IRC channel logs


back to list of logs

<brettgilio>rekado: Whatever your solution is, I plan on stealing it. Because I currently dislike the solution of propagating.
<bandali>brettgilio, will do shortly, thanks :)
<leoprikler>is there a better way than propagation for elisp though?
<leoprikler>(in the general case)
<bandali>would be nice if loaded..
<brettgilio>bandali: Yeah it s been having issues today
<brettgilio>Want me to throw you the debbugs url?
<bandali>sure, that'd be great
<bandali>brettgilio, thank you
<brettgilio>bandali: It's mostly the last two paragraphs I think you'll find interesting, but let me know what you think :)
<bandali>brettgilio, certainly ambitious, but i do very much like it! thanks for sharing :-)
<bandali>i would love to be able to all of my FM work in guix
<bandali>... to *do all ...
<brettgilio>Agreed. The two concepts seem synergistic to me.
<brettgilio>We need a Formal Methods committee for guix.
<rekado>sorry about issues.*
<brettgilio>I say that half-heartedly, but who knows.
<rekado>seems like fibers is borked and I can’t build it with guile 3 to see if that fixes it.
<leoprikler>Formally prove and synthesize g-exps?
<rekado>so now it’s running without fibers and leaks fds.
<brettgilio>leoprikler: I was mostly meaning getting a lot of formal methods utilities packaged in guix, but I wouldnt be opposed to seeing things like gexps and guile monads formalized too.
<brettgilio>bandali: ^
<brettgilio>Formal Methods in Guix Committee when?
<brettgilio>Ludo' works in INRIA, he should be co-signing this inititive. haha
<brettgilio>I'd love to see a "Formal Methods" section on the "Guix in your field" portion of the website
<bandali>so, as it stands, there's no reasonably small ML compiler written in, say, C?
<brettgilio>bandali: It seems not. But I'm not opposed to trying to write one. Doesnt need to be complete like mrustc is as say an alternative compiler. Just a compiler to bootstrap the big compilers.
<bandali>brettgilio, right
<brettgilio>I kind of wonder if the nyacc goal of a small compiler could be useful here
<brettgilio>I honestly might weigh down our guix-devel mailing list proposing a formal methods committee not gonna lie.
<brettgilio>everybody needs more emails from me. (jokes)
<digash>- 'compress-documentation' phaseoutput (`/gnu/store/6rnhfi8zrdajbahbq9ii8bpj7mwfcjba-info-reader-6.6') is not allowed to refer to path `/gnu/store/ziinjmbnq004866mwjrczsk12wf35qb8-perl-5.30.0'
<digash>build of /gnu/store/nnfvr10713khxp1bkbpwi5wxbj8cm4wb-info-reader-6.6.drv failed
<digash>still stuck on this
<digash>tracked it down to this code:
<digash> #:disallowed-references ,(assoc-ref (package-inputs texinfo)
<digash> "perl")
<digash>oot@xyz ~# guix refresh --list-transitive info-reader
<digash>info-reader@6.6 depends on the following 23 packages: pkg-config@0.29.2 ncurses@6.1-20190609 tar@1.32 gzip@1.10 bzip2@1.0.6 xz@5.2.4 file@5.33 diffutils@3.7 patch@2.7.6 findutils@4.6.0 gawk@5.0.1 sed@4.7 grep@3.3 coreutils@8.31 make@4.2.1 guile@2.2.6 bash-minimal@5.0.7 ld-wrapper@0 binutils@2.32 gcc@7.4.0 glibc@2.29 glibc-utf8-locales@2.29 perl@5.30.0
<bandali>brettgilio, hehe. more seriously though, i'd love to find other fellow formal methods people in the guix community. it may not be a bad idea to email guix-devel, cc'ing/naming me if you wish, and see if there are any others? :-)
<digash>looks like info-reader depends on perl, but also disallowed?
*rekado restarted mumi / issues.*
<brettgilio>bandali: I will draft an email tonight with some proposals.
<bandali>are these things best called "committees", or some other thing? working group, perhaps?
<brettgilio>rekado: thank you
<bandali>brettgilio, great!
<brettgilio>working group is probably better
<brettgilio>committee sounds white-collar and makes me cringe quite frankly lol
<brettgilio>we have working groups for other parts of guix as is. like rust bootstrapping, wip-gnome, mes so on
<brettgilio>so the "Formal Methods Working Group" sounds great.
<bandali>haha yeah
<brettgilio>bandali: I will definitely mention your name. Would be very valuable for others to see that there is more than just me wanthing this
<brettgilio>Anyways, gtg for now. Stay tuned for the email
<brettgilio>also there is this guy who is a big free software activist and formal methods researcher. I dont think he uses guix, but if he knew about it I bet he would.
<bandali>brettgilio, thank you :-) and thanks for the link, will check it out!
<brettgilio>bandali: on thought. We might be able to start an official gnu project on writing a bootstrapping compiler for ML
<brettgilio>Tied to guix
<bandali>brettgilio, i was actually thinking of that a few minutes ago :-)
<leoprikler>we should call it mml :)
<bandali>i think it would be a worthwhile contribution/addition to the gnu project
<brettgilio>Mml. Haha
<brettgilio>I'll include this thought in the formal methods in guix working group email
<brettgilio>leoprikler: what is the first m?
<bandali>minimal, perhaps?
<leoprikler>that, but it was actually a jab at our boostrapping stuff (mes et al.) happen to all start with an m
<leoprikler>at least as far as I'm aware
<bandali>yea. i think MES stands for Maxwell Equations of Software
<brettgilio>I thought it stood for Mostly Esoteric Stuff. Hahaha /s
<leoprikler>one does not exclude the other
<bandali>another good one could be cml, for C ML compiler, also short for caramel ;-)
<leoprikler>then you should get a rust component into that for crml
<bandali>lol :D
<brettgilio>leoprikler: lmfao
<brettgilio>I'm going to cook dinner. Be back later
<leoprikler>or you write it in the CR language, which is a variant of whitespace
<leoprikler>the language consists of only carriage returns and line feeds
<bandali>i guess caramel also works for a lisp implementation
<leoprikler>and compiles into guile bytecode because why not?
<leoprikler>or caram.el an Emacs Lisp library
<bandali>what a cool robust name, if i say so myself ;-)
<leoprikler>speaking of which
<leoprikler>you could make an emacs lisp library of the biblical angels, e.g. gabri.el
<bandali>yup, there's a lot of 'em *el
<leoprikler>Btw. in the Emacs conference, it was said that Emacs could perhaps use a new extension language.
<leoprikler>But before that, they should implement an easy loader for extension languages (elel.el)
<leoprikler>in my overly long attempt at no longer funny comedy
<bandali>i was following it very seriously until elel.el :-p
<Blackbeard[m]>leoprikler: I disagree with the person that had that opinion
<Blackbeard[m]>That person also told me he thinks Emacs should be for programmers
<Blackbeard[m]>I think emacs is an excellent tool for writers
<Blackbeard[m]>And lisp is easier to understand for non programmers than other languages
<leoprikler>Well, I certainly disagree on the "Emacs is for programmers" thing if it is meant exclusively. (That said, Emacs is also for programmers and we should not forget support for the various tools they need.)
<leoprikler>As far as programming Emacs itself is concerned, I wouldn't trade Lisp for another language family, but I would trade Emacs Lisp for another Lisp.
<Blackbeard[m]>leoprikler: yeah a better lisp would be nice.
<brettgilio>leoprikler: agreed. I've been saying for a long time I'd rather see another lisp over work on concurrency
<brettgilio>I think both are needed. But I'd rather see a new lisp
<leoprikler>I'd really like to see Guile Emacs to do well, but that has faded into obscurity after some while. I also think Guile's features can make for a way smaller C core.
<brettgilio>bandali, leoprikler: I'm working in that email now
<bandali>brettgilio, +1
<bandali>i was looking at and
<brettgilio>bandali: check out Robert harpers cmu website tio
<brettgilio>Or MLton.orh
<brettgilio>Whichever one it is
<brettgilio>I'm going to cc you in this email
<brettgilio>And Leo too
***oriansj` is now known as oriansj
<brettgilio>leoprikler, bandali: You should have the email now
<bandali>brettgilio, confirmed; will read shortly!
<brettgilio>Hope it reads well
<bandali>brettgilio, it reads great; thanks a lot for putting this together and sending it!
<brettgilio>bandali: am really happy to. I'd love to see this become a trend in formal methods. I think our thesis and desires goals make a lot of sense.
<bandali>brettgilio, cheers to that!
<bandali>i also took the liberty to register #guix-fm
<bandali>but first let's see what the folks reading guix-devel think about the proposal
<brettgilio>Thank you for registering that. Agreed on waiting and seeing. We might be surprised at the interest in a good way.
<bandali>cheers, and fingers crossed!
***jonsger1 is now known as jonsger
<brettgilio>janneke: Hey, are you online?
<brettgilio>janneke: NVM. I emailed you.
<alextee[m]>hi guix, i want to package this
<alextee[m]>which file should it go in?
<alextee[m]>oh, guix doesn't have yarn?
<bandali>isn't that an electron app?
<alextee[m]>it's not an app, it's only a website that you open locally and use it as a client
<alextee[m]>parabola has it too
<bandali>hmm, i'm not sure
<alextee[m]>actually, i think i don't need yarn, i was misled by this
<alextee[m]>the web package only does cp and install so it's probably fine
<alextee[m]>guix doesn't have npm either lol
<bandali>yes, the entire js ecosystem and packaging is a mess
<alextee[m]>i know, but there's no other good matrix client out there....
<alextee[m]>i really don't like quaternion
<bandali>yeah... i've kind of given up on it for the time being
<bandali>mostly just use irc
<alextee[m]>it looks like there existed a "guix build-system node"
*alextee[m] uploaded an image: Screenshot from 2019-12-16 04-25-48.png (30KB) < >
<peanutbutterandc>`guix install glib gtk+` tells me that the profile contains conflicting entries for glib. Is this a bug, or am I not doing something right?
<brettgilio>peanutbutterandc: likely right. But it should just be a warning that resolves
<peanutbutterandc>brettgilio, sorry, I don't understand. It exits with a return value of 1. `guix pull` didn't fix anything either
<alextee[m]>just sent a patch for python-sphinx-intl
<alextee[m]>btw, i also noticed guix doesn't have pandoc
<alextee[m]>what's up with that?
<alextee[m]>or is it ghc-pandoc hmm. i dont know what ghc is supposed to be
<peanutbutterandc>Okay so glib works just by installing gtk+ (because propagated inputs) and there is no need for a separate installation of glib
<peanutbutterandc>(Thank you for coming to my TED talk.)
<alextee[m]>peanutbutterandc: yeah i only have gtk+ too, it brings in glib with it
<peanutbutterandc>alextee[m], If I am not mistaken, propagated inputs are not recommended, right? The proper way is substitutions and-or wrappers?
<alextee[m]>no idea, i'm still confused as to what's the difference between propagated inputs and native inputs
<alextee[m]>or simple inputs *
<alextee[m]>i think inputs = depends on the package but doesn't need to provide it in its store
<alextee[m]>and propagated inputs = bring the package along in your store
<peanutbutterandc>It seems that simple inputs aren't directly made available in the profile. They don't "pollute" it. But inputs are hidden away inside /gnu/store - nicely tucked in
<alextee[m]>ok yeah ^
<peanutbutterandc>It seems the maintainers (I know atleast nckx does) prefer anything to propagated-inputs because of the 'conflicts' and it seems it goes against the idea of functional package management
<alextee[m]>i guess its main use is for meta packages
<peanutbutterandc>alextee[m], I'd never thought about that before. Yes, it does make a lot more sense now. Thank you
<alextee[m]>but don't take my word for it, i'm just as confused :-)
<brettgilio>alextee[m]: ghc is the Haskell compiler. And I pushed your commit. Thanks
<alextee[m]>ah great thanks
<alextee[m]>anyone else get this? `- 'compress-documentation' phaseoutput (`/gnu/store/6rnhfi8zrdajbahbq9ii8bpj7mwfcjba-info-reader-6.6') is not allowed to refer to path `/gnu/store/ziinjmbnq004866mwjrczsk12wf35qb8-perl-5.30.0'`
<alextee[m]>happens with `guix install info-reader` after the install phase
<efraim>It's a known issue. We're working on it
<efraim>Disallowed references is needed for cross compiling but it creates issues for regular use
<alextee[m]>ah cool
***Server sets mode: +cnt
<nixo_>Ehy guix, anybody knows what SOURCE_DATE_EPOCH should override? I'm following civodul suggestion to patch julia to use it, instead of using libfaketime. So far so good, matched jl_gettimeofday. Julia compiles, tests run successfuly.. but it never ends. Looking at the source, it's calling a function that does this: while (time() - start) < waitfor. Since time is always the same, this won't end
<nixo_>but IMHO is their bug. They should use a monotonic clock there
***janneke_ is now known as janneke
<civodul>Hello Guix!
<janneke>guix pull: error: qt: unbound variable
<efraim>referenced from where?
<efraim>IIRC we have qt-4 and qtbase
<janneke>nevermind, a collegue couldn't get things to work, but had all kind of .go files littered through their worktree; shadowing guix stuff
***ng0_ is now known as ng0
***jonsger1 is now known as jonsger
<zap`>Hi guix!
<zap`>Is anyone going to chaos congress this year?
<efraim>git grep guile-gcrypt | grep define | wc -l => 3
<civodul>zap`: not me, but it looks very nice
<civodul>efraim: what conclusion would you draw from this? :-)
<efraim>civodul: we should add 3 more for guile-gcrypt-next ;D
<mjw>BTW. I setup my own mirror which others are happy to use if they keep tripping over savannah outages.
<mjw>echo '(list (channel (name 'guix) (url "git://")))' >> ~/.config/guix/channels.scm
<mjw>Of course I guess that means you'll have to trust me. Or is there some signed tags in git that shows that is an official channel mirror?
<zap`>civodul: Yeah it's super awesome. There will be NixOS assebly. I thought that it would be cool to see guix people there too.
<leoprikler>iirc git signatures are not that reliable
<leoprikler>you could still provide a fake repository with outdated or perhaps even bad code. On top of that, I don't thing guix verifies the signatures
<leoprikler>(or even requires them in the first place)
<civodul>mjw: neat, thanks!
<civodul>leoprikler: yeah, that's the whole problem :-/
<civodul>zap`: i guess you should email guix-devel asking who's going
<mjw>civodul, also first bug report :) You can also use https to clone guix from that URL, but guix pull doesn't seem to like that (it apparently sees the HTML page). Using git:// works fine though.
<zap`>civodul: thanks for suggestion
<civodul>mjw: weird, does it pay attention to user-agent or somethign?
<mjw>civodul, it is cgit, it implements dumb HTTP transport
<civodul>but then why would "git" behave differently from "guix"?
<mjw>civodul, I don't know, does guix simply spawn git?
<civodul>mjw: no, it uses libgit2, which is a different implementation
<civodul>so it could be that it doesn't send the same "user-agent" or "accept" headers, for instance
<alextee[m]>oh no i stumbled upon this python bug
<alextee[m]>hope someone updates it to 3.8 :-) it sounds like an arduous task
<mjw>civodul, I cannot find how cgit does that.
<mjw>But I think I really should implement the git smart http protocol on my server anyway. That might magically fix it.
<mjw>(and should be more efficient than the dumb http transport)
<jahor>Hello. Is it ok that substitutes are updated without any additional information?
<leoprikler>yes, this is "normal" behaviour
<jahor>Thanks. Why in quotes? Guix community considers building everything from source?
<civodul>new post!
<snape>alextee[m]: python 3.8 is in Guix already
<leoprikler>Well, spamming the same message over and over serves no real purpose other than confusing user and leading them to ask the question you've asked.
<jahor>I see :)
<leoprikler>As for why it is spammed to stdout like this: substitutes are always updated first when considering a package to build and since this step is done for multiple packages it is the only output you will see before things start actually building.
<leoprikler>(or is it stderr? not sure)
<jahor>I have one more question. Manual recommends doing export PATH="$HOME/.config/guix/current/bin:$PATH"
<jahor>export INFOPATH="$HOME/.config/guix/current/share/info:$INFOPATH" when speaking about guix pull but installation chapter says nothing about is. Is it necessary?
<snape>civodul: very nice!
<snape>jahor: it seems useful indeed
<snape>I think it's already done for the users of Guix System though
<alextee[m]>snape: oh thanks, guix pull must have missed it a few hours ago o.o guix search tells me 3.7.4
<raghav-gururajan>Hello Guix!
<leoprikler>as snape said, your ~/.guix-profile is already added for you
<raghav-gururajan>I have a patch in ".diff" format. I was told to do `git am --3way --directory=~/patchfile.diff`. But the process is still going on and does not seem like anything is progressing. What should I do?
<leoprikler>check the diff file for errors
<leoprikler>sometimes line wrapping in emails can break diffs
<jahor>Cool answers as usual : D
<snape>well I'm not sure what leoprikler meant by '.guix-profile is added'
<raghav-gururajan>I am not sure what to look for in the diff file
<snape>because we were talking about .config/guix/current ;)
<raghav-gururajan>It received the diff file as sttachment via email.
<jonsger>civodul: nice post :) but I spoiler read it already in the artwork repo :P
<raghav-gururajan>I downloaded and placed it in my home dir
<nixo__>Hi civodul! I was working ont he julia SOURCE_DATE_EPOCH thing. But the more I play with the time, the more bugs in julia I find .-. (see I'm pausing this for a couple of days (also beacuase for some reason, even if time() now returns values according to SOURCE_DATE_EPOCH, mtime of cahce files is still correct, and I don't know how that's possible. But since faketime was
<nixo__>working, probably there's another function used to get the time? or maybe libfaketime does something more. I'll try to figure this out)
<raghav-gururajan>leoprikler Can I pass --ignore-space-change, --ignore-whitespace
<leoprikler>No, because lines are broken.
<leoprikler>jahor: btw. Guix system also handles ~/.config/guix
<snape>which is a discrimination against foreign distros users
<leoprikler>snape: I meant 'the PATH variables for ~/.guix-profile are expanded bla bla bla'
<civodul>nixo__: sure! did you check if shows any patches, for instance?
<leoprikler>IIRC there is a gist for foreign distro users that adds an /etc/profile.d/
<snape>well it's not documented
<raghav-gururajan>leoprikler I see. When I opened the diff file, the lines does not seem to be broken.
<leoprikler>that's because it has only recently been shared with us on IRC
<snape>leoprikler: jahor wanted to see the Guix's info page. And it's not in .guix-profile unless you install the 'guix' package, which is recommended against
<leoprikler>yeah, but ~/.config/guix/current is handled by Guix System at least
<snape>indeed :)
<leoprikler>Which again leads to the question whether or not jahor is using that or a foreign distro.
<snape>anyway I think it's a bug
<snape>I'll report it some day
<leoprikler>On a foreign distro, you pretty much have to emulate our /etc/profile
<leoprikler>and on Guix System, if you don't already have your ~/.config/guix/current set up (i.e. before your first pull)
<leoprikler>you will have to set up those variables as well
<leoprikler>(or log out and back in)
<nixo__>civodul: yes, with no luck. But since they do not provide packages in their repo, the patch we need is useless for them
<str1ngs>also on foreign disto you need to do some tricks to get locales working system wide
<jahor>Well, I'm getting started with GuixSD.
<snape>yes, but there is a reason why it differs with GuixSD about locpath
<snape>however there is no reason why it would differ about 'guix pull'
<snape>s/there is/I can see ;)
<jahor>Sorry for a dumb question. But how do I mention people in IRC?
<snape>you just say their name
<leoprikler>just write their name
<jahor>Oh thanks
<jje> i can't reconfigure due to this
<leoprikler>jje +1
<snape>jje: what does 'guix describe' say?
<snape>(and leoprikler)
<jje>one second
<leoprikler>I don't have a clean pull on this machine, but on the other one I pulled 4 hours ago.
<jje>guix describe
<snape>thank you, it's broken upstream actually
<jje>oh ok
<jje>also why can't my system find opensmtpd this morning when sudo herd status says it is running as a service.
<snape>you don't need to have service executables in your PATH to make them work
<snape>with the herd
<snape>jje, leoprikler, you can 'guix pull' again, maybe it'll work ;)
<leoprikler>brb debugging
<rekado>I’m trying to use guile-next with mumi, but I’m stuck on gnutls.
<rekado>1) gnutls somehow doesn’t benefit from the 2.2 -> 3.0 rewrite, so I had to add it to the inputs of mumi; 2) it fails trying to connect to debbugs over HTTPS: “Unbound variable: set-session-server-name!”
<leoprikler>Hello from Polari :)
<brettgilio>rekado: thx for the fix on guile studio
<alextee>is there a way to take a screenshot of a context menu?
<alextee>gnome screenshot ignores it
<alextee>oh it works if you take the whole screen instead of just the window
<alextee>i guess the context menu is considered a separate window since it can extend beyond its parent
<leoprikler>Perhaps, but it will likely require a program, that uses tool-specific magic to figure out *only* the context menu.
<leoprikler>IMO taking an image of the entire desktop and doing some imagemagick is likely the better option.
<leoprikler>(If you wish for a generic tool, that is.)
<alextee>oops, i meant to ask in #gtk sorry
<jahor>Is it ok to do (keyboard-layout (cons foo '(bar)))? Or what is a right way of doing setxkbmap -layout "by(latin),ru" -option "grp:toggle, grp_led:scroll, ctrl:swapcaps"?
<alextee>hexchat auto-jumped me here
<leoprikler>jahor: as far as I'm aware Guix only changes the layout.
<leoprikler>meaning if you want those group options you have to find some other way to append them to setxkbmap
<jahor>I see
<jahor>Still how do I set two or more languages to layout?
<leoprikler>You don't, at least not in Guix itself. GNOME and other DEs might have tools for that.
<jahor>I guess I should autostart this by XMonad... Thanks anyway :-)
<leoprikler>"this" being the keyboard layout thing or IRC?
<jahor>Layout thing
<Minall>Hello guix!
***ng0_ is now known as ng0
<leoprikler>So Polari can be closed while still being connected to IRC, but all of the buffers will be emptied.
<leoprikler>Which is quite a non-feature if you ask me.
<Minall>leoprikler: Wait, there's polari in guix?
<leoprikler>I wrote a package for it and I'm about to send it.
<Minall>leoprikler: I can't believe it!!!! I want to test it right now!!!!!!!!!!
<leoprikler>Sadly I have to gc before I can lint for some weird reason.
<Minall>Please send it !!!, I do use polari
<Minall>THat's strange...
<Minall>So, you're going to send it and then how long would it take for it to be upstream?
<Minall>I'll like to test it since I do like polari
<leoprikler>Usually a week or two, but you can grab the patch and put it into your local guix build or some channel.
<Minall>Or you can pass me the .scm file too!
<Minall>Where can I find it?, is it on git?
<leoprikler>I wrote it in gnu/packages/gnome.scm
<leoprikler>meaning I cloned guix and patched it in-place, so to say.
<snape>and if it's a simple package it can be faster ;)
<leoprikler>It also depends a little on the maintainers' whims. If one of them happens to want Polari for a long time, they'll pick it up quickly.
<snape>or any reviewer
<leoprikler>Not sure if I used up all my luck through telega, or if me waiting a week for some shell extension still has my karma in balance.
<leoprikler>By "maintainer" I meant "people with commit access", but yeah, reviewer would likely have been more correct.
<leoprikler>People mistaking their browser for IRC. 😢️
<leoprikler>Using Polari's feature to clear the chat history (and thus my mind), I have achieved ultimate enlightenment!
<digash>alextee[m]: looks like the issues with info-reader is fixed!
*alextee[m] reconfigures
<digash>efraim: thank you for fixing info-reader, it bit me when I was installing new os on the laptop :)
<Minall>leoprikler: So I shouold just clone guix and get gnome.scm with polari?
<leoprikler>clone guix, git am the patch and install through ./pre-inst-env guix
<leoprikler>for which you'll have to `./bootstrap && ./configure && make` inside `guix environment guix`
<Minall>Thanks leoprikler !
<leoprikler>Let's go linting
<leoprikler>Okay, lint results are harmless. Time to send mail.
<alextee[m]>digash: how did you use the new info-reader? i am doing sudo guix pull and then sudo guix system reconfigure /etc/config.scm, but i still get an error on info-reader
<alextee[m]>i can see the commit in master, weird
<leoprikler>guix pull as normal user
<leoprikler>or wait, is this the installation medium?
<leoprikler>(well, there sudo is just noise, since you're root anyways)
<alextee[m]>no i'm just upgrading my system
<alextee[m]>trying now with normal guix pull
<digash>alextee[m]: i've build a USB image, boot into it and then ran guix system init as root
<digash>alextee[m]: it worked today, but had the same issue with info-reader yesterday
<jje>today emacs is telling me this so how can opensmtpd not be in the store but still running as a service?
<str1ngs>jje: maybe you need to guix reconfigure?
<jje>yes i just did and rebooted.
<str1ngs>or emacs needs to restart if the opensmtpd input has moved
<str1ngs>also guix update?
<nixo__>note: currently hard linking saves 106530.35 MiB
<nixo__>(I've not rung guix gc in 6 months :D) But doesn't it --optimise automatically?
<str1ngs>nixo__: as I recall guix does not run gc automatically
<nixo__>str1ngs: not the garbage collection, just the optimization thing
<nixo__>The daemon performs deduplication after each successful build or archive import
<str1ngs>nixo_: based on that then it seems it should optimze on it's own accord.
<digash>Does GuixSD support LUKS encrypted root / partition?
<janneke>Digit: yes
<janneke>we call it the Guix System nowadays, however
<janneke>digash: see "Mapped Devices" in the manual
<digash>janneke: thank you, just making sure that i am not going down a dead end, set it up with mapped devices but have issues booting, probably something else
<digash>after successful guix system init config.scm /mnt got to grub error no such device: 96d1-...
<janneke>digash: i am not using anything special for grub, simply (target "/dev/nvme0n1")
<digash>janneke: so you do not have mapped devices and encrytion setup?
<janneke>digash: i do, but grub only needs the target and finds my mapped-device and ecrypted root
<nixo__>can I set native-search-paths to non-existing dirs?
<nixo__>julia has the special empty element that is expanded automatically
<janneke>nixo__: no, non-existing dirs are pruned; you may have to use a wrapper?
<nixo__>janneke: uh, is it done manually like in packages/uml.scm:69 ?
<digash>janneke: maybe you can spot the problem?
<sneek>Welcome back nly, you have 1 message.
<sneek>nly, str1ngs says: oh great . is that when exporting to html?
<janneke>nixo__: yes, or use `wrap-program'
<mehlon>how is Guix on ARM? has anyone tried installing it on a rpi?
<nly>what's the standard way to request for package updates?
<bavier>nly: ask here :), or if you want to include rational, can send to
<nixo__>janneke: thanks!
<janneke>digash: it used to be that as mapped-device (source "/dev/xxx") did not work
<janneke>digash: i am using (source (uuid "50d96f54-1dbb-48f8-bca5-2f1feb5ff144"))
<janneke>with uuid the output of cryptsetup luksUUID /dev/nvme0n1p1
<janneke>...but you mentioned a grub error; i don't do separate boot, nor efi...
<nly>thanks bavier
<nly>i'd like to update emacs-shroud and nomad soon
<nly>hi janneke since you are here
<janneke>hi nly
<nly>do you have a moment
<janneke>yes ... i'm bootstrapping :)
<nly>i want to have the docs for emacsy online
<janneke>ah, we only have a git archive, right
<nly>i did the same for nomad
<janneke>nly: ah, nice
<nly>it's not too much work since emacsy already has nice docs :)
<nly>i'll bother you if i hit a roadblock
<janneke>sure, that's great!
<janneke>nly: i really dropped/paused the ball on emacsy, with exciting progress on bootstrapping
*janneke goes afk for a bit
<apteryx>hello :-)
<apteryx>wouldn't it be nicer if file-name-predicate would match a regexp against the absolute path of a file instead of its basename?
<apteryx>that would enable more possibilities when specifying regexps in a search-path specification, for example.
<apteryx>it's trivial to fix (remove the basename call in the file-name-predicate procedure of (guix build utils) but means rebuilding the world (core-updates).
<apteryx>I believe the change would be backward compatible too. Correct me if I'm wrong.
<str1ngs>janneke: both nly or myself I think can contribute anything we need upstream to emacsy if you are busy
<KE0VVT>What's Emacsy?
<str1ngs>emacsy is a guile library to create emacs like applications
<str1ngs>it provides things like keybindings, buffers, windows. it has a nice abstraction. in that it's not dependent on any GUI toolkit either.
<janneke>str1ngs: yes, i feel that way too -- but i don't like to think that i'm letting you down
<str1ngs>janneke: nope not at all. the only think we might need immediately is static html docs for emacys. maybe later when re visit documentation introspection. or I'll probably run any major API change by you.
<str1ngs>won't be for some time yet.
<nly>sounds good to me
<apteryx>ugh; it wouldn't be backward compatible...
<apteryx>matching the start of a file name with ^ would no longer mean the same thing
<str1ngs>janneke: I think it would be okay to generate the static manual @ wdyt?
<dongcarl>Hey all, is there any way to `guix graph` a manifest easily? I don't see it in the help...
<leoprikler>apteryx: you could try a two-pass solution or a conditional one depending on whether or not the name contains a slash
<leoprikler>dongcarl: Currently not. You'd have to evaluate the manifest and transform it into a list of packages for that to work.
<janneke>str1ngs: yes, suer
<str1ngs>okay thanks
<janneke>str1ngs: thank *you*
<Parra>how can I set an environment variable during build time?
<Parra>and how I can get pwd?
<Parra>there is some place where I can find all those questions?
<Parra>Like.. finding the framework and functions that I can use
<Parra>I don't know even what IDE should I use, or autocomplete..
<Parra>I find those questions looking for other packages implementation in the repo... :/
<bavier>Parra: some of those answers can be found in the 'guile' manual
<PotentialUser-54>Parra: If you're into Emacs, I suggest you to install geiser ( and emacs-guix ( Then your quality of life as a guile developer will be increased by a lot. If you're not using Emacs then I don't think I can advice anything
***PotentialUser-54 is now known as valignate
***valignate is now known as valignatev
<Parra>I want to try emacs, but it's so vi like, I'm more used to nano
<bavier>fortunately nano has decent scheme syntax highlighting
<Parra>I mean, I just like that style, I come from Visual Studio C++ which is very different, I moved to Linux many years ago but I only use vi like mode if I only have that option, and emacs seems similar to vi, maybe it's time to start learning it in deep
<valignatev>I'm not sure what do you mean, but emacs is very different from vi in a sense of having a modal editing workflow
<valignatev>e.g. you don't have to enter an "insert mode" to write text
***drrrty is now known as drrty
<Blackbeard[m]>Parra: what? Emacs bindings are very similar to nano
<Blackbeard[m]>If anything emacs is more like nano than vim
<Parra>I don't know, for me was strange the first contact
<Parra>anyway, I will try the emacs-guix, and try to find more info in guile web
<Blackbeard[m]>Parra: but why?
<Blackbeard[m]>It is not been actively developed
<Blackbeard[m]>Why not learning emacs from emacs ?
<bavier>Blackbeard[m]: you're thinking guile-emacs I think
<Parra>because I don't care about emacs right now, I have been using too many IDEs, I'm tired of learning new IDEs, I just want to develop guix fast and have a good enviroment
<Parra>but meh: ++learned_ides;
<drainful>Emacs could be the last one
<drainful>But yeah, it's probably not worth picking up just for one task
<Parra>I did a bug report recently which was solved by @brettgilio recently, about ruby compilation (static instead of dynamic)
<Parra>I'm doing the same right now for NodeJS too
<Parra>Nodejs is compiled statically as a design decision but my software needs it as a dynamic library
<Parra>can this be interesting for guix? should I make another bug report or patch?
<Parra>I don't know many softwares that embed NodeJS thought
<Parra>but it may be interesting in any case, linking dynamically is the standard behavior of Guix
***drrrty is now known as drrty
<raingloom>what am i supposed to set NIX_REMOTE to if i wanna `guix import nix` something? i installed nix through guix system wide and the daemon is running.
<raingloom>ok, nvm, i thought i was supposed to be a path
<raingloom>so setting it to "daemon" is enough, but then why can't i import frama-c...
<raingloom>hm. looks like i'm using it correctly, it's just buggy.
<raingloom>welp, gonna just do it by hand.
<raingloom>i guess the real `guix import` was inside me all along.
<Blackbeard[m]>bavier: ah yes I read it like that. Hahaha my mistake
<valignatev>raingloom: I couldn't figure out nix import yesterday as well. I tried to import it from the local copy of nixpkgs though, didn't want to spin a whole nix stack for it. Ended up "importing" the package by hand as well :)
<raingloom>hmm, which package has the manpages for stuff like calloc?
<raingloom>doesn't seem to be in glibc...
<leoprikler>+1 on nix import being badly documented
<lispmacs>hi, I've got a HP LaserJet Professional P1102w USB printer, which worked for me under Debian 9 (free system) but in Guix, cups Web administration tool does not see it. Do you have any ideas for me?
<civodul>raingloom: it's "man-pages" :-)
*civodul -> zZz
<raingloom>ah. makes sense.
<pkill9>lispmacs: does dmesg show any errors?
<lispmacs>pkill9: I'm not seeing any errors. dmesg showing that the device is found, with serial no, etc.
<pkill9>lispmacs: do you have the hplip/hplip-minimal extensions added for the cups service?
<lispmacs>pkill9: wait, I just found an hplip package, not yet installed
<lispmacs>i think that is what I need
<lispmacs>installing now with my fancy emacs-guix interface
<pkill9>you need to add it to your cups service as shown in the manual:
<pkill9>installing it as a package won't work
<lispmacs>oh, okay, that would make sense
<lispmacs>still adjusting mentally to the Guix way
<wdkrnls>Hi Guix, can anyone tell me what package the urandom man page is in? I don't seem to have it available on my Guix system.
<wdkrnls>thanks :)
<lispmacs>I was just wondering if somebody was working on (or planning to work on) some kind of package install queue interface? I have this frequent problem where I'll start a 10 minute package install, and then two minutes into that decide I want to install another package as well.
<lispmacs>but by the time the first installs are done, I forget to install the other thing I wanted
<leoprikler>you can start the second installation right away
<leoprikler>it will build things in parallel rather than sequential, but it's super duper safe to do so in guix
<vagrantc>really was happy to see "configuration file" listed with "guix system list-generations" !
<nckx>leoprikler: Have you tried that recently? Guix locks profiles during building now (I understand why, but it broke that common operation for me).