IRC channel logs

2022-09-29.log

back to list of logs

<civodul>nckx: i started looking at qa.guix a couple of weeks ago and benefit from it indirectly via substitutes on bordeaux.guix
<nckx>itd_: Ah, so it's not reliable? Then I said nothing.
<nckx>podiki[m]: 3.
<nckx>civodul: I saw that! That is great (now that we have the storage…). True, we're all benefiting from that, but I wondered if there were other ways to use it more directly.
<civodul>i also have an keymap in debbugs-mode that jumps there :-)
<nckx>I don't know if the ‘Comparison unavailable’s I'm seeing are transient or just TODO.
<civodul>recommended view: https://10years.guix.gnu.org/video/progress-on-automating-parts-of-guix-patch-review/
<nckx>That would presumably make a big difference.
<civodul>ah yes, i don't know
<nckx>civodul: Thanks.
<civodul>i've seen these sometimes
<nckx>Ah, so they are not universal? Then I just chose a bad link to randomly click.
<nckx>Thanks II.
*nckx snarfs the keybinding idea.
<itd_>nckx: I don't understand the questions, sorry. https://git.savannah.gnu.org/cgit/guix.git/tree/guix/import/print.scm#n50 relies on the scm file path to determine a package module. This doesn't seem to work if the file path has a prefix (e.g. ./; because (. myhello-...) is not (myhello-...)).
<itd_>s/questions/question/
<lechner>Hi, guix package --search=libtool reports version 2.4.7 as available, but I cannot find it in the guix sources. How may I install that version, please?
<zpiro>Perhaps mbakke has some useful insight for addictive mannerisms and convetion that should be promoted as common short hands where the practices is largely missing and highly desired.
<zpiro>So let me ask this, https://www.smartrecruiters.com/CERN/743999852309083-data-acquisition-engineer-ep-lbc-2022-83-ld-, is this something that ussually comes with a professorship and something that i am ocer qualifed for, mbakke?
<jonsger[m]>XXX-done@debbugs.gnu.org or XXX-close@debbugs.gnu.org What is the correct way for closing a bug? I always forget
<pkill9>nice to see talks on guix
<lispmacs[work]>jonsger[m]: -done
<jab>jonsger[m]: use debbugs. It's easier.
<jab>jonsger[m]: when you see the bug in debugs, you can press "C" to send a control message, and it will prompt you...do you want to close? append?
<jab>that's how I prefer to do it.
<jonsger[m]>jab: its an Emacs thingy yeah?
<jab>jonsger[m]: yup! Let me see if I can send you a link...
<zpiro>Most of human history beings, "there are two types of people". In IT that more often involves terrible sysadmins, and even worse programmers. One does not think inside the box, and the other cant think outside the box. The bare minimum involves extorting a bare mnimum of of short hand for system work flow.
<jab>jonsger[m]: https://10years.guix.gnu.org/video/emacs-debbugs-and-public-inbox/ shows some examples of debbugs usage
<zpiro>There should be roomn for pestering about wiriting guides, articles and stories about how to get popular things done and working. At some point it also involves a commnity narrative about common practices and short hands central developers should give up.
<nckx>itd_: Sorry, I was away. I think the JSON importer should be patched in this case. I don't think generating (. hello) is ever a good idea, regardless of how we do (or don't) normalise file names elsewhere.
<nckx>Does that make sense?
<nckx>NB I'm not a user of the JSON API.
<nckx>Good lord. *the importer. I'm distracted again.
*nckx o/
<gabber>what do you think about the idea mentioned in the discussion of Konrad Hinsen's talk to write separate Guix manuals for each target group?
<zpiro>From a personal persctive, not having finisahed high school, mbakke being a firend. Peter M. Hanstee being the uncle of my class mate in Berge. Having worked at CERN, attended RMS lectures, and being primary responnsible for the computing of a LHC experiment. Is bascially, fuck it! this is not normal.
<chungy>gabber: I think that's a great idea.
<zpiro>Neither of us has met Hansteen, but by and lafge not average.
<gabber>what are all the (possible) target audiences? sys-admins, software developers, embedded engineers (?), devops, scientists, everyday gnu+linux-libre users?
<ilmu[m]>that list will never be exhaustive. just write for the audiences that you know and trust others to do the same. the problem should be about organizing the information so that people can do that and that the target audience can find the information they need.
<ilmu[m]>all of this documentation stuff is always just about organizing the information, how to keep it up to date, how to keep it accessible, how to keep it relevant...
<chungy>There's some broad categories that will sweep many of the more niche ones into them. At minimum, I'd say a user's manual and developer's reference.
<sughosha>Hi, I am having problem with %desktop-services. I want to modify gdm-service-type to add (wayland? #t) in %desktop-services, but the problem is that gdm-serviece-type is not declared straight, but as in condition "(cons* (if ...", so I cannot use modify-services to modify it. Any help is appreciated and thanks in advance.
<gabber>maybe my question was more towards: what do you think would be good boundaries for such a guide? i touch most of the aforementioned categories and am enticed by the idea -- it might make advocating Guix to my peers way easier
<ilmu[m]>gabber: I would love to help but my context is incompatible because I've thought too much about this and ended up with an approach (https://matrix.to/#/#datalisp:matrix.org) which I normally would be silent about but right now I'm kinda pissed off that the guix matrix room was destroyed by incompetence so my trigger is a bit loose at the moment.
<ilmu[m]>the problem is that people don't accidentally share knowledge easily enough
<ilmu[m]>we need a tool that is useful and extensible and simpler (i.e. easier to agree on) than emacs.
<gabber>i'm not sure i can follow. WDYM "extensible and simpler [tool]"?
<ilmu[m]>basically I think we should have a "main menu" esque program
<ilmu[m]>and that the behaviours you want to access should be indexed and documented using that tool.
<nckx>The #guix:matrix.org room is fine, or as fine as it can be within matrix.org policy, which is to kill & resurrect rooms for whatever reason. Calling it incompetence is a bit much. I'm sure they have their reasons.
<ilmu[m]>idk why you feel like you have to defend it but it's off topic so I will limit my response to this answer
<gabber>ilmu[m]: i was more thinking about text documents targeted towards, let's say software devs that happen to be interested in the easiest way to build software reproducibly. starting off with why every software developer should care and a few examples on how easy this is in guix towards examples (with references to the relevant sections in the Reference Manual) of more complex use cases and explanations on why some things are done in a
<gabber>specific way in guix
<ilmu[m]>gabber: yeah I know, I am also talking about that just in a roundabout way.. for me opening anything is a behaviour that can be indexed in a menu.
<ilmu[m]>it's like boiling the internet down to just hyperlinks
<nckx>I'm not defending it, but their server, their rules.
<lechner>sughosha: i'm not sure about your situation. this probably does not help you http://paste.debian.net/1255357
<gabber>ilmu[m]: still not sure i can follow -- it all sounds pretty "emacs" to me, but you ruled that out before...? but i agree, interactive discoverability and automatically and discoverable help/documentation is a thing only few softwares manage to excite in
<ilmu[m]>I'm just saying that "package managers" should go all the way to the main menu
<ilmu[m]>it's not enough to bring the software onto the machine I also need to know how to invoke it, because I am a stupid user and `man` is too much for me.
<ilmu[m]>emacs is not something we can get people to agree on because it does everything. we just need something that does that one very specific thing (be a menu)
<sughosha>lechner: The problem is that the configuration error occurs like this: "gdm-service-type: Unbound variable"
<lechner>sughosha: do you have something like (use-service-modules desktop ssh xorg)
<lechner>sughosha: or perhaps (use-package-modules certs gnome) ?
<sughosha>lechner: Ohh I didn't have "xorg". Now no error. I will relogin and see if it works. Thank you!!!
<gabber>ilmu[m]: i'm not sure i agree. imho most computing users don't care about a thing concerning the internals of their machines and wouldn't care to know what even "package manager" means. the simplest of users might wish to know their system is "safe" or at least "up to date". since Guix is simply not "one kind of a system" but more like "whatever you imagine your system to be with advanced techniques to realize exactly that" your idea
<gabber>might just not make sense: what level of information, what language would you use in a manual lookup for "software update"? would you explain guix's packaging, guix pull, guix profiles, the whole roll-back mechanisms? at which level would you integrate that -- since `man` is too much to ask?
<ilmu[m]>uh
<ilmu[m]>I would just press enter on the option I want in the menu?
<ilmu[m]>us people who care about internals can nerd out about how to do all this stuff (that's what that matrix link was for, place to nerd out..) but users should just have a menu.
<apteryx>nckx: it's actually 64k
<apteryx>(cp /proc/config.gz /tmp && du -h /tmp/config.gz)
<gabber>ilmu[m]: so you propose having better documentation available for surface level Guix things in say the Gnome desktop?
<ilmu[m]>no not really
<ilmu[m]>my proposal is convoluted and overly ambitious but correct afaict
<ilmu[m]>it's not summarized in any one line but I've made an effort to keep it at ground level in these messages here
<ilmu[m]>but there are more people that care about reproducibility and authenticated datastructures than build system people
<ilmu[m]>and if we want all of these interests to work together then we should start by aligning our assumptions a bit.. which is what I want with datalisp.is
<zpiro>mbakke: https://github.com/bldur/aguix cheers!
<zpiro>I were told no, and I am not wishing for a 2017 time machine over it.
<zpiro>Not to be diffcult, but I will actually just walk away from this with my high school dregree and rejection on proposal for a ph.d.
<zpiro>Didnt intend to kill the channel, but my thoughts on the matter, and how quickly and to what leve, towards a high level of absurdity. I am not going back to CERN without a professorhiip, regardless of how important and 10s of millions of euros the IIT eqipment cost at the time, there is this problem of lacking a high school degree. And for most, prfoessor and profession mean somehintg, and not bye virtue of
<zpiro>Snowden not having a degree.
<zpiro>I'm sure there are speccial snowflakes in politics it may be wise to challenge intelectually.
<ilmu[m]>sorry zpiro but you are not coherent from my pov
<ilmu[m]>what are you even talking about?
<nckx>Night Guix. Enjoy #guix late-night. It looks fun.
<zpiro>nckx: https://www.smartrecruiters.com/CERN/743999852309083-data-acquisition-engineer-ep-lbc-2022-83-ld- I'm highly qualifed for this job. 100k USD after tax with diplomatic attache positon and role. Cant you apply?
<zpiro> https://github.com/bldur/aguix/blob/main/readme.txt this was my ph.d proposda, nixos was one, you know?
<oriansj>zpiro: what does being qualified have to do with getting a job?
<zpiro>oriansj: it is a crious place of work many in open source should consider. You're average co-worker is a professor or ph.d student. And you are hired to mkae things work surrounded with people that can do math.
<zpiro>So, you need to understand algorithms and infrastructure, but by an large you can take the the jobde to code, but is is not what is offered.
<ilmu[m]>zpiro, I sympathize with your problem (credentials out of sync with competence) if you want a fair assessment then you'd need something that doesn't exist right now (which is why frustrations ensue)
<zpiro>nobody will apply, because the basic requirement is to ensire one of the four largest physics experiments in the world works and the basic level.
<oriansj>zpiro: who cares if you lack credentials but can demonstrate that you can do the job?
<zpiro>*at the basic, but it is the gist of it. The worm description dicuss a reality of 40Tbits as the minimum functional requirement.
<zpiro>*work
<zpiro>most reasonbble people in networking would generally go, "40Tbits, naham bot my terf
<zpiro>"naha, not my terf", not my yard -- typos.....
<zpiro>oriansj: you get that bit, 40Tbits is alot as a work requirement? Most that know the field, would rightly concluide that it is awfully bad and porley paid work.
<ilmu[m]>now I am sus, are you this incoherent or is this elaborate tactics of some kind
<ilmu[m]>anyway gn
<oriansj>zpiro: actually seems pretty low but then again physics experiements are not my domain of work.
<zpiro>oriansj: you are right, for me it is about 1 Peta yte per second, and that job at 40Tbit is at the low end. But you may end up sitting there, and you probably need the authority to bitch at kernel driver developers to realistically do it.
<zpiro>but you have a better chance ffor such a job ast an open source community memeber. It stands positvely out more tyhan the the guy whom rose above it as a normal nerd.
<zpiro>somehow it seems to involve finding that guy with a physics degree with talents. As something narrow, and it is a favor to everyoone not being offered tyhrough the standard mold of candiates.
<zpiro>go figure, it pays much. and on average, about 100k physiticsts pass through CERN every year.
<jab>sneek: later tell nckx that he is a comic genius!
<sneek>Will do.
<zpiro>Physics experiments wasn't my domain of work either. But had my stint of Iy League Professors, and world leaders in Standard Model physics bareques. Where I met all of the who's whom in physics and above 200 h-index. If they have a high h-index in physics, I have met them privately.
<zpiro>not as it souds, but there is such a thing for standard model physics and CERN. Where in physics, you could be stephgen hawking at camdribe, but if you do not fall in line. CERN stems on you and your research group.
<podiki[m]>let's keep this more on topic
<zpiro>absurd and ture, but so is also. Even torvalds bow down on C code abd higher OS principles - nothing about it.
<lechner>Hi, how may I depend on libtool 2.4.7 in a package declaration, please? It's available on CI, presumably via core-updates.
<zpiro>*true, pitful or what not. But as a matter of fact. After personal authroity and stardom, boils down to being able to appeaal. Biggest world institutions hafe no ceiling or floor to cover this need, nationally or internationally. And nationally they are always famous.
<zpiro>lechner: more often than not, it involves using corcodile sing "<" or ">". *I want anything that match this number*, or sophistically less than. Universal advice to check where we are in the world of text files and unix.
<zpiro>unix ideaals, "you work with text files or crash", "you have to work with regular expression or crash". and so and so on.
<zpiro>so absurdly specific, about 2.4.7, you can specificy it, or the requirement was done by somone whom just needed it to run once.
<lechner>zpiro: you are right! i need to run it once because the maintainer told me he fixed my bug
<zpiro>or said aanother way >2.4 ans <2.4 id some such isn't accepted as input instead of 2.4.7. It was written by a teen ager and you can spend two days of your life to do it right.
<lechner>zpiro: compared to you, i know nothing about physics but a little bit about people. do you need some sleep?
<zpiro>aww, s/id some/or some/g.
<lechner>no worries, i also type poorly
<lechner>sometimes
<zpiro>lechner: maybe! I may be admitting to knwing more about the laws of nature than the laws involving the impressions of others.
<zpiro>lechner: but only a british personn can articulate as such in the way that you do.
<lechner>when i get tired the two seem the same
<lechner>i envy the british humor and have been trying to imitate them ever since
<lechner>but you look right through people
<zpiro>I am 40 now, and can go: Well, I am told I worked at Frankfurt Institute for Advances Studies and the Turing Institute. But I have never had a job or wokred in the cities invled, but I can, and they can, claim alumnui.
<oriansj>zpiro: then let me give you an foundational property of people. They don't care who you are or what you can do. They care about what they believe you can do for them. That is the basis of all power that exists in credentials; it enables others to believe you can solve their problem and that is why they may choose to hire and pay you money.
<zpiro>think i have been in frakfurt airport, but i have never been to the city. I am howver listed as FIAS alumni with no formal high school degree.
<zpiro>Oh, and IEEE refuse to unlist my associatoion to FIAS and CERN in spite of direct threats.
<zpiro>and this is the only way i can nornmalize this, but it is true.
<zpiro>as I have a fucking h-index, or used to have one, without formally finishihg high school.
<zpiro>oriansj: you want to kick me for disruption?
<zpiro>I am sure that founder of #nixos and many things related related will find me in their logs, and now you, oriansj.
<zpiro>"03:57 <oriansj> You will want to stop before you get kicked and banned for disruption." Try, and fuck you!
<zpiro>this is the one thing I am fairly certain nobody in here can do and get away with as a community.
<oriansj>zpiro: 1) I am not a Guix mod. 2) Reading comprehension doesn't seem to be a skill you have right now. 3) this channel is for Guix and Guix related work/problems.
<apteryx>I'm trying to test issue #30093 on Fedora 33, and the guix-daemon won't start; SELinux, I guess?
<oriansj>apteryx: well if you do: sudo tail -100 /var/log/audit/audit.log | audit2why you should be able to see if it is
<oriansj>journalctl -t setroubleshoot is also another route to that info
<apteryx>thanks
<oriansj>and if you see nothing: There might be dontaudit rules in play and running semodule -DB to disable all dontaudit rules
<apteryx>seems it was that
<vagrantc>zpiro: i am also not a moderator, and i don't know what was said before i joined the channel, but please show some empathy and kindness at the very least https://git.savannah.gnu.org/cgit/guix.git/tree/CODE-OF-CONDUCT
<vagrantc>maybe also "Giving and gracefully accepting constructive feedback"
<zpiro>vagrantc: https://github.com/bldur/aguix/blob/main/readme.txt while 2016, along your lines?
<vagrantc>zpiro: my lines? i don't follow.
<zpiro>vagrantc: oh, well my room mate was coding VHDL when i wrote that.
<zpiro>binary resporducible down to firmare and VHDL was somehintg I thought open scienve would be head over heals for.
<vagrantc>sure, bootstrappability and reproducibility ought to really catch on at some point
<zpiro>my assumptions being in error, i am sure cern have many good positions to offer.
<vagrantc>and guix seems to be promising for achieving some of that
<oriansj>but as of yet. It isn't culturely accepted.
<oriansj>now there are certainly people here who would value having bootstrapped hardware and an entirely free software stack from the microcode up.
<zpiro>CERN.CH still have the entire IRC protocol banned by deep packet inspection.
<oriansj>zpiro: please try to stick to things actually related to guix
<zpiro>I discussed the topic with those involved, and it was unfortunately a tradisonally botnet measure and nothing that could be done.
<oriansj>zpiro: well you don't need to use IRC to use Guix
<zpiro>The position I pasted here advertised 40 Tbits.
*vagrantc waves
<vagrantc>good luck folks, gotta head out
<zpiro>Where I can offer the ancetode, from Ben Segal, from internet hall of fame from first link to Europe from the USA. That it was sold as, "Don't worry, ethernet is encapsulated, it wint affect older networks".
<zpiro>that being said, cern network an firewalkl team is 4 peeople, with terrabit links. And good guys, but they ban IRC and I got trouble, but major experiments are actually expemppted from general banter by automation.
<podiki[m]>please stay on topic (guix), there's an off topic room if you prefer, though the name escapes me (maybe guix-offtopic)
<zpiro>OOr there are these other people, but generally there are four people that operate. You can run torrent and irc from central experiment links. If nnot you are told off that IRC is for botnets aaand torreents steal sccience bandwidth in automated manners.
<zpiro>so, ffor LHC acyperiments, you have 100Gbit or more for dfowntime.
<zpiro>IIRC 100Gbit is lowest common demoniator and avbout what direct link to Japan is.
<lechner>zpiro: time to call it a night? i think you might get kicked. let's get some good rest. i'm taking off too
<zpiro>lechner: as long as wee understaand that outside of beamtime, that these numbers are what is ensured and guaranteed?
<zpiro>lechner: I'm not saying anyone should do it foir porn, but outside of LHC beam time, science guarantees excess of 1 GB second.
<lechner>i'm definitely outside of beam time now. let's go home
<zpiro>alright, but there are bandwidth aguarntees in world academic netowork for CERN.
<zpiro>lechner: I get that is is hard to comprehend, but guaranteed bandwidth is why CERN is not on IRC, and the guarantees are very high. At major experiments the four people doing sofphisticated networking and firewlling stop askin qquestions.
<zpiro>lechner: just we are clear baot that LHCb. i if you need 100Gbit for anime, you have real world perfomrance and deliverables - guaranteed and prioritized.
<zpiro>you can be that guy using more than half of a nations internationakl bandwidth.
<zpiro>lechner: always a bright tomorrow to be had. I am faaairly certain i will not get kicked. By an large I have always been able to carry.
<zpiro>It may not be ideal to ben ranting and raving, but aslong as you can held your own. It is left to channel operators, and also here I believe that more ranting and raving to be better. Unless you are berting and abusing someone? That is, you can mute me withouty relying on operators.
<zpiro>With more tha 20 years on IRC, with how silent this channel is. The only reasonbole thin is to rant and complane with me before anything else becomes a question.
<zpiro>Lest so eone fail to sneak in help if the channel is asked and those active dont know.
***cedb is now known as yeet
***yeet is now known as Guest697
***Guest697 is now known as cedb
<zpiro>lechner: Gave it some thought, and to be clear about. It is not that I am drinking, and drunk ranting; becuase I am. Aand should be tolerated for fun effect. It is not that I: first came to IRC in the late 90's. Or that I contributed as an IRC op, or that I helped run 200+ channels prior to servers. Or that I can referr to people, and appeal if things should go bad. While having friends involved in
<zpiro>irc.blackened.com being besides the point. Fair is a mninumum demand.
<zpiro>So I have passed the stage, where, well; you can, but you may never hear the end of it, and the old decrepted, and ghosts of old may show up.
<zpiro>may not be a fine line, but it is very old and respected; for how far ranting old farts may go.
<jackhill> /topic
<podiki[m]>please keep the chatter on topic; it is quiet here now due to the time for most users, but that's no excuse
<apteryx>zpiro: #guix-offtopic exists for a reason; consider yourself warned
<PotentialUser-56>Hi
<abralek`>Hi Guix
<itd_>nckx: No worries, so was I (away). Then I'll try to patch the JSON importer. Thanks for your opinion/time.
<unmatched-paren>is there a way i can get ``halt'' to work without root?
<unmatched-paren>hmm, setuid?
<abrenon>hello guix
<mfg>:bp
<jpoiret>unmatched-paren: `loginctl poweroff`?
<sneek>Welcome back jpoiret, you have 1 message!
<sneek>jpoiret, jgart says: cool makes sense thnx
***jess is now known as jess-o-lantern
***jess-o-lantern is now known as jess
<nckx>apteryx: It does exist for a reason, but this is definitely not it…
<sneek>nckx, you have 1 message!
<sneek>nckx, jab says: that he is a comic genius!
<nckx>He is?
<Not_Leader[m]>what does sneek do?
<nckx>sneek: yeah, besides eating botsnacks?
<nckx>Aw, it didn't work. Anyway, sneek exists to deliver messages and eat botsnacks.
<Not_Leader[m]>oh
<nckx>They are also a repository of questionable knowledge:
<nckx>sneek: what is Guix?
<sneek>I've heard Guix is a functional package manager for the GNU system and beyond.
<nckx>sneek: what is npm?
<gabber>nckx: jab meant you
<nckx>Apparently, sneek is polite.
<nckx>Not_Leader[m]: 99% of pracital usage is leaving messages. ‘sneek: later tell x y’.
<Not_Leader[m]>oh wait on irc <name>: is a mention isn't it
<nckx>I guess? I've only heard it called a ping.
<nckx>There is no formal @foo syntax like on Matrix.
<Not_Leader[m]>oh right
<nckx>String contains your nick? A client ding and highlight for you.
<Not_Leader[m]>oh so still like matrix
<Not_Leader[m]>Not_Leader test
<Not_Leader[m]>oh wait
<Not_Leader[m]>can you type in my nick
<nckx>Matrix does some magic, because I've ‘mentioned’ (in the literal sense) names before that didn't turn into Matrix ‘mentions’ for whatever reason.
<nckx>Not_Leader[m]: I'm sure the IRC <-> Matrix bridge does some conversion.
<nckx>I (very) occasionaly use Matrix, and I type foo: out of habit and it doesn't work, and here it always seems to, Q.E.D.
<Not_Leader[m]>nckx: oh i think names also mention in irc
<Not_Leader[m]>i mean also in matrix
<Not_Leader[m]>because that happens on a discord bridge too
<Not_Leader[m]>works on my side all of the time weirdly enought
<Not_Leader[m]>anyway we're going off-topic
<nckx`>Not_Leader: This is a mention?
<Not_Leader[m]>nope
<nckx>It didn't format that way in my client.
<Not_Leader[m]>so bridges mention but regular accounds don't
<Not_Leader[m]>anyway i should probably ask this before i forget about it
<nckx>:)
<Not_Leader[m]>does guix-home have a flatpak service?
<Not_Leader[m]>also how did you immediately recognise that i'm on matrix
<Not_Leader[m]>is there something in the nick
<Not_Leader[m]>oh
<nckx>There's a big honking [m] across the IRC bridge.
<Not_Leader[m]>Not_Leader[m]
<nckx>Yup.
***Not_Leader[m] is now known as Not_Leader
<nckx>Not_Leader[m]: I'm no expert but I'd not heard of one and grep says no.
<Not_Leader>oh ok
<nckx>And I tab-completed that before you removed the [m], just in case you thought it might not've worked. It did.
<Not_Leader>huh
***Dynom_ is now known as Guest6495
<f3n1x>Heya. From the terminal, how do you invoke 'suspend' ?
<nckx>loginctl suspend?
<f3n1x>ah... i'll give it a try !
<nckx>There are other ways, but this is supposed to be the user-friendly one. When it works.
<nckx>(I forgot there is an actual ‘suspend’ command, but nobody I know has ever run that deliberately :)
<nckx>(It does not do what most people expect, at all, so probably don't run it.)
***wielaard is now known as mjw
<abralek>I am finally deployed guix on my Turris Omnia (armv7l) router (yay!) and Now slowly install services. For the network, I need to install dhclient AND dhcpd services and a some static network. I am getting 'guix system: error: service 'networking' provided more than once'. What would be the right way to allow dhclient and dhcpd? Can we just replace `provision' symbol for dhclient with something else?
<abralek>I have to say that I upgraded dhclient-service and transform it to use configuration record, so to be able to specify interfaces it suppose to listen on
<jpoiret>does QT not pick up QT_PLUGIN_PATH anymore?
<jpoiret>(it's for an application using qt5)
<PotentialUser-30>Hi all, I'm trying to build a guix package for the first time but seem to be getting stumped on a simple error:
<PotentialUser-30>    (sha256 (base32 "885e5b5f5f25eec6f9e4a1e8be3d0ac71a686331ee1cfb442dba391111bd32bd"))
<PotentialUser-30>    > guix package: error: exception thrown: #<&invalid-base32-character character: #\e string: "885e5b5f5f25eec6f9e4a1e8be3d0ac71a686331ee1cfb442dba391111bd32bd">
<PotentialUser-30>I'm guessing because it looks like a base16 string, but everytime I try to decode it re-encode it as base32 I run into errors (e.g. `echo hash | basenc -d --base16 | base32`). What am I doing wrong?
<civodul>PotentialUser-30: yup, it's a base16 string instead of base32, hence the error
<civodul>see "guix hash" and "guix download" on how to obtain it: https://guix.gnu.org/manual/devel/en/html_node/Invoking-guix-hash.html
<ryan1>PotentialUser-30: yeah, looks like base32 should only have uppercase characters
<PotentialUser-30>Ahhh perfect -- this solves my problem, thanks for the speedy response
<ryan1>Hello everybody! Does anybody know if your first email to guix-patches@gnu.org requires moderation before it's submitted? I submitted a patch last night, but I'm not seeing it in the archives or on issues. I just want to make sure I set up git send-email correctly.
<civodul>ryan1: hi! yes, there's a delay (could be a couple of hours) for your first message to the list
<civodul>you'll have to be patient :-)
<minima>hi, i use a slight variation of this function to convert some bash scripts into executables that i can launch from my guix home https://paste.debian.net/1255408/
<the_tubular>Wait, we supporting bcachefs now ?
<minima>one of these scripts contains a non-ascii character (a mid-dot '·'), which seems to be lost in translation somehow and is rendered as '??' in the final copy of the script
<minima>is there anything obvious in the function above that's not unicode friendly? or that might otherwise get a bit touchy when served non-ascii input?
<civodul>minima: it's kinda annoying: before 'substitute*', you'll need to install a UTF-8 locale
<civodul>to do that you first need: (setenv "GUIX_LOCPATH" #+(file-append glibc-utf8-locales) "/lib/locale")
<civodul>and then: (setlocale LC_ALL "en_US.utf8")
<civodul>or similar
<civodul>once you've done that, file I/O will default to UTF-8
<civodul>actually it might be enough to do: (%default-port-encoding "UTF-8")
<minima>ah! i had a vague idea that i needed to add unicode support at that stage, but i'd never figured it out by myself - super thanks!!
<minima>oh let me try with that first then
<jpoiret>the_tubular: some people here use bcachefs but it's still out-of-tree
<jpoiret>(in the kernel)
<the_tubular>Yeah, I know that, but is it supported by guix ?
<the_tubular>There's a few commits that makes me think so
<jpoiret>I think nckx uses it
<jpoiret>he could tell you more about it
<ryan1>civodul: No problem, thanks for letting me know. It's been a fun packaging process, BTW.
<minima>civodul: amazing, i managed to get it work with %default-port-encoding, as you said, thanks!
<minima>with '(with-fluids ((%default-port-encoding "UTF-8")) ...)' to be precise
<the_tubular>Ok, I'll ask him jpoiret
<civodul>minima: cool!
<PotentialUser-30>Hi again, different problem this time: I'm getting a:
<PotentialUser-30>>  starting phase `bootstrap'
<PotentialUser-30>running './autogen.sh'
<PotentialUser-30>patch-shebang: > ./autogen.sh: line 10: autoreconf: command not found
<PotentialUser-30>which makes sense since my `PATH' does not contain that binary.
<PotentialUser-30>> whereis autoreconf
<PotentialUser-30>autoreconf:
<PotentialUser-30>> /gnu/store/zl9bv553iwrkyiqw5wvwcq0yh745syfj-profile/bin/autoreconf
<PotentialUser-30>which looks like it's included by the package "profile", although the package 'profile' does not exist. What am I missing here?
<jgart[m]>What is .PHONY mean in a Makefile again? 🤭
<PotentialUser-30>the closest thing I've found in the archives is this:
<PotentialUser-30> https://lists.gnu.org/archive/html/guix-devel/2014-01/msg00032.html
<PotentialUser-30>which talks about a setting up the package environment first
<jgart[m]>s/is/does/
<PotentialUser-30>> What is .PHONY mean in a Makefile again? 🤭
<PotentialUser-30>the rule isn't a real file dependency, but a fake one you can use for other rules
<jgart[m]>A contextualized example speaks a thousand words.
<jgart[m]>> A phony target is one that is not really the name of a file; rather it is just a name for a recipe to be executed when you make an explicit request.
<jgart[m]>But the above quote still confuses me about what exactly it is
<jgart[m]>The main purpose is in case you have an actual file called whatever comes after the arguments to .PHONY?
<jgart[m]>PotentialUser-30: Why do I need a fake one?
<jpoiret>PotentialUser-30: you need to add autoconf to native-inputs
<PotentialUser-30>jgart[m] it's just good for naming. `make build-just-these-files` has a nicer ring to it than `make file1.txt file2.txt file3.txt`
<jpoiret>if you're building a Guix package, you have to remember that the build process is isolated from the rest of your system, your PATH doesn't influence the build at all
<abrenon>jgart[m]: it's just a rule that will be applied any time you name it
<abrenon>no matter if there's a corresponding file or not
<PotentialUser-30>jpoiret - Ahh, okay -- thanks I'll have a look around for native-inputs.
<abrenon>make's goal in life is to make files appear in your directory, if there's here and more recent than their dependencies, then as long as make's concerned, there's nothing to do
<abrenon>if you have a rule named "clean", and, if it should happen that a file named "clean" appears in your directory, then you can't run "make clean" anymore
<abrenon>or rather you can, and make will stare at you and say: "ok, done already, I'm good, ain't I ?" without applying the clean rule
<jpoiret>apteryx: wrapping QT_PLUGIN_PATH exactly causes applications to no longer pick up the user-installed wayland backend
<jpoiret>this is an issue
***flatwhatson_ is now known as flatwhatson
<minima>i liked the answer here: '$ herd restart root' -> 'You must be kidding.'
<minima>my home config includes a mcron service (home-mcron-service-type); after reconfiguring my home the mcron script doesn't seem to be updated
<minima>'herd schedule mcron' seems to be still reporting the old mcron job
<civodul>ah yes, that's a bug in Guix Home i think
<minima>i did try with a 'herd restart mcron'
<minima>ah
<minima>cool, no worries then
<civodul>"restart" doesn't help?
<minima>nope it doesn't seem to help
<civodul>'guix home reconfigure' deals with shepherd service upgrades in a custom way, different from 'guix system reconfigure'
<civodul>ok
<civodul>well, we should fix that
<civodul>maybe you can open a bug to keep track of it?
<dthompson>I really need to check out guix home.
<pkill9>one issue with guix home is that mcron is only run once you login
<minima>yes, happy to do that now - should i first double check if there's anything similar open already?
<civodul>i don't remember a bug report in that area
<pkill9>or maybe that's not an issue, but it may make more sense to have a user's shepherd run on boot by the system's shepherd run on boot as a
<minima>ok, cool, i'll try and put something together
<minima>thanks
<civodul>dthompson: daviwil demystified it in this talk: https://10years.guix.gnu.org/video/an-introduction-to-guix-home/
<dthompson>civodul: I watched it yesterday!
<dthompson>just gotta find some time to check it out.
<dthompson>years ago I used to use an unprivileged shepherd to as a proto guix home.
<dthompson>s/to as/as/
<unmatched-paren>jpoiret: "`loginctl poweroff`" <- is that an elogind thing? cause i don't have it
<pkill9>hmm actually you could just add a system service that runs the user's shepherd and still manage it with guix home
<pkill9>not sure if guix home creates shepherd dotfiles actually
<pkill9>ah well
<minima>i tried a reboot and that did update the cron job
<minima>i'll put my findings in a bug report, in case it's of any help
<unmatched-paren>I think I'll just have to tell people to add (file-append shepherd "/sbin/halt") to their setuid-programs
<unmatched-paren>Woot, batsignal works as a home service
<abhiseck>is there any difference between function(char const *arg) and function(const char *arg)?
<unmatched-paren>abhiseck: I don't think so, no.
<abhiseck>I was reading coreutils and gnulib sources and found many uses of the first there
<unmatched-paren>But ``char *const arg'' is different iiuc
<abhiseck>unmatched-paren: yeah I know those cases
<abhiseck>thanks!
***cedb is now known as mnrmnaugh
***mnrmnaugh is now known as cedb
<pranavats>Hello, can inferiors be used to access services (along with packages) which have since been removed from the guix codebase?
<pranavats>I'd like to setup a wicd daemon. I'm able to access the wicd package using inferior, but not its service type.
<unmatched-paren>No, I doubt it
<f3n1x> I'm puzzled... why tramp-mode looking for a file à-la '/sudo:hostname:/etc/hosts' says buffer is read-only ? what am i
<f3n1x> missing ? How do you access 'by root owned' files on your Emacs on Guix ?
<tricon>f3n1x: you can still modify files in /etc, such as /etc/config.scm for your sys config.
<tricon>f3n1x: you just need to disable RO mode in Emacs: C-x C-q
<tricon>(once you open the file.)
<unmatched-paren>sneek: later tell Kabouik: https://issues.guix.gnu.org/58170 :D
<sneek>Okay.
<unmatched-paren>sneek: botsnack
<sneek>:)
***ChanServ sets mode: +o litharge
***litharge sets mode: +b *!*@2001:470:69fc:105::2:8e37
***binance_group[m] was kicked by litharge (You are banned from this channel)
***litharge sets mode: -o litharge
<f3n1x>thx! tricon
<f3n1x>Just to learn from, is it because Guix idiosincrasy ? ... i guess is just a matter of getting used to ? tricon
<tricon>f3n1x: it's because the files in /etc are marked with only -r- permissions, which Guix appears to do by default. if one were building a package that utilized files in /etc, you could chmod them during the build process if you wished to.
<Not_Leader>f3n1x: i'm assuming /etc/hosts is a symlink to something somewhere in /gnu/store
<unmatched-paren>f3n1x: don't try to edit /etc/hosts
<unmatched-paren>instead use the service for it
<unmatched-paren>Oh, wait, actually
<unmatched-paren>use the (hosts-file ...) field of operating-system
<pkill9>does guix home generate dotfiles in the home directory for shepherd? or does it point to the store path
<unmatched-paren>pkill9: it seems to generate a ~/.config/shepherd/init.scm
<pkill9>nice,that means you could have a system service that starts the user's shepherd at boot, while managing it with guix home
<Not_Leader>would it be theoretically possible for a flatpak service for guix-home to exist?
<pkill9>yes i think so
***mark_ is now known as mjw
<pranavats>Hello. I'm trying to run wicd service by copying the code for it from networking.scm from an older commit in guix. But I get an error stating "#<record<origin>>: record ABI mismatch; recompilation needed". I have all the code relevant to wicd-service in a file, which I load from my config via (load ...). Can someone help me understand the error and how to resolve it?
<euandreh>I was happy to see the recordings of the "10 years of Guix" event. Thanks for all those involved on making it available
<acrow>+1
<the_tubular>Yes, I'm still catching up on the few recordings I have missed, it was a great event, good job everyone!
<unmatched-paren>Next up: home-mako-service-type! :D
<antipode>I found a test case for the 'low pitch in IceCat' bug (37224): https://hls-js.netlify.app/
<ryan2>g-expressions are neato!
<Andronikos>Does someone know the current state of Guix for Raspberry Pi and ZFS?
<lechner>Hi, what does this mean when working on the core-updates branch, please? guix build: error: failed to connect to `/usr/local/var/guix/daemon-socket/socket': No such file or directory
<viivien>lechner, you forgot to pass --localstatedir=/var at configure time maybe?
<mbakke>lechner: it means you forgot to pass '--localstatedir=/var' to your ./configure invocation
<viivien>Must be true then
<lechner>viivien mbakke: that was it. thanks!
<civodul>jpoiret: i have some bad news: https://git.savannah.gnu.org/cgit/guix.git/commit/?id=168a7933c0e138dc7061a3f0dc96871e16da5c5f
<civodul>ryan2: great that you like them :-)
<podiki[m]>civodul: wow! though I'll keep an eye for it after next reconfigure....and pour one out for our departed error: success message
<attila_lendvai>civodul, yay! that gave a rather bad impression when i saw it at the first boot...