<quiliro>what do you use to backup a gitlab repository besides the pushes? <quiliro>one copy is on the users' machines, another is on the server (gitlab) and another must be on a safe...what software should be used for that? <quiliro>ison[m]: the problem with beammer was not the upgrade...it was the contents of the org file <quiliro>when i removed the comment, it recovered exporting to beammer PDF! <quiliro>thank you for your help with Org to create Beammer slides <tidux>quiliro: the best way is to not treat it as "a gitlab repo" but "a git repo that happens to have a webui" <tidux>ssh key auth + "git pull" in cron on another box <tidux>if you mean the non-git features of gitlab I have no idea <apteryx>hello, should we gitignore the t-home-*/ and t-profile-*/ directories produced by make check (or maybe it make system-check) <apteryx>Sum2minus1:what kind of font issues? <Sum2minus1>yes some of it is missing unicode, however there who letters/blocks of text missing. strangly though i can copy paste the missing text into the terminal and read it just fine <Sum2minus1>nvm I just fixed it. turns out I installed the fonts on the wrong profile by mistake. seems to be working fine now *faceplam* <GNUtoo>in nginx-server-configuration there is index <GNUtoo>I want to enable directory listing <GNUtoo>and whatever valid thing I do, index index.html; is generated <GNUtoo>(index '("foo.txt")) -> index index.html; <GNUtoo>(I've not defined locations in the scm file but a .well-known one is genrated, probably from letsencrypt) <nly>jgibbons[m] there's no new release yet for emacsy and nomad <nly>you should send a bug report <nly>jgibbons[m] i think this might be related to window manager <str1ngs>jgibbons[m]: thank you for using nomad. can you define what you mean by logged out? <leungbk>Does guix provide the ldconfig command? <sneek>leungbk, you have 2 messages. <sneek>leungbk, efraim says: looking forward to it! ludo's crate import fixes worked for me <sneek>leungbk, rekado_ says: The CRAN importer retries imports from different repositories. If a package isn’t found on Bioconductor it will retry to import from CRAN. Same for imports from Git which are retried on Bioconductor (and then CRAN). <sirmacik>hmm, I've installed network-manager-{vpnc,openvpn} but even after restart networkmanager won't see them :/ <sirmacik>or maybe I'm just completely wrong, but it rings the bell under output formatting section <rekado>gnu_srs1: it’s a placeholder in a format string. <gnu_srs1>sirmacik: Tks. Found scheme code : !~a; ~a $out; etc ***sneek_ is now known as sneek
<sirmacik>another trouble, seems that currently there is no way to manage extensions enabled in gnome <sirmacik>I cant see package with required chrome, local component for gnome extensions website <sirmacik>an gnome-tweaks won't show either extensions installed with guix or those downloaded manually <roptat>sirmacik, I don't use gnome, but maybe it's just an environment variable issue? <roptat>what if you set XDG_DATA_DIRS=$XDG_DATA_DIRS:~/.guix-profile/share or something? <sirmacik>still doesn't work, seems like na issue with missing libsoup-2.4.so.1, there is also an error with detecting shell <bluekeys>Hi guix. I've just installed emacs-geiser. How do I load it? <xavierm02[m]>I have a problem with pdflatex. I think it's running mktexfmt every time I compile my latex file, which takes a long time, and doesn't seem to be run by the pdflatex of my colleagues. Is this normal? How can I fix it? https://paste.debian.net/1098732/ <bluekeys>I'm so dumb... stuff that gets added into site-lisp is loaded automatically. It's in the docs and I found it the second I posted here. <apteryx>bluekeys: yes, I don't have anything in my .emacs to load geiser <apteryx>but I do set the default implementation to: guile with: (with-eval-after-load 'geiser-mode (setq geiser-active-implementations '(guile))) <bluekeys>The only reason to add load statements would be if you had explicitly loaded emacs with --no-site-file for some reason or you want to load something not installed by guix. <bluekeys>Thx apteryx. That'll go straight into my config. <roptat>civodul, I forgot I answered to that thread ^^' <roptat>it looks good to me, I can take care of pushing these patches this evening <apteryx>is it possible to have geiser evaluate at the REPL only when my cursor is at the end of the line? I'd like if when my cursor if before a closing parenthesis, RET would give me a line break to nicely format statements rather than jump to eval. Paredit is kind of unusable with the Geiser REPL as it is <civodul>there was an empty commit log but the rest looks good to me <roptat>yes, I can take care of the commit log :) <civodul>apteryx: yeah, good question, not sure <roptat>I won't push all these packages either, since they are not finished <civodul>apteryx: some people do use Geiser with Paredit, so i always thought i must be missing something <civodul>roptat: ah yes, i was just referring to the last two patches in that thread <civodul>i guess you could ask nixo to submit the packages separately <apteryx>civodul:maybe they stick to a single line... this makes it convenient to paste on #guile ;-) <divansan`>I tried installing webkitgtk and sbcl-cl-webkit but it fails with the same error. <roptat>I mean, I can't see your paste, I see this error message instead <civodul>divansan`: does the file in question exist? <civodul>unfortunately the Common Lispers have yet to address it <divansan`>civodul: no it doesn't. I just found that bug. Should have checked first. *civodul shamelessly distributes work :-) <civodul>nckx: i cannot reach dmitri/sergei on ports 5551/5552; any hints? <apteryx>like, what how does Emacs handles indentation of packages, given that we use (put 'package 'scheme-indent-function 0) ? It seems the first level (package\n(first-level field is still indented normally (2 spaces) but that 2nd level + nested fields/expressions are indented at one space. <apteryx>OK, when scheme-indent-function is set to an integer, the indentation code of Emacs in (lisp-mode.el) executes lisp-indent-specform <apteryx>i'll try to replicate this in import.scm to fix the imorters indentation issue. <GNUtoo>is there some example code to just run a command at boot? <GNUtoo>like for instance to do that in defining a oneshot shepherd service <GNUtoo>what I'm looking for is where to put all that <GNUtoo>like in the system definition I guess in the list of services <GNUtoo>and I'm also looking for an example for (start '()) <rekado>apteryx: I use C-j to break lines in the Geiser REPL. <rekado>dmitri and sergei are not yet accessible from berlin due to firewall settings <rekado>I asked IT to track the IPs of these hosts, but they haven’t even responded and ignored my pings :-/ <Minall>Is there a reason why xorg is using fallback drivers even though I 'assume' I should be using intel drivers?, which are installed by default? <Minall>Maybe my system does uses fallback drivers? <Tazy>how come most of guix is build with gcc 5.5.0 ? (and clang 3.9.1?!) <roptat>I think it's from before we introduced oneshot services <roptat>I should look into that to make it slightly better <civodul>Tazy: the core-updates branches we're about to merge switches to GCC 7 by default, among other things <civodul>(Clang is not in build environments by default) <civodul>rekado: i figured there was a firewall issue but i can't reach them from home either <civodul>GNUtoo: you can also look at the existing services in (gnu services ...), some are simple <civodul>and see "guix system shepherd-graph", too <Tazy>clang was just noticeable for icecat. Welp lets hope gcc 7 works out, hopefully enables more aarch64 packges build.. <pkill9>it would be good to be able to see the daemon output <Minall>Is there a guix package for Odoo? <sneek>Welcome back nexgen2, you have 2 messages. <nexgen2>Can you please help me with a few questions about GUIX compared to Debian <nckx>civodul: No shame, it's your job. ;-) Re: our Russian friends: I can ping them through the LAN (when you ping them, you're just pinging my shitty residential router), but I can't SSH into them (‘Connection refused’) :-/ I'll have to hook up the USB TTY this evening to find out what's wrong. Surely ssh-daemon can't have crashed on both… <nexgen2>Are all 100% of GUIX packages reproducible? <nckx>I don't know if there is one, it's just an easy question to answer categorically 🙂 <nckx>I don't think this is currently tracked in an organised/public manner. <nexgen2>how can I count at least amount of non reproducible packages <nexgen2>something like this would be preferred <nckx>Something like that is what I was referring to above, but somebody still needs to write it. You can test individual packages using ‘--check’, it's not far from that to automation & a pretty HTML report. <nckx>Main blocker I see (apart from writing the code) is that you need double the CI capacity to provide up-to-date pretty stats. <nckx>We're not swimming in free CPU cycles as it is. <nexgen2>Can you please just give an idea about how much percent of packages are not reproducible <nckx>10%. There you go. Source = mah butt. Could be 90 too. I really don't know. <nckx>‘Reproducible’ means different things. It's often narrowed down to ‘bitwise reproducible packages’, which can be very important, or completely irrelevant. It's often overemphasised. <nexgen2>what you meant under 10% non reproducible? <nckx>I care that my two servers with identical Guix system configurations do exactly the same thing. I don't give a whiff about timestamps in libfoo. It's just an easy shortcut, sometimes, to check whether two systems are identical. <nexgen2>which type of reproducibility referred? <nckx>nexgen2: I consider <10% a very safe guess considering my experience, but why ask for a hard number from someone without a clue? <nexgen2>is it about 10% of non reproducible as ‘bitwise reproducible packages’ <nckx>Using that definition, yes. *nckx launches a crappy bash script on their build farm to find out. <mbakke>nexgen2: if you run the `guix challenge` command above you'll get to know exactly how many derivations are non-reproducible <nexgen2>does it mean only 2% are reproducible? <nckx>Going for 100% source(-equivalent) bootstrapping, as Guix does, is IMO far more important than the overhyped ‘let's get together & compare our bits’ approach. Those bits after the fact don't mean much. Which isn't to disparage the Debian reproducibility folks' work, bit-reproducibility is good, but it's not currently resting on any foundation. <nckx>mbakke: Won't that only work if you haven't downloaded substitutes from that server? <nckx>Ah, yes, hence the ‘inconclusive’. <nckx>Those are the downloaded ones. <nexgen2>inconclusive means it is not good enough to be treated as reproducible? <nexgen2>how it is related to be downloaded from some server? <nckx>nexgen2: No, it means you've downloaded (for example) the icecat from the substitute server you're challenging, so of course your icecat is identical to the server's. <nckx>There's no way to find out whether packages are bit-identical without rebuilding *every single* package yourself, or finding 2 unrelated substitute servers that you can compare. <nexgen2>4+1 packages have been built on my host? <nckx>So I've started a ‘build everything again and check it’ script on my substitute server, but that's a) not scientific, I'm just curious b) will be a single snapshot in time and c) will take a hell of a long time 😛 <nckx>nexgen2: I think that's lower than average but not implausible. *nckx wonders which item differed in nexgen2's analysis. <nckx>I don't know. I've never used ‘guix challenge’ (I don't use substitutes anymore). <nexgen2> local hash: 0swd4hi7vahbx2ivjkaw8bliayns1wrk8qkn0c9lhi7cj58k0lbh <mbakke>nckx: yes, you will need to check against a server that you do not have substitutes from, which is why I used bayfront (which did not appear to have any substitutes, unfortunately) <nckx>That's a derivation produced by ‘guix pull’. <nckx>mbakke: Ah, bayfront was down the last times I tried. <nexgen2>Can you please explain what is "100% source(-equivalent) bootstrapping" <nckx>nexgen2: I'd like to repeat that Guix takes reproducibility *very* seriously (in fact, I think it's ahead of everyone else in bootstrappability efforts like <https://www.gnu.org/software/mes/>). <nckx>nexgen2: Ah, by coinicidence, what I just wrote. 🙂 <nexgen2>well, I already indicated MES advantage on the forum where I am comparing GUIX to Debian <nckx>nexgen2: A lot of current ‘reproducibility’ is just: if I feed source code into this huge binary blob [e.g. Debian], and you feed the same source into the same blob, do we get an identical new blob? That's all those reproducibility numbers mean. <nexgen2>Debian guy states it is not important MES thing <nckx>That can be valuable, but it's not valuable in a vacuum. It's good to know if someone tampered with your system. It can't tell you if someone tampered with *all* systems. <nexgen2>though I find it very good by myself <nexgen2>because it would help to avoid backdoors in the GCC <nckx>nexgen2: Uhm, well, considering we're lacking all the context (and I don't feel like joining yet another discussion) I won't comment on that, but I might roll my eyes. <nckx>‘you see a desire to mention this dirty trick as many times as possible and tell more about it, without disdaining lies. Doubtful advertising.’ <nckx>Unless this is a mistranslation, the tone of this discussion seems very aggressive and I don't really feel a desire to participate. <nckx>(Nothing to do with our conversation above, of course 🙂) <mbakke>That means 92% of the store items that were found on the remote servers were reproducible <nexgen2>Russian forums are very often aggressive <nexgen2>that is why russians generally used by ZOG to opposite them to the rest of world <nckx>Okay imma stop you there, as far as #guix goes. It's off-topic to say the least. <mbakke>civodul / rekado: can you kill evaluation 7158 in Cuirass? <nckx>nexgen2: I certainly disagree with the >quoted poster. It's pure opinion. I don't think it's a good opinion, no. <rain1>nexgen2: they can join #bootstrappable if they want to discuss it <rain1>reproducibile builds is orthogonal to bootstrappability <nckx>nexgen2: Sure, I understand (and I agree with what I assume is your response, although I'd state it less… Russian ;-) ) The whole premise of ‘usefulness’ is just so subjective. I mean, you don't want to know what I hear said about plainn software freedom by people not in our bubble. <nckx>This is the same thing. ‘Eh, who cares’. That's not something you can sanely argue against. It's a dead end. <nckx>Which is to say I can't; I backtrack and try another avenue. <GNUtoo>hi again, is it possible to eable directory listing with nginx? <nckx>GNUtoo: Does "autoindex on" not work? <nckx>It definitely works here with a ‘raw’ nginx.conf. <quiliro>what is necesary to make a git server accessible by ssh? git daemon and openssh as services? <GNUtoo>nckx: I'll try that, thanks a lot <nckx>quiliro: Nothing but ssh-daemon. <quiliro>so if i want to git clone from a remote machine, i do not need to start git daemon? <nckx>For the same reason that you don't need a daemon to use scp: ssh runs the commands, as long as the server has git installed (but nothing needs to be running). <quiliro>all i have to do is install git and openssh on both server and client and then just git clone? <GNUtoo>quiliro: normally yes, you just need to make sure that the directory you're cloning is a git bare repository <nckx>quiliro: Yes, although ‘install’ is a bit vague in Guix :-) OpenSSH needs to be running as a service, git can just be installed in your user profile. <quiliro>something like git clone ssh://home/user/git_repository <nckx>GNUtoo: I clone a non-bare one regularly and it seems to work fine? <nckx>quiliro: That's just not the syntax; it's ssh@host:/path/to/repo. <nckx>(Path can be relative, but I can't say what it's relative to, so I always use / on my own machines.) <GNUtoo>nckx: maybe for cloning, for pushing it's more problematic <GNUtoo>The issue is if you push in a branch that is checked out <GNUtoo>I don't remember exactly how it failed though <quiliro>is there simple documentation for making it pushable? <nckx>GNUtoo: I assume it's equivalent to pushing to a local directory, which I've never tried. <quiliro>you can pull and push to a local directory? i did not know that! <GNUtoo>I tried to see if I could shorten the time spent to push and compile code <nckx>quiliro: I'm pretty sure pushing should just work, on the condition that it's a bare repo. <pkill9>cool, i didn't know you could git clone over ssh like that <GNUtoo>so if my memory is correct, you cound't push on the branch that is checked out, so I ended up having to puth to a second branch and do a checkout of that <nckx>quiliro: Well, I meant that more as an illustration of how naive (in a good way) the SSH layer is, as I said I've never actually tried 🙂 <nckx>I was pleasantly surprised (as you are) that you don't need to run any extra daemons to do simple self-hosting. It's HTTPS and fine-grained access control that needs those. <quiliro>by the way...i am testing Jami on Guix System...if anyone wants to try, talk to me....my id is 54504e52f558e9ac83b1368ec6aaa54748a0e8e5 <nckx>(By the way… Guix's Jami packages are out of date, hint-hint…) <GNUtoo>I didn't try that in GuiX yet but the ssh daemons and ssh command line tools can do that <quiliro>oh! an oportunity for me to learn by packaging Jami...thanks nckx <GNUtoo>nckx: I have: (nginx-server-configuration [...] (raw-content '("autoindex on;")) [...] ) <quiliro>what if my git repo does not have a public ip..may i use onion addresses for accessing it remotely? <GNUtoo>quiliro: the daemon enable that indeed, not sure about guix config though <GNUtoo>nckx: I then do "guix system reconfigure <the scm file>" and then with ps aux I find the nginx config file and grepping for autoindex in it founds nothing <nckx>GNUtoo: I don't use the Scheme wrappers meself, by ‘raw’ I meant that I write my own ‘native’ nginx.conf and use (nginx-configuration (file "/etc/guix/nginx/nginx.conf")) in my system.scm. <GNUtoo>for tor, the old wiki had some infos on how to configure things for it <GNUtoo>I use it myself to bypass NAT as well <nckx>GNUtoo: I didn't know there was an (unrelated) raw-content field in Guix's wrappers or I would have used a different word. However, the docs for nginx-location-configuration say that it takes a ‘list of strings’, not whitelisted fields, so you should be able to add it there at least: (body (list "autoindex on")); or so. <GNUtoo>Note that if you care about privacy too you might need additional measures <pkill9>Jami looks good, i wonder how well it works <quiliro>but it is possible it is not the fault of Jami <quiliro>how can i verify if it is a problem of Guix System or not? <pkill9>quiliro: do you get any errors? what does the output say when you run it in the terminal? <quiliro>pkill9: i have no one avialable to test <quiliro>i have started gnome-ring from the terminal <GNUtoo>(raw-content (list "autoindex on;")) worked, thanks a lot <quiliro>if anyone would please want to help me test what is my jami problem, my jami id is 54504e52f558e9ac83b1368ec6aaa54748a0e8e5 <nckx>Nobody knows your hackin' in bed until someone asks you to video chat. <GNUtoo>lol, it's not always convenient to hack in bed though, it depends on what you do <GNUtoo>If it involves disassembling hardware you risk loosing the screws... <quiliro>i should think about wearing a shirt! <nckx>GNUtoo: Word… I have to downclock my CPU to a GHz because ‘it's too loud otherwise’. <nckx>Granted, the soldering iron might be a bit much. <quiliro>for updating a package, shold i use emacs guix-edit? <nckx>quiliro: I get ‘Note: file is write-protected’ because ‘guix edit’ tries to ‘edit’ files in the store by default [this is indeed unfortunate, I dream of an elegant copy-on-write solution some day], so you'll have to point emacs-guix to a writable checkout first. <nckx>quiliro: I never use guix edit, I just… edit ~/guix. With emacs, of course. <rekahsoft>Hi all, when using 'guix system docker-image ...' what would a minimal os file-system configuration look like? I notice there is a %container-file-systems variable exported from the file-systems module, however I can't seem to get it to work when using (file-systems %container-file-systems). <nckx>quiliro: If this is about jami, I see that 201904 includes a ‘Sets up video streams upon receiving the first video frame’ fix so yeah, might be worth trying. *nckx knows nothing about running Gocker so that's about it, unfortunately. <rekahsoft>nckx: all good, I was just playing with it today as I wanted to build a little sample python project that is managed with guix to show a friend <rekahsoft>There are actually two slightly different ways to produce docker images from guix (at least from what I can tell). One is 'guix system docker-image ...' and the other is 'guix pack -f docker ...'. <nckx>rekahsoft: The former creates a complete operating system, the latter an ‘application bundle’ (application + dependencies, or basically a big static binary). Maybe you knew that already, but I don't think either is redundant or the overlap huge? <nckx>I do think all your questions are perfectly reasonable and not having a clear introduction in the manual is a bug. Which would be nice to report, if you'd have the time. Hint hint 🙂 <nckx>That's not even a hint, is it. <pkill9>wish jami provided a distro-agnostic portable binary/bundle/appimage <tune>maybe they could use a guix container <rekado>mbakke: I don’t know how to cancel an evaluation. Is it enough to delete the evaluation from the cuirass database…? <pkill9>and that git repository has the sources within submodules that i think are also pointed to non-existent repositories <rekado>(it’s inconvenient that cuirass has no admin interface for people like me who don’t want to mess with SQL) <nckx>pkill9: I was just getting super-annoyed at my failing to find the ‘Download tarball’ page. So it's not just me. <rekahsoft>nckx: I had just made the discovery earlier today :) <nckx>Yeah, I got that far, and eventually found their GitLab URL (their big ‘GitLab’ button not pointing to it was a fun prank). <nckx>Screw flapimmages, I wish projects still released sane sourceballs. The real distro-agnostic format (™). <nckx>rekahsoft: Not the first time that #guix is a self-help group for those traumatised by bad packaging. Take a seat, have some booze. <mbakke>rekado: perhaps killing the relevant cuirass processes is sufficient? <mbakke>rekado: The PIDs are 48035 and 48077, and 8811 and 8696 (for 7c93edb). <rekado>wouldn’t cuirass restart them the next time around? <mbakke>hmm dunno, maybe the evaluations need to be deleted too <mbakke>in related news, 11d73fb broke cuirass <rekado>I don’t see anything relevant in the cuirass logs. They are so full… <rekado>I think the build output really shouldn’t end up in the Cuirass log. <rekado>I see this in the log for the guix-extra derivation: icecat-makeicecat.patch: patch not found <tidux>rekado: are you doing a source build of 60 or packaging 68? <nckx>rekado: Weird: someone posted that exact error here, turned out to be infinite module recursion, don't see how that can be the case here. <nckx>It must be the standard failure mode of things being wrong 🙂 <nckx>tidux: This is just while trying to ‘guix pull’, AFAICS. <nckx>(IceCat version numbers?) <rekado>nckx: you’re probably right about mutually recursive module references. <nckx>That would imply things being pushed without evaluation testing…? <nckx>It's a big day ☝ for this guy. <rekado>+ #:use-module ((guix build julia-build-system)) <nckx>[Genuinely curious] How could that ever not blow up while developing?