IRC channel logs


back to list of logs

***zacts is now known as zacts_pi
<usney4>okay guys
<usney4>it is finally up and running
<usney4>how do I install a browser?
<rvgn>usney4 `guix package -i icecat`
<nckx>usney4: ‘guix install icecat|…’
<rekado>usney4: there’s icecat, epiphany, eolie and more
<usney4>thanks guys
<usney4>how do I add whiskers to the panel?
<usney4>the menu thingy
<usney4>why is it taking so long to install icecat?
<usney4>how do I search for packages?
<nckx>usney4: ‘guix package -A NAME-REGEXP’ or ‘guix search REGEXP | less’ (searches descriptions, too; very verbose).
<nckx>And Icecat took 6 seconds to install here, so… you'll have to help us help you. What do you see?
<nckx>(Granted, I only needed to download 2 substitutes @ 39 MiB. You likely have fewer dependencies pre-installed.)
***drrrty is now known as drrty
<Minall>Hello guix!
<Minall>Is there a os-probe method that I can use on guix?
<lfam>Minall: What do you mean?
<Minall>I have 2 drives, one with guix, and one that I use to test distros, I want the guix one to be able to detect the other distro
<vagrantc>os-prober is available, but i've never figured out how to use it
<nckx>And it isn't magically integrated in any way with Guix.
<Minall>yep, I don't see any way to add it on the manual
<Minall>nckx: o/
<vagrantc>should really file a wishlist bug about that
<Minall>Now what... I can't test the new installed system...
<nckx>Minall: Why not? You should be able to choose a boot drive when your PC boots.
<Minall>When I try to do that it doesn't boot grub, so I thought I could do it from guix
<vagrantc>i have a system with both Debian and Guix System on it ... but i manually drop to the grub shell to load debian's grub.cfg
<nckx>You can. You can load the other grub by running ‘configfile (hdX,Y)/boot/grub/grub.cfg’ in a GRUB command line.
<Minall>Well that's interesting, maybe I'll try that
<nckx>Normally, I'd say ‘add that as a menuentry to Guix's grub.cfg’, but Guix's grub-configuration is… extremely limited. I don't know how you would.
<vagrantc>my debian install is in lvm, too, so i have to also "insmod lvm" from the grub prompt ... but it works, and it just means i boot guix more than debian on that system :)
<vagrantc>thought I filed a bug about that at some point, but can't find it ...
*vagrantc re-adds it to the list of things vagrantc will likely forget
*dongcarl is building core-updates...
<Minall>Ok I'm back
<Minall>It doesn't work
<Minall>But it's also the installation, my pc has a strange bug that if I plug a debian usb, it just freezes, and I'm unable to do anything, so I have to reboot manually, just by plugin a usb with debian or anything that bases on it
<Minall>It's strange, but welp, no problem at all, I was just going to test something, but nope
<rvgn>dongcarl Nice!
<dongcarl>just building to test my new package works... hopefully I can get it in before freeze deadline...
<usney4>nckx icecat took for ever to install
<quiliro>yep...icecat and epiphany are biggies
<quiliro>the best is to have an offload server to do the heavylifting
<nckx>quiliro: They should have received substitutes though. The latest icecat is built.
<quiliro>nckx: even though a small machine takes a lot of time
<nckx>quiliro: I don't understand that last remark in the context of available substitutes. Do you mean the overhead of extracting the substitutes & profile generation & such? That can be slow (it is here, with my 215-package manifest) but it shouldn't take the 1+ hour that it apparently did.
<nckx>With a new(ish) profile (a safe assumption), it should take minutes at most.
*nckx is now too tired for Esperanto 🙂
<rvgn>dongcarl Good luck :)
<dongcarl>built! and pushed...
<vagrantc>dongcarl: several of the build-dependencies for guix are now in Debian unstable ... still need guile-git, guile-ssh and scheme-bytestructures ...
