.cache//guile/ccache/..." ? I know how to fix this: I need to run make. But my question is why? My understanding of how Guile works is that if the source is newer than any compiled version it knows about in its compiled path, this gets printed *and* the file is compiled and stored in the .cache directory. On subsquent use of Guix I
<Apteryx>would not expect to see these messages anymore (I already should have matching compiled versions under $HOME/.cache...).
<thomasd>It seems that changing the configuration of a system service only takes effect after reconfiguring AND rebooting (I think shepherd needs to be restarted). Am I right about this? Or is there a way to restart/reconfigure shepherd without rebooting?
<Apteryx>thomasd: I think by default when a service configuration changes at least shepherd should be restarting it. It's been a while I've studied how services work though.
<thomasd>Apteryx: it restarts it, but with the same configuration, I think :/ maybe it depends on which part of the configuration is changed
<thomasd>e.g. for CUPS, /gnu/store/...-shepherd.conf refers to /gnu/store/...-shepherd-cups.scm, and this shepherd-cups.scm contains the command line used to (re-) start CUPS, which points to a fixed /gnu/store/...-cupsd.conf.
<thomasd>this makes me think that, until shepherd itself is restarted (with a new shepherd.conf), no new cups configuration will be used?
<ng0>is someone working on elm on guix? I just found this language :)
<thomasd>(I'm guessing here, I've only just started to look into services :-) )
<alezost>youri: it looks like .scm are not compiled, that's why you have those warnings and why guix is slow; how did you install guix?
<jmd>ng0: mutt is the nachfolger to elm.
<ng0>but that's not here either, at least only the mailclient mutt is in guix .)
<jmd>Both mutt and elm are mail clients
<ng0>I meant the language
<jmd>(with almost identical user interfaces).
<ng0>one momen
<youri>@alezost, I just copies the two directories ./var/guix and ./gnu to /var and to root, then I ran a systemd command to start the server
<alezost>youri: copied from where?
<youri>exactly those steps
<youri>wget << linux x64 and I checked the signature too
<quiliro>i cannot send some emails
<quiliro>others are delivered
<ng0>quiliro: what do you mean?
<adfeno>quiliro: Which email client?
<ng0>if one of your email applications and/or server side can enable logs, I would use them and llok at the logs
<youri>alezost: how are scm files typically compiled, is this done automatically with a guix pull?
<alezost>youri: I've never tried binary installation, but I guess, all the needed files (including the compiled .go files) already should be in store, so I don't really understand why you have this problem.
<alezost>do you have GUILE_LOAD_PATH envirnoment set?
<alezost>also did you run "guix pull"?
<ng0>you still need to guix pull I think (not sure)
<youri>yes, i did
<quiliro>ng0 adfeno: it is in claws-mail
<youri>under normal user and under root, but it doesn't make the behavour change
<youri>I think I will try the AUR package later this week
<youri>Honestly, thanks for helping alezost :)
<alezost>youri: well, there was no help, so no problem :-) but could you check "which guix"?
<alezost>and "file ~/.config/guix/latest"
<youri>guix which guix -> /usr/local/bin/guix
<youri>0 lrwxrwxrwx 1 root root 54 Feb 28 13:34 /usr/local/bin/guix -> /var/guix/profiles/per-user/root/guix-profile/bin/guix
<youri>and /home/youri/.config/guix/latest: symbolic link to /gnu/store/pgv1bk6xa3zvmhr8bjljjbqrfv0rxjz0-guix-latest
<alezost>youri: what about "echo $GUILE_LOAD_PATH"?
<youri>that is empty I guess
<youri>it is not: /gnu/store/7pw8yb83y59lz2v6kpi1pcg6cmkvlkif-profile/share/guile/site/2.0:/gnu/store/sd8jm2rw7cp3bnrk421kr97ki6sqxnhz-guile-json-0.5.0/share/guile/site/2.0:/gnu/store/w90isin9pnm9ri8w9njxby8h98lfnkzq-guile-ssh-0.10.2/share/guile/site/2.0:/gnu/store/dkq6dzl1j0j1dkc6hsmif4c64llbaa69-gnutls-3.5.4/share/guile/site/2.0
<youri>but I can't remember I did set that explicitly
<alezost>that's probably from sourcing ~/.guix-profile/etc/profile
<alezost>everything looks good to me, so I'm lost; could you paste a full output of "guix ..." command that gives you the compilation warnings?
<adfeno>quiliro: What exactly happens when you try to send email?
<adfeno>Is there an error message:
<youri>;;; note: source file /gnu/store/7fisf4frrgsjzmknjbab1dal23wxrp8d-guile-2.0.13/share/guile/2.0/ice-9/eval.scm
<youri>;;; newer than compiled /gnu/store/7fisf4frrgsjzmknjbab1dal23wxrp8d-guile-2.0.13/lib/guile/2.0/ccache/ice-9/eval.go
<youri>guix: missing command name
<youri>Try `guix --help' for more information.
<youri>it depends what I run, for different arguments I get different (more) errors
<quiliro>adfeno: the result when the email cannot enter the queue folder and this is the result when it is delivered
<alezost>youri: oh, that's looks like a different version of guile is used to run guix then was used to compile it. Try to run guix like this: "GUILE_LOAD_COMPILED_PATH= GUILE_LOAD_PATH= guix ..."
<adfeno>quiliro: Reading it now.
<adfeno>quiliro: I wonder if has a queue limit.
<quiliro>adfeno: it is not that....because the email weight is low
<quiliro>4390 bytes it should be able to deliver
<alezost>youri: this can be a guix bug, could you please send a message to <> with that "newer than compiled" guix output
<adfeno>quiliro: Not exactly size limit.
<adfeno>quiliro: ... but sending or queue limit (number of messages).
<youri>alezost: I won't be surprised if it is me doing weird stuff
<adfeno>I have read some email service providers do that.
<youri>I'm just getting started
<quiliro>but i can still send other messages
<quiliro>just some messages don't get delivered
<ng0>in that case I would ask riseup
<alezost>youri: could you also check timestamps of those files: "stat /gnu/store/7fisf4frrgsjzmknjbab1dal23wxrp8d-guile-2.0.13/share/guile/2.0/ice-9/eval.scm | grep Modify" and "stat /gnu/store/7fisf4frrgsjzmknjbab1dal23wxrp8d-guile-2.0.13/lib/guile/2.0/ccache/ice-9/eval.go | grep Modify"
<alezost>(there is probably a better way to check the time, but I don't know it)
<quiliro>i did not have that problem in trisquel with icedove
<ng0>I think the 'vim-full' description should not be equal to the vim description. I haven't patched it because it's trivial, but I really think the description should reflect the difference, not just the list of inputs
<adfeno>quiliro: Let's see...
<ng0>something like %vim description + "this variant of vim includes all possible build time options etc etc..
<youri>scm: Modify: 2017-02-28 16:32:09.157043253 +0100 go: Modify: 2017-02-28 16:32:09.143709920 +0100 (current time is 20:26, and I am sure I did guix pull after 16:32)
<adfeno>quiliro: I'll look it up briefely.
<youri>alezost: kan I remove the go and will it re-built it autoatmicallty?
<adfeno>It seems to be an IMAP issue (because the IMAP connection gets broken).
<alezost>youri: no! don't remove anything from the store manually
<alezost>youri: I see the problem: everything in the store should have an epoch date (1970-01-01) but you have 2017-02-28 which is definitely a problem
<adfeno>It happens during a data transfer (as seen in "IMAP> [data").
<adfeno>↑ quiliro
<youri>A crap
<youri>i know what it is
<alezost>youri: that's why it tells that .scm is newer than .go
<youri>I did extract the .xz archive
<youri>and I did not explicitly preserve timestamps
<alezost>yeah, that's it
<youri>damn, that just took me 10 hours and 29 minutes
<youri>you rock
<alezost>:-) glad to help
<youri>very glad people are willing to do this in their own time, thank you so much
<adfeno>quiliro: Which version of Claws Mail?
<quiliro>adfeno: version 3.14.1
<alezost>youri: no problem, you are always welcome to ask here if you have questions
<Apteryx_>Hi! I'm trying to cleanup a bit the way we use XML_CATALOG_FILES across our packages. Ludovic pointed to me that libxml2 defines it and can be added as an input to provide it. This is already neater than redefining ourselves everywhere, but libxml2 is a 10MiB dependency which is not useful for many packages which require XML_CATALOG_FILES. Thus, I'm trying to simply set a package's "native-search-paths" field to that of libxml2:
<Apteryx_>Unfortunately it fails with "ERROR: Unbound variable: libxml2", and I don't understand why. libxml2 symbol is defined in (gnu packages xml), which the wm.scm module uses.
<Apteryx_>From the REPL it seems like evaluating the native-search-paths from either libxml2 or i3-wm (with my modification) results in the same thing.
<snape`>Apteryx: you link does not seem to work
***snape` is now known as snape
<snape>or Apteryx_
<Apteryx_>snape: Hm. Works here :o.
<Apteryx_>How about this one:
<snape>this one is good!
<Apteryx_>OK :)
<Apteryx_>Would anyone know why "(native-search-paths (package-native-search-paths libxml2))" is not valid? To see the full context: