IRC channel logs

2019-06-27.log

back to list of logs

<rvgn>nckx Have you built libreoffice on your ThinkPad before?
<rvgn>Marlin1113: For tablet, there is a dedicated service called "inputattach-service-type". :)
<nckx>rvgn: Sure, overnight. It doesn't take *that* long on an i7, it just takes LibreOffice long.
<nckx>I've also put 16 GiB of RAM in this bad boy, don't know if that makes a difference for LO.
<nckx>rvgn: Note that that's only needed for the old (X200 and earlier) models.
<nckx>X220 and up have a USB-connected touchscreen which works out of the box.
<nckx>efraim: Heads up: I've reverted your two changes to the openssh package since they break the service and lock people out of their boxes.
<rvgn>rvgn Cool!
*nckx 's recent join/part spam may be somehow related ;-)
<Marlin1113>oh, thanks rvgn
<Marlin1113>how about for the tearfree thing?
<rvgn>Marlin1113: Ignore mine. nckx informed that service is for x200 or earlier.
<nckx>rvgn talking to themselves on the Internet again.
<nckx>šŸ™‚
<rvgn>? lol ??
<nckx>(I've missed some history with all those reboots so I don't know which ThinkPads are being discussed.)
<chewzerita>is guix's packaged gnome work with wayland by default?
<nckx><rvgn> rvgn Cool!
<rvgn>nckx Oh damn LoL
<rvgn>That was for you
<nckx>chewzerita: Not ignoring you, I just don't know.
<rvgn>Marlin1113: Not sure about the other thing. But regarding keyboard layout, the layouts can be declared as a list.
<chewzerita>nckx: I used to use tiling wms for a long time, just switched to full bloat: debian with gnome
*nckx literally just finished installing Ubuntu MATE on their mother's laptop because Gnome couldn't handle the heat wave we're having.
<nckx>(No, i3 is not an option for her.)
<rvgn>nckx: xfce?
<nckx>chewzerita: Why'd you switch?
<nckx>rvgn: De gustibus desktopibusque non disputandum.
<Marlin1113>aw
<Marlin1113>what do i do ;-;
<chewzerita>nckx: I wanted everything to Just Workā„¢ (notifications, firefox audio, consistant theme)
<rvgn>nckx: What in the world that means?? xD
<chewzerita>I always did everything on the command line, now the only cli thing I use is beets (http://beets.io/)
<nckx>Yeah. Guix is awesome but does not do *that* yet.
<chewzerita>nckx: what's *that* (I have used and contributed to guix before btw)
<rvgn>Marlin1113: keyboard-layout ("layout1" "layout 2" ... "layoutn")
<nckx>rvgn: Taste is subjective. Also, I've never noticed much difference between the two, and I think MATE is slighty user-friendlier for someone coming from Gnome 3.
<nckx>Marlin1113: Does that pasted configuration work as-is?
<rvgn>Marlin1113: That's the syntax. For clarity, current desktop.scm example in the manual portrays listing multiple keyboard layouts.
<rvgn>nckx: I concur that taste is subjective. But gnome and xfce uses human sensory perception something research data to design the UI.
<rvgn>Since data is based on experimentation, the results are quite objective.
<rvgn>Damn, I lost the reference. Was reading about a while ago.
<nckx>rvgn: Then why do I prefer Mate over XFCE? Checkmate, science nerds.
*nckx actually prefers tiling WMs that don't waste screen space on ā€˜pixelsā€™ but you get my point.
<Marlin1113>hmmm
<tune>tiling WMs are fantastic
<Marlin1113>rvgn i wanna get us qwerty and us dvp
<Marlin1113>i'm trying "us qwerty", "qwerty" and "us" "qwerty", none of them work
<rvgn>nckx: Well, that research did not use good sample population as it was preliminary.
<dadinn>roptat: the glibc-utf8-locales package is installed and it seems the en_US.utf8 typo is not the issue either :(
<rvgn>Marlin1113: Oops! I think the format should be "en_uk.utf8" like that..
<Marlin1113>doesn't work
<Marlin1113>:c
<rvgn>Marlin1113 Just a sec
<nckx>Marlin1113: That's because all of them contain "qwerty", which doesn't exist ;-) It's just "us".
<nckx>And "us" "dvp" for the superior layout.
<Marlin1113>i tried both, didn't work
<rvgn>Marlin1113 My bad, that formart was for locale.Sorry Yes, as nckx mentioned it's just "us"
<nckx>tune: True, until you get a touch screen.
<rvgn>I will send you the link
<rvgn>Marlin1113 (keyboard-layout (keyboard-layout "us" "uk" "foobar"))
<nckx>Marlin1113: "us" didn't work? I find that hard to believe.
<Marlin1113>ooh, i got it wait
<nckx>As the variant.
<rvgn>Marlin1113 Check out: https://guix.gnu.org/manual/en/html_node/Using-the-Configuration-System.html#Using-the-Configuration-System
<rvgn>Marlin1113 View the 2nd example. You will see keyboard-layout declared and how it is also used inside xorg-configuration service.
<Marlin1113>now "us" works, but "us" "dvp" doesn't
<Marlin1113>got it
<rvgn>Marlin1113 Glad :)
<Marlin1113>had to do '("us") '("us" "dvp")
<Marlin1113>needed a quote and parens
<rvgn>(y)
<nckx>(#t)
*nckx needs sleep, sorry.
<nckx>Marlin1113: That alternative syntax seems to me undocumented, or did I miss something?
<rvgn>nckx I wanted to ask you something, but will you be leaving anytime soon?
<nckx>Have you tested it (not just whether it evaluates without error, but if the resulting reconfigured system offers the correct layouts)?
<str1ngs>can someone please test this. guix remove guile; guix install emacsy . then from a new terminal emacsy-webkit-gtk-w-buffers
<str1ngs>it should fail with (no code for module (emacsy emacsy))
<nckx>Looking at (gnu system keyboard) I really don't see any code to handle this or how it could work but as noted, nckx is a šŸ˜“ boy.
<nckx>str1ngs: I'm not going to remove guile, sorry. Could you reproduce that using a (probably pure) environment?
<str1ngs>pure is not enough it still maintains your profile
<str1ngs>I need to determine if this is a guix bug or a emacsy bug.
<str1ngs>plus you can rool back your profile ... so
<nckx>str1ngs: I don't understand what ā€˜maintains your profileā€™ means.
<str1ngs>it means that your profile is still used. pure will not test this bug.
<nckx>The whole point of --pure is to not use another profile.
<nckx>rvgn: Eh. In 10 minutes or so.
<nlyy>1. guix package -S 1(or the oldest gen around) then 2. guix package --rollback (now you have a blank env) 3. guix package - i emacsy (install a guile library) 4. emacsy-gtk-webkit (run code that uses guile lib)
<rvgn>nckx Ah I see. It's about my issue with inputattach-service-type. Can I CC you in an email?
<nlyy>it will fail with no code for module (... )
<str1ngs>#I'm not in the mood for arguing. I have a patch that fixes this already that's been ignored. and now I need to determine which is at fault guix or emacsy. so if you don't want to be help test
<nckx>str1ngs: I am helping, impatient person. I also want to know what you mean by ā€˜[--pure] means that your profile is still usedā€™, since that makes no sense to me.
<nckx>One can do two things at once.
<str1ngs>again, please either test with my parameters since this is the case I'm dealing with
<str1ngs>if you can't test with those parameters that's understandable. maybe someone else is willing to help.
<nckx>^C
<nckx>test it yourself.
<nlyy>my ^ comment was related to str1ngs 's issue, which i was able to reproduce
<nckx>Cool. I gave up trying.
<str1ngs>I have it fails now I need to determine which is at fault guix or emacsy
<nlyy>is it a bug?
<nckx>rvgn: That's fine, but I don't use it?
<nckx>So I won't be offended but there's little point. If it's on a guix ML I see it come by anyway.
<rvgn>nckx I remember us discussing this issue long back. So thought you may have an insight.
<nckx>I remember, but I've never used inputattach myself. It's unlikely.
<nckx>I don't have the hardware to test.
<rvgn>nckx Ah I see. That's okay.
<rvgn>nckx By any chance, do you know who the maintainer of that service is?
<Minall>Hello guix!
<rvgn>Minall o/
<Minall>rvgn: greetings!
<nckx>rvgn: It was added by Tim Gesthuizen <tim.gesthuizen at yahoo dot d e> this year, it hasn't been touched since.
<nckx>Minall o/
<rvgn>nckx Thanks!
<Minall>I'm having some trouble with my configuration :(, I request help
<Minall>if it is possible of course
<nckx>Minall: Don't ask to ask, just ask šŸ™‚ However, I'm shutting down for the night.
<Minall>nckx: haha, thanks!,
<Minall>I don't know how to add some xorg configurations on my config.scm, my driver and my keyboard layout
<Minall>How can I show you my config?(I know it's a little stupid, but I really can't add those 2 options)
<PotentialUser-25>when you rewrite a package dependencies, does it mutate it in place >'
<Minall>I know of the existence of some command that allows me to upload the file
<Minall>Maybe with -cat-
<nckx>Minall: Try guix install wgetpaste && wgetpaste < your-file.
<rvgn>Minall You may use https://bin.disroot.org to show your config. :)
<Minall>Which one do I try? welp, I'll go with wgetpaste first
<nckx>Either wgetpaste or manually pasting it into a pastebin in a Web browser is fine, I just thought wgetpaste might be ā€˜easierā€™.
<nckx>As the topic mentions, we encourage the use of paste.debian.net (which is also what wgetpaste uses now IIRC) because it's friendly to Tor users; I don't know that about bin.disroot.org but would be disappointed if they weren't.
<Minall>I'm installing wgetpaste now!
<Minall>Wow, what a nice command you showed me, wgetpaste!
<Minall> https://bpaste.net/show/3514d91537d4
<Minall>That is my config
<Minall> I'm a little new, so if you see something strange, I'm still a noob, so please teach me
<ItsMarlin>nckx doesn't work ;-;
<ItsMarlin>it had an error in the end
<ItsMarlin>i'll leave this stuff for tomorroe
<ItsMarlin>tomorrow*
<ItsMarlin>just got home from a trip
<chewzerita>Just listened to https://librelounge.org/episodes/episode-23-guix-with-ludovic-court%C3%A8s.html pretty cool
<dongcarl>Hey all, trying to pic up mikegerwitz's changes here: https://lists.gnu.org/archive/html/guix-patches/2018-03/msg00049.html
<dongcarl>I can't figure out what Ludovic is saying...
<dongcarl>Why does this fail for him?
<emacsomancer> rvgn: nice! I didn't realise disroot hosted a pastebin
<efraim>nckx: thanks for taking care of it
<nckx>efraim: I would have done more if I'd had time.
<efraim>I probably should've committed the two openssh changes in the opposite order
<nckx>Never trust a trivial clean-up commit...
<civodul>Hello Guix!
<civodul>efraim, rekado_: i'm reconfiguring bayfront
<efraim>ok, I was building out to gobject-introspection on core-updates to see if it was still a problem there
<efraim>but I stopped it just now
<civodul>yeah, i don't know how that happened
<civodul>but i fixed it
<civodul>i'll post the details on guix-sysadmin
<efraim>good news (maybe?) the CVEs affecting python and python2 don't affect the ones on core-updates
<civodul>efraim: cool!
<civodul>though we'll still have to rebuild stuff with Guile
<civodul>rekado_: logs.guix.gnu.org works (but doesn't do the right thing, i think)
<roptat>hi guix!
<roptat>I found a newer guix in the overdrive's store, I'm using it to pull the latest guix
<roptat>it built python, so I think I'm good, I just need to be very patient :)
<civodul>heh :-)
<civodul>there was something wrong with your overdrive?
<roptat>openssh doesn't start at boot
<roptat>so I'm trying to upgrade guix and reconfigure
<roptat>because you told me the bug was supposed to be fixed :)
<jonsger>roptat: you need the fix for http://issues.guix.gnu.org/issue/36395
<roptat>when was that pushed?
<jonsger>roptat: revert commit 10h ago, which fixes the bug
<roptat>so I think I'm good, I started guix pull two hours ago
<jonsger>then you should
<jonsger>it costs me like 1 or 2 hours yesterday...
<roptat>ok, I'm not at home anymore, so I'll have to wait until this evening to see the result
<roptat>I should add myself to the list of sysadmins too...
<roptat>is it ok to push a new configuration file for that machine, adding myself as an admin? should I find it a proper name, or can I call it "overdrive2"? :)
<rekado_>civodul: Iā€™ll take a look at this logs issue.
***jonsger1 is now known as jonsger
<jonsger>heh, yes
<civodul>roptat: sure; factorization with the other overdrives would be cool, but a committed file is better than no file :-)
<civodul>nckx: speaking of which, did you eventually manage to get your overdrive up and running?
<civodul>let us know if there's anything we can do to help
<civodul>it'd be great to have it build stuff!
<civodul>for those who haven't seen it: https://wingolog.org/archives/2019/06/26/fibs-lies-and-benchmarks
<civodul>cbaines: we can have data.guix.gnu.org point to milano anytime you want now
<roptat>civodul, also I'll be moving this summer, so my IP will probably change again
<roptat>and the server will probably be done for a few days too, because I have to think about where to put it and how to bring a cable to it
<roptat>I'll remind you when the time comes :)
<civodul>roptat: sounds good!
<roptat>civodul, have you looked at the small patch I sent yesterday about guix pull suggesting running guix pull?
*civodul checks
<civodul>roptat: just replied!
<roptat>thanks
<rekado_>civodul: I totally overlooked your big derivation-build-plan patches!
<rekado_>civodul: they just popped to the top of my inbox because of your note about having pushed them.
<rekado_>civodul: I just wanted to say that I like these changes, especially that <derivation-input> contains a <derivation> now.
<rekado_>(I guess Iā€™ll need to change cuirass to adapt to these changes)
<roptat>do we want the "delta" to be purely static, or do we want to allow user content (comments, stars, ...)?
<cap>Greetings. I'm trying to define a package. How do I compute the checksum of a git repository?
<roptat>usually I write the package definition and put the hash of some other package, then guix build -S it, and use the actual hash that is shown at failure
<roptat>:)
<rekado_>roptat: Iā€™d do without.
<rekado_>comments mean that you have to filter spam and remove harrassment; stars are without context (what version do they apply to? What do they mean?).
<rekado_>cap: guix hash -rx
<cap>Thank you both.
<civodul>rekado_: cool!
<civodul>normally Cuirass doesn't need to be changed
<civodul>it's all backwards-compatible
<civodul>roptat: i guess the "delta" could be developed incrementally: let's just make it a dumb registry at first, and eventually add more features
<civodul>it'll have to be moderated anyway, so we might as well make a pretty static version of it initially
<roptat>ok, I'll try to have a first proposal before this week-end
<rekado_>civodul: for the display of dependency failures in Cuirass I had to deal with derivation-inputs and read the derivation files.
<rekado_>so I guess this has to be modified now that derivation-input no longer holds a file name.
<civodul>rekado_: as long as you use derivation-input-path, that's fine
<civodul>if that code was matching on the record type, then it's indeed broken
<civodul>those record types should never have been exported
<civodul>rekado_: BTW, the ability to view which dependency failed is... great :-)
<rubic>sshd has just stopped working for me. 'sudo herd start ssh-daemon' -> "Service ssh-daemon could not be started".
<civodul>rubic: apparently a bug was introduced a couple of days ago and fixed a day later
<civodul>so you might need to pull && reconfigure
<civodul>rekado_: we'll have to update build-farm.el to take advantage of the new features (search, failed dep, etc.)
<civodul>it seems alezost is less active these days
<rubic>civodul: thanks (me = relief)
<civodul>yeah, that's unfortunate
<user_oreloznog>Hi Guix! Discovering Guix (on foreign Debian distribution) ... I installed it with the automatic install script 'guix-install.sh' and it works.
<user_oreloznog>A look at https://www.gnu.org/software/guix/manual/en/html_node/Application-Setup.html#Application-Setup
<user_oreloznog>... And I asked myself a question:
<user_oreloznog>Is it necessary to launch the nstd daemon yourself before using Guix?
<user_oreloznog>If yes, should it be launched from the host distribution or from guix's sheperd?
<roptat>debian probably already has it, no?
<user_oreloznog>I think so :)
<user_oreloznog>S I can launch nstd from debian ?
<user_oreloznog>So
<roptat>I don't remember having to run anything from guix before using it on my fedora
<user_oreloznog>ok
<roptat>I don't remember having to run anything special from my system either
<user_oreloznog>It's ok since i have installed it :)
<user_oreloznog>Thanks roptat!
<roptat>basically the first paragraph says if /var/run/nscd/socket exists, you're fine
<user_oreloznog>:-)
<roptat>mh... I don't have it on my fedora, but it's probably elsewhere, because I don't have any network issue with my guix packages
<user_oreloznog>In debian, I don't have it apparently...
<user_oreloznog>And no network issue here... :)
<user_oreloznog>Maybe I'll try to install nscd...
<roptat>I have a /var/db/nscd/hosts
<user_oreloznog>oh...
<rekado_>nscd is necessary as a ā€œbridgeā€ for non-local user accounts.
<user_oreloznog>ok...
<roptat>oh, so if you only have local accounts, it's not necessary?
<rekado_>if your host system uses SSSD or talks to LDAP or whatever then the glibc from Guix would not be able to ā€œseeā€ those accounts as it cannot load the binary plugins needed for that.
<rekado_>but it *can* talk to NSCD over a socket and have NSCD take care of the details.
<rekado_>if you only have local accounts NSCD might not be needed, but Iā€™m not sure about this.
<rekado_>(there could be other problems than just account discovery and authentication)
*rubic ssh-daemon service is working after pull && reconfigure
<user_oreloznog>rekado: thanks for these explanations...
<davexunit>speaking of nscd... does anyone have good knowledge of how it works and/or has anyone ever written a custom pam module?
<davexunit>I have an issue where nscd seems to forget about users (whose details are acquired via a custom pam module) a little while after they've logged in. presumably cache expiration.
*rekado_ doesnā€™t know
<davexunit>was worth a shot. :)
<civodul>davexunit: what do you mean by "forget"?
<civodul>nscd does some caching by default
<civodul>you can run "herd statistics nscd" to see what's up
<davexunit>civodul: what happens is that upon login my custom pam module will validate the user account using a custom remote data source, but at some point after login when I try to run a command the shell responds with something along the lines of "I don't know who the heck you are".
<davexunit>running "id my-user-account" results in calls that lookup user info again and I continue my session until the next time it forgets.
<davexunit>so it seems like I am missing something in the shared library that implements all the _nss_* functions
<davexunit>unfortunately this low-level stuff is mostly undocumented. the glibc manual has a small section about it but mostly I had to crib from other modules that figured some of it out already.
<davexunit>anyway, not looking for anyone to solve this problem for me. but if anyone happens to know of good reference material for this sort of stuff, I'd be happy to receive a link to it. :)
<civodul>documentation is not that bad (info "(libc) Extending NSS")
<civodul>but yeah, it can be tricky
<davexunit>I referenced that section a lot, but it is really light.
<davexunit>the function names are all templates and figuring out what the function name should actually be was difficult.
<cap>I'm forced to install iwlwifi. I looked it up in the logs and on the mailinglist but I don't understand there explanations. Does someone knows a tutorial graded for guix beginners?
<Marlin1113>hey guix
<Marlin1113>there's something wrong with one of my packages :c, i'm trying to figure out which one it is
<Marlin1113>when i try to guix pull from my testing repo i get
<Marlin1113>(repl-version 0 0)
<Marlin1113>(exception out-of-range (value "bv-u8-ref") (value "Value out of range: ~S") (value (1)) (value (1)))
<Marlin1113>this is the log, so i don't really understand what's happening
<Marlin1113>i'll have to debug each package
<bavier>Marlin1113: maybe a missing sha256 ?
<Marlin1113>oh, i'll take a look
<Marlin1113>oh, missing sha :P thanks bavier!
<bavier>Marlin1113: np ;)
<bavier>not sure if we could provide a better error message...
<Marlin1113>Maybe put in "missing sha256?"
<rvgn>Hello Guix!
<quiliro>saluton
<rvgn>emacsomancer Yep :)
<rvgn>Hello Folks! Do anyone used GNU Privacy Assistant (gpa) in Guix System? For me, it's not even running. :(
<quiliro>Kial homoj estas en #guix?
<Marlin1113>hey guix, what is the recommended way of testing packages?
<Marlin1113>atm i have a local repo which i edit, commit and pull, but this seems pretty inneficient
<bavier>Marlin1113: if you put the repo path in GUILE_LOAD_PATH you should be able to use 'guix build foo'
<Marlin1113>not sure if i understand what you mean
<rekado_>Iā€™m comparing potential build nodes and one of the options has a dual CPU (24 core per CPU) AMD EPYC 7451. I wonder a) if maximizing cores is a good idea for the build farm, and b) whether I should pick 128GB RAM or 256GB.
<rekado_>any ideas?
<efraim>Is there a big difference in the single core performance of the machines?
<efraim>As much as I hate it some phases and some whole builds are single threaded
<efraim>128/24 is 5.333, I would think we'd be hard pressed to be memory starved, /tmp in RAM?
<Marlin1113>i guess it would allow more stuff to be built at the same time, which is a advantage in terms of time usage
<rekado_>efraim: single core boost is significant, but Iā€™d want to just increase the number of parallel builds.
<rekado_>we may get about 23 of these nodes
<jonsger>rekado_: only for guix?
<rekado_>yes
<jonsger>so 23x 2xEpyc 7451
<rekado_>yes
<rekado_>1104 cores.
<jonsger>uff, that is really nice
<Marlin1113>dang, guess the money is being put in good use :P
<Marlin1113>there are some epyc models with 32 and 64 cores i think, take a look to see which has a better price/performance
<jonsger>rekado_: I don't now how your "time pressure" is but maybe it's worth to wait on Epyc 2 (Rome): https://www.computerbase.de/2019-06/amd-epyc-7002-rome-spezifikationen-preise/ they are around the corner
<rekado_>we have a big spreadsheet where we compare AMD and Intel server CPUs by price/100MHz
<rekado_>jonsger: yeah, itā€™s too bad, but we have to spend the money now or not at all :-/
<jonsger>oke, then the 7451 is maybe a good choice
<rekado_>Marlin1113: thereā€™s an EPYC variant with 32 cores which we are also considering, but the base frequency is quite a bit lower, which would probably be noticeable in the many builds that are single core only.
<jonsger>are you intending to make the build workers only with one core or with more?
<jonsger>maybe it would be an option two have to machines or so with EPYC 7371 16x3.1/3.6/3.8GHz to get some builds faster
<ItsMarlin>i don't know if EPYCs also do it, but threadrippers have a way of boosting single core by using multiple cores
<ItsMarlin>dunno much about it
<rekado_>jonsger: not a bad idea.
<jonsger>btw 18core/72threads POWER9 sells at 1425$ without VAT from the US :P
<rekado_>jonsger: we get really big discounts from Dell; for POWER9 weā€™d need to have a public tender.
<ItsMarlin>power9 is great
<jonsger>rekado_: the POWER9 "tip" wasn't so serious :P
<jonsger>maybe I should ask my colleagues at SUSE, because we use also Epycs...
<Marlin1113>where does most of the money come from?
<sammich>guix in funded by inria no?
<rekado_>Marlin1113: our instituteā€™s budget :)
<rekado_>sammich: I donā€™t think so. Inria employs one of the maintainers.
<sammich>oh i see,
<Marlin1113>apart from packaging, what more can i do?
<rekado_>Marlin1113: patch review, for example
<rekado_>or development of our tools, such as mumi and cuirass.
<rekado_>hereā€™s mumi: https://git.elephly.net/?p=software/mumi.git
<Marlin1113>i'm still learning programming stuff tho :c
<Marlin1113>i still have issues with packaging
<str1ngs>Marlin1113: don't have to be a programmer to contribute. there is always documentation and translation. and bug reporting.
<rekado_>it might be easier to learn programming when you have a real problem to solve.
<rekado_>some parts of Cuirass that need improving are restricted to the web interface, which isnā€™t very difficult to change.
*rekado_ ā€“> afk
***drrrty is now known as drrty
<rvgn>Hello Guix!
<Marlin[m]>Howdy frend
<sammich>i was looking to write a zig package for guix, it has a check if the running system is guix and uses $NIX_LDFLAGS and $NIX_CFLAGS_COMPILE does guix set equivalent flags does anybody know?
<rvgn>Do you anyone know free software desktop client for facebook?
<nckx>sammich: No, just use CFLAGS &c. zig using NIX_* is probably a bad idea.
<nckx>Goes to show that if you expose something, people will use it...
<sammich>nckx: ah cool, does guix set LDFLAGS i presume it does it seems like a normal thing to set
<nckx>sammich: I don't *think* so. Is it? When you build a package by hand there's no global LDFLAGS floating around either. It should be set by the package's build system. That said, some packages do need tweaks added so you'll find plenty of LD/CFLAGS examples in gnu/packages/*.scm.
*nckx is less sure about CFLAGS though.
<rekado_>nckx: http://logs.guix.gnu.org works now; the HTTPS version has to wait until some other time.
<nckx>\o/
***ChanServ sets mode: +o nckx
***nckx changes topic to 'GNU Guix | 1.0.1 is out! get it at https://guix.gnu.org | videos: https://guix.gnu.org/blog/tags/talks/ | bugs and patches: https://issues.guix.gnu.org | paste: https://paste.debian.net | Guix in high-performance computing: https://hpc.guixsd.org | This channel is logged: http://logs.guix.gnu.org/'
<rekado_>nckx: thanks!
<nckx>rekado_: There's a stray .log.tmp file in there by the way.
<rekado_>Iā€™ll add the code to maintenance.git soon
<jonsger>t
<jonsger>t
<jonsger>^was only a test about how it creates the links for two messages with the same time stamp :)
***ChanServ sets mode: -o nckx
<rekado_>itā€™s primitive!
<rekado_>and it swallows the spaces before and after URLsā€¦
<jonsger>but this time stamp link stuff is not a real problem :)
<rekado_>and it lacks an actual stylesheet
<jonsger>but it's super fast :)
<rekado_>heh, even though itā€™s really naive
<rekado_>I guess Guile is just really fast
<sebboh>hi all. I don't think I know how to handle my guix environment variables. I'm using guix on a foreign distro. I especially don't know how to handle the environment variables when I'm using `sudo guix ...`.
<sebboh>I have to manually export GUIX_LOCPATH= periodically to silence some 'guile: warning: failed to install locale' messages. But even when I silence them successfully during normal operation, they come back when I use `sudo guix ...`. I could surely tell sudo to preserve the environment, or manually pass some variables, or whatever, but I thought I'd check here for advice first.
<sebboh>Frankly I don't like having two copies of guix installed on this foreign distro.. one for uid=1 and one for uid=1001...
<rekado_>hereā€™s the code in case someone would like to make the logs prettier: https://git.savannah.gnu.org/cgit/guix/maintenance.git/tree/hydra/goggles.scm
<rekado_>sebboh: you donā€™t need to have two versions of Guix installed.
<roptat>when searching for a build on berlin, how are results sorted, and how can I get results for a specific package (e.g. "coq" but not "coq-mathcomp")?
<rekado_>roptat: ^coq$ might work
<rekado_>or rather ^coq-1.2.3$
<rekado_>or whatever the version is
<quiliro>Kial homoj estas en #guix?
<rekado_>roptat: the order is by build number.
<quiliro>245 homojn
<rekado_>246ā€¦
<quiliro>rekado_: ho!
<quiliro>:-)
<rekado_>:)
<rekado_>roptat: ^ and $ are the only special characters that the search supports, though. You can further restrict the results with ā€œspec:ā€¦ā€ and ā€œsystem:ā€¦ā€.
<sebboh>rekado_: for my foreign distro guix, I just used the ... the download a script and pipe it into bash method. Which I wasn't able to find just now. Was that method deprecated?
<roptat>rekado_, thanks
<civodul>rekado_: nice work on IRC logs!
<quiliro>Mi estas komenzanta esperantisto. Sed mi volas kontribui al la tradukon de la Guix-a manualo. Kio estas la proceso kaj la iloj por ĝin? Mi uzas la Emacs-on
***ChanServ sets mode: +o civodul
<civodul>oh the entry message already has the new URL for logs, wonderful :-)
<civodul>quiliro: mi kredas ke oni nur devus traduki en lingvon ke oni bone konas
<quiliro>civodul: la traduko de Google-o estas fari. Mi preferas la traduko de komenzanto.
<quiliro>s/fari/farita/
<ngz`>I just had an idea about "guix search". Once the list of matches is built, would it make sense to re-order them by graph depth, i.e., with those that are not inputs of any result above, then those inputs of one result, etc.
<ngz`>For example "guix search ivy" should return "emacs-ivy" before "emacs-ivy-yasnippets".