<vagrantc>but it's coming along...
<dongcarl>vagrantc: That's really good to hear... Reminds me I need to put up a PPA
<vagrantc>oh, and guile-gnutls ... which might be the total blocker in the end
<dongcarl>or APT repo
*vagrantc wonders if it would be possible to build guile-gnutls out-of-tree
<quiliro>nckx: what i mean is that there are always builds to make even if there are substitutes...i understand that they could be done by offloading
<dongcarl>vagrantc: out-of-tree?
*dongcarl is a n00b
<vagrantc>dongcarl: it used to be built from the gnutls sources in Debian, but was removed due to issues in guile that have (presumably, hopefully) been solved years ago
<vagrantc>e.g. guile 2.0 vs. 2.2
<dongcarl>vagrantc: Oh, is guile 2.2 packaged in debian? If so, why not put it back?
<vagrantc>so i was wondering if there was some way to make an alternate source package that just build-depends on gnutls-dev or something and build it
<vagrantc>dongcarl: the gnutls maintainer in debian is so burned in the past on it that i don't think they want to even think about it
<nly>is packaging on Debian harder?
<nly>ah i get it
<vagrantc>it has it's issues
<nckx>quiliro: Sure, we're probably talking about the same builds. These shouldn't take ‘forever’, but there's no way to tell without more info.
<vagrantc>in this particular case, it's one of the things i like about guix ... you can build multiple packages from the same sources and they don't interfere with each other
<nckx>dongcarl: ‘gnu: Update to 0.18.0.’
<dongcarl>vagrantc: That's really unfortunate...
<vagrantc>in debian, normally if you build N variants of a package, it needs to be built from the same source package ... e.g. dynamically applying and unapplying patches, etd.
<nckx>See, now you'll get yelled at.
<dongcarl>nckx: OH NOES
<nckx>(No you won't.)
<dongcarl>can someone add a `bitcoin-core` in there? I don't wanna force push on my first day
<nckx>dongcarl: We don't force-push, period.
<nckx>It is there for eternity.
<dongcarl>vagrantc: Okay, well we can maintain our own APT repo and prove that it works, until the gnutls maintainer relents
*vagrantc will try to do a lightning talk at debconf this year about all the things vagrantc forgot to mention (and has since learned) since giving the "My Crush on GNU Guix" talk last year
<dongcarl>nckx: eternal shame, lol
*nckx watched the original and looks forward very much to the rev.
<vagrantc>dongcarl: yeah, it might come to that
*dongcarl actually upgraded all of GNU to 0.18
*dongcarl is looking forward to watching that too
<nckx>dongcarl: There's an ‘gnu: 0.32: Update to $package’ of mine in there somewhere. Welcome to the club of shame.
<quiliro>0.18? don't you mean you mean 1.0.1?
<quiliro>woops, i'mlistening to myself speaking and that comes up in my writing!
<dongcarl>quiliro: nah, GNU v42 is where it's at
<quiliro>dongcarl: what package?
<nckx>quiliro: bitcoin-core. It was just a typo.
<quiliro>i did not know GNU had versions
<dongcarl>quiliro: Haha the actual package I upgraded was bitcoin-core
<dongcarl>Yeah I made a silly mistake
*dongcarl drowns in eternal shame
<quiliro>thaugh it would be cool to have a GNU version
<quiliro>useless but cool
<nckx>quiliro: We'd need more cool features for that. Although a debugged ‘guix deploy’ would definitely be bump-worthy IMO.
<quiliro>dongcarl: thank you for clarifying
<dongcarl>GNU UBI (user-binary interface) v1
<nckx>* M-x butterfly reference.
<quiliro>guix deploy? GNU UBI? what are those?
<dongcarl>Well I literally just made up GNU UBI
<dongcarl>but `guix deploy` is new, and is supposedly very useful
*dongcarl needs to read up
<nly>does anybody use system wide tests?
<quiliro>butterfly? mi estas en alia mondo kaj mi pensas ke sonĝas
<nckx>I'm guessing your not subscribed to guix-devel 🙂
<dongcarl>nothingmuch: You might be interested in
<nckx>nly: I'm running ‘make check-system’ at this very moment if that's what you mean.
*nckx just updated GRUB.
<nly>i am not aware of that, what does it check?
<nckx>quiliro: Carl's ‘UBI’ quip made me imagine grey-bearded hackers twiddling bits by hand which reminded me of this comic
<nckx>It's all very unimportant, sorry.
<nckx>nly: It's a test suite for (gnu system), basically ‘Guix System’, using mostly VMs.
<nckx>‘make check’ tests Guix-the-package-manager, ‘make check-system’ checks Guix-the-distro. Whether it boots and starts X and runs mail servers well &c.
<nly>thanks that's it
<nly>thanks dongcarl. i found this,
<nckx>nly: That's awesome and horrifying.
<dongcarl>it is really awesome tho not gunna lie
<dongcarl>very upset they don't have "blockchain" on there
<nckx>(The horrifying is ‘This only works if your *browser* supports ligatures’ which implies all kinds of horrible things about how fonts on Teh Modern Web work, but that's not their fault.)
*nckx tries typing ‘html5’ in there; is satisfied.
<nckx>dongcarl: 2017:
<quiliro>will guix version 4bac5bba51383f5c01cd5a9a5efe7b9198be86c0 have guix deploy?
<quiliro>i will just tab
<quiliro>nope... does not
<quiliro>guix pull time
<quiliro>will have to close all windows so it does not take that long
*rvgn is suprised that nckx has not gone to sleep yet xD
*nckx isn't.
*nckx is trying to figure out how locale data installation works in Guix System.
<rvgn>nckx nice1
*nckx conses ‘eo’ onto %default-locale-definitions.
<nckx>Although I'm still unsure how or whether "UTF-8" is needed or even desired here. glibc just calls it ‘eo/’.
*dongcarl was gunna go to bed but remembered glibc 2.27 is broken in core-updates
***amuck_ is now known as amuck
<zacts_pi>nckx: I got my laptop early today
<zacts_pi>I'll try to test guix again sometime by the weekend
<nothingmuch>dongcarl: iirc the plan is to use ipfs? i don't recall where i got that impression though, perhaps librelounge podcast?
<roptat>hi guix!
<roptat>(if I remember well, you're into this sort of fonts :))
<dongcarl>Weird behavior on core-updates: `./pre-inst-env guix build gcc-toolchain` gives me `gcc-toolchain-9.1.0`...
<rekado_>dongcarl: is it weird?
<dongcarl>rekado_: I thought the new default version was 7? Or am I mistaken?
<rekado_>dongcarl: that’s just the default for the *-build-system environments.
<rekado_>you’ll even get 9.x on master right now
<dongcarl>rekado_: Oh I see
<dongcarl>that makes sense, thanks!
<rekado_>roptat: heh, nice. I’m on a limited train wifi now and save my bandwidth for the texlive work, but I’ll take a look at this later in the hotel.
*rekado_ packages dozens of hyphenation pattern packages…
*rekado_ is really bored
<rekado_>only 52 more!
<efraim>If you're still bored after that you can check the usages of string-append and see if they should be git-file-name instead
<rekado_>efraim: yeah, I’m replacing most texlive packages on the wip-texlive branch.
<rekado_>many of them should really be merged, others should not exist at all (texlive-dvips).
<rekado_>gotta switch trains now
<bricewge>I would like to add `` to `/etc/pam.d/login`
<bricewge>Looking at looked at `login-pam-service` and `unix-pam-service` it seems to me that it's not possible to do so with `modify-services`. How can I do it then? Do I have to patch Guix?
<bricewge>s/Looking at looked/Looking/
***OrangeShark2 is now known as OrangeShark
<raingloom>heyyy, which package has the gio command? it's apparently not in gvfs.
<raingloom>oh wait, nevermind, i think i found it through Arch's pkgfile
<raingloom>...nope, false alarm
<bricewge>raingloom: It's in `glib:bin`
<raingloom>bricewge, thanks! i forgot about the multiple outputs thing >_>
<raingloom>ok, well, it says "unable to find supported ssh command" when i try to mount an sftp endpoint
<ruffni>hello. i'm having the issue where icecat does not display ANY number-characters (see here: any ideas on how to fix this?
<ruffni>sorry, link doesn't work.. it's
<raingloom>can the openssh-configuration be empty? i'm getting some huge error if i leave it empty, even though i don't want to modify any of the defaults
<nckx>ruffni: Please be aware that is a horrible image host (they won't even display your image if your viewers disable/limit JavaScript, like I do). I wish I knew a better one, but I don't.
*nckx AFK but that sounds like bug-guix material, raingloom.
<raingloom>nckx, nvm, i think it was a syntax error, i'm just not used to guile's syntax errors yet
<ruffni>nckx: i could also describe the image :) it's the english wikipedia page on Pi viewed on my guixsd installation running icecat. instead of each numeral character there is a wide white space. unfortunately it's hard to use the machine when i can't verify whether the ip address in my address-bar is correct or not :)
<raingloom>nckx, or if you were referring to the gio error, looks like setting up the ssh service fixed it
<raingloom>but this should be better documented tbh. or the gio error should be better. i'd contribute but i'm not sure if it's a glib or a guix issue.
<raingloom>ruffni, nckx, btw i have the same error in Icecat, in case more eyes are needed for the bug
***spk121_ is now known as spk121
<ruffni>raingloom: do you use a workaround?
<raingloom>ruffni, with icecat? no, i used to use netsurf for a while.
<roptat>ruffni, have you tried to install fonts in your user profile? have you tried "fc-cache -f"?
<ruffni>i don't think so. do i need any specific font packages?
<ruffni>just ran fc-cache -f ... should i restart icecat?
<roptat>I think so
<ruffni>nope, fc-cache -f didn't help. also, i think some special characters like # and * are not shown either...
***emyles` is now known as emyles
<quiliro>saluton amikoj
***cehteh_ is now known as cehteh
<janneke>the `location' in channels, is that the subdirectory to look in?
*janneke wonders how to specify a subdirectory in a git archive
<nly>janneke i am also interested in this if you find this out
<janneke>nly: it's not implemented, the code has a FIXME about this
<janneke>nly: ...but wait, maybe there's a hook already: build-aux/build-self.scm
*janneke is gonna try that
<rvgn>Hello Guix!
*rvgn is very excited about today's core-updates --> master merging \o/
<efraim>I thought today was the freezing, I didn't realize it was ready
<rvgn>I was scheduled for merging today right? o.O
<rvgn>*It was
<mbakke>rvgn: It was scheduled for a 'feature freeze', there is still a long way to go before it can be merged to master.
<rvgn>mbakke Oh :/
<mbakke>rvgn: Substitutes should be available in a week or so, it would be great if you could `guix pull --branch=core-updates` then and report on anything that is not working ;-)
<rvgn>mbakke Cool! :)
<str1ngs>janneke: while you are here :). I was thinking of proposing a buffer-kill-hook method for emacsy what do you think?
<janneke>str1ngs: does emacs have that? then: always OK
<janneke>sounds reasonable, i'd like to find out how emacs handles such things, that's all
<str1ngs>janneke: the issue is scm_from_pointer (buf, pointer_destroy_cb) is reliant on guile's GC
<str1ngs>janneke: for GUI pointers this is problematic because there is lag when a GUI component is destroyed.
<janneke>str1ngs: right
<str1ngs>I'm not sure if this translates will to emacs, since emacs does not handler pointers or foreign objects. only in the context of modules, where are a special case
<janneke>yes, i see
<janneke>let's do it!
<nly>in Emacs there is a hook kill-buffer-hook
<str1ngs>I'll create a feature branch, so you can review before merging. and I'll update the examples to properly destroy GUI objects. this should fix the scm_from_pointer FIXME as well
<str1ngs>nly: thanks for the reference nly :)
<jonsger>hm glibc-2.29 fails on core-updates for me...
<mbakke>jonsger: How does it fail?
<mbakke>jonsger: Ah right, you need the patch from <>.
<mbakke>(that failure is "just" for glibc-locales)
<jonsger>oke, I'll try it with that now :)
<mbakke>I realize we should have pushed the Python 3.7.4 release candidate, since we were waiting for the final one anyway.
<mbakke>Updating now unfortunately requires a full-rebuild change.
<mbakke>jonsger: Last one, promise! :P
<kkebreau>I'll go ahead and push the glibc-locales patch, since it fixes things on the current branch and we can revert changes if need be.
<jonsger>kkebreau: +1
<mbakke>kkebreau: Great! :)
<ruffni>i fixed my "no numbers in icecat" problem by installing various packages containing the string "font" in their names... i guess there's some missing dependency in the icecat package
<rvgn>Does anyone have experience working with GNU Privacy Assistant? Hoping someone could fix
***jackhill_ is now known as jackhill
<Dynamicmetaflow>Greetings all! I'm a new user to Guix, been using it for almost 2 weeks on all of my systems and couldn't be happiier. Thank you everyone for their contributions and support!
***Connor[m]1 is now known as Adroit[m]
<bandali>hello Dynamicmetaflow, happy to hear that, and happy to have you aboard!
<Dynamicmetaflow>Hi bandali, thank you! Looking to slowly take my time to study the manual and also Guile.
<bandali>Dynamicmetaflow, cheers!
***Adroit[m] is now known as Connor[m]2
***Connor[m]2 is now known as erudition
<dongcarl>Someone here might be more experienced with `libssp`, but it seems the general recommendation is to use the `glibc` one if available. Would it be appropriate to add `gcc_cv_libc_provides_ssp=yes` to `#:make-flags` in `cross-base.scm` if we're building a cross-gcc with a libc?
<dongcarl>Source for the make flag:
<pkill9>will it ever be reasonably possible to package electron/electron applications?
<jonsger>mhm core-updates produces some heat :(
<Dynamicmetaflow>What do you mean by produces some heat?
<roptat>it means many things need to be compiled ;)
<jonsger>Dynamicmetaflow: I need to compile a lot of heavy stuff
<Dynamicmetaflow>Ah ok, I misunderstood.
<erudition>pkill9: yes please. Chatting from Riot right now, and it's not available on guix, probably due to electron
<Dynamicmetaflow>Has anyone been able to create a package for hugo, the static site generator? I've attempt packaging hugo unsure about how to troubleshoot the backtrace,
<Formbi>erudition: you can install Flatpak (through Guix) or Nix (through a script) and install Riot through one of them
<Formbi>pkill9: I don't think it won't, there are electron packages in Nix
<Formbi>oh, there is Nix in Guix too
<pkill9>Formbi: it looks like nix doesn't build electron from source, it just grabs the pre-compiled binary release
<Formbi>well, yeah
<Formbi>but Guix has ungoogled-chromium
<pkill9>ah, yea
<Formbi>btw I wonder why it's almost never available as a substitute
<pkill9>i'm looking at nix's riot-desktop package, and i can't tell if it compiles that from source or not
<pkill9>maybe the latest build is failing
<pkill9>yea it is:
<Formbi>like almost all build were failing
<pkill9>this is the latest build that succeeds: you can find which guix git commit that one is and then use an inferior to build chromium with that guix commit and hopefully the build server still has it available as a substitute
<Formbi>how to use an inferior?
<jonsger>gcr fails in the testsuit on core-updates: