<roptat>~/guix/repo/master/gnu/../gnu/installer/////////////////////////... <roptat>that's the filename that's too long <civodul>well well, where does that come from? <roptat>it's the guile from /run/current-system/profile/bin/guile <roptat>oh no that's indeed the guile from guix environment guix <civodul>you could run "guix environment guile3.0-guix" <civodul>though i don't know if that'd solve your problem <roptat>there seems to be a loop from ~/guix/repo/master/gnu/../gnu/installer/ to ~/guix/repo/master/gnu/../gnu/installer//////.... which add a / at each turn (there are many lstat64 calls before we reach the filename length limit) <roptat>I can't go further back in my terminal's history though <civodul>roptat: anything fishy in $GUILE_LOAD_PATH? <roptat>civodul, make just failed with the same issue <roptat>nothing weird, the environment profile, and some 2.2 from /run/current-system <roptat>oh there's this warning: WARNING: Use of `load' in declarative module (gnu system install). Add #:declarative? #f to your define-module invocation. <roptat>but it wasn't there with guile 2.2 <civodul>any difference if you revert 82d8959e5d137b2061a68878d78a8f74a238ac44 ? <roptat>btw it's perfectly fine on x86_64 <roptat>civodul, reverting that commit still gives me the warning, but no error anymore <civodul>can you spot the error in that commit? :-) <roptat>is there a way to test that procedure? <roptat>from guix repl, there doesn't seem to be anything wrong... <civodul>it could be an incorrect offset in %struct-dirent-header/linux <civodul>but i don't see why it would happen now <roptat>oh my system is a few commit older, so I don't see the issue I think <roptat>bah I can't, guix-modular-master is failing on armhf since then... <rain>anyone here who knows gcc well enough to help me figure out the flags for statically linking libc? i can't get a statically compiled fsck.f2fs together. ;_; <atw>is there a service in the distribution that generates a yaml config file? I'm looking to crib from something and avoid running afoul of yaml's escaping <sneek>Welcome back xelxebar, you have 1 message! <apteryx>raghavgururajan: OK, linphoneqt works well with pulseaudio, it's just that when testing with guix environment --pure, the pulseaudio environment variables aren't defined, which causes it to use ALSA directly. <apteryx>At this point all the features I regularly use on Linphone for Android (text messages, calls, TLS/SRTP) work well with linphoneqt on my Guix System, which is quite nice. Thanks for this useful contribution! <sb-nipper>anyone her up to answering an installation question on installing from "guix-system-install-1.0.1.x86_64-linux.iso"? <sb-nipper>I succeeded in creating a boot-able memory stick. However the <sb-nipper>gui installer hangs while connecting to the wifi network I select. <sb-nipper>The script sees the interface, and the available networks, but gets <sb-nipper>stuck at "connection is progress Connecting to ..." <sb-nipper>HOWEVER, when setting up the network in manual and <sb-nipper>specifying the wpa_supplicant.conf file, it can connect <sb-nipper>Is there a way to specify the correct wpa_supplicant.conf file <sb-nipper>or break out of the gui script to do this manually then return? <raghavgururajan>apteryx I am glad :-) Btw, did you get it working on master? For me it does. <apteryx>raghavgururajan: I'm giving it a fresh try now. <apteryx>I've cherry picked all the linphoneqt patches from core-updates <raghavgururajan>apteryx Thanks so much. I just wanted it to get into master, so that many can start using it :-) <apteryx>raghavgururajan: 48acdefe2e adds the ring sound files <apteryx>OK, I'm hitting the #include_next <stdlib.h> error, without adding gcc5 <apteryx>still the libGL error, failing to load my driver (nouveau) <raghavgururajan>Then, this issue is not global, so need not be an blocker to add it to master. However, we can try to fix this. <apteryx>yes, I'm using an nvidia card. Are you on intel? <raghavgururajan>apteryx Also, could you try running the application in two ways by [1] `./pre-inst-env guix environment --ad-hoc linphoneqt --pure` && `linphone`. [2] `./pre-inst-env guix environment --ad-hoc linphoneqt ` && `linphone`. *raghavgururajan rebuilds linphoneqt <raghavgururajan>apteryx I think I know how to fix your issue. The app crashes on your end because for socket and mutex. I sent a patch for this, but got lost in the email thread. <apteryx>raghavgururajan: yeah, I already tried with and without --pure, it doesn't help. You can replace && by -- to use a single guix command: ./pre-inst-env guix environment --ad-hoc linphoneqt -- linphone <apteryx>raghavgururajan: why were you saying that Qt on master is using GCC 5? <raghavgururajan>It is using gcc5. You can see it in the pack def of qt4, which is inherited by qtbase <apteryx>I don't see qtbase inheriting from qt (qt4) <apteryx>From my understanding, qtbase is qt5, and qt is qt4. <apteryx>The problem on master is caused by the order of include directives. Master was (is still?) using CPATH for GCC > 6, which causes issues with -isystem directives. <apteryx>the commit which fixed that on core-updates is 2073b55e6b964cb8ca15e8c74cb32dac00f05f0d <apteryx>so, using GCC 5 is not a proper solution, I'm afraid (especially without removing the headers/libs that would come with the default GCC 7 -- it seems to me it'd potentially create some Frankenstein binary :-)) <apteryx>swig has an example of a workaround that should do it <apteryx>well, transposed from Automake to CMake <raghavgururajan>apteryx Cool! I am just working on a patch to fir your error. I'll send you shortly. <MYMYRSRS>How do you resolve environment variable conflicts when using Guix on other distributions? <MYMYRSRS>For example, the CA certificate environment variables required by curl sometimes conflict <MYMYRSRS>I always write these variables in .bashrc <apteryx>raghavgururajan: but we aren't using qt-4 unless I'm missing something <apteryx>we are using qtbase, which is Qt 5, and doesn't need any GCC 5. <apteryx>I should try just the socket patch at first? <apteryx>OK, the socket one doesn't seem to fix the issue <apteryx>yes, that's good, but it seems not the cause, more an effect <apteryx>But why does it work on core-updates? The fix is probably in the core-updates commits. <apteryx>I'm not too keen on adding layers of stuff to work around problems that only exist in master; it's a bit sad to not have linphoneqt on master right now, but it doesn't seem trivial to make it work reliably (my attempt to not use the gcc-5 hack failed also :-/). <raghavgururajan>So let's linphoneqt be in core-updates for a while, until we find a solution. <apteryx>I'd like a 2nd opinion on this one, as usually the policy is that icons are left to install by the users, the rational being that we don't want to force any particural icon style on anyone. <apteryx>In the case the package appear severely broken (I haven't tried the old linphone gtk yet, so I don't know) without a special set of icons, perhaps we can make an exception, but otherwise this has been consistently followed, AFAIK. <raghavgururajan>It is logical to leave the choice to users, when the application itself provides that choice. <apteryx>raghavgururajan: I see! This kind of information as a comment above the added inputs would make the review easier :-) <apteryx>Something like: XXX: Specific icons are hard-coded, so include them given the user doesn't have a choice anyway. <apteryx>I'll add this and push if you don't mind, after testing a bit the GTK client (not now, it's a bit too late :-) <apteryx>the last experiment for today is guix copying' my linphoneqt built on master to an intel laptop <apteryx>so we have an intel only linphoneqt on master ;-). I'm not convinced that cuts it yet... I'd like to find what was done on core-updates that resolved that Qt/libGL problem. <apteryx>thanks for entertaining the conversation! <apteryx>ah, that's not true. All the parts were functional, I just added some polish here and there. <xelxebar>What's the purpose of the label string when listing inputs to a package? <xelxebar>If we're just listing packages, it seems redundant, so I assume it's to give names to more complicated cases (origins etc?) <lle-bout>bricewge: network to ci.guix.gnu.org looks fine now, I didnt send anything yet. <lle-bout>so it looks like 9pzqxdr9j8vyfp3z4m52sqmknrynnw0h-vigra-1.11.1, a dependency of libreoffice fails to build on master <lle-bout>bricewge: hmm MTR says it's still losing packets. <efraim>it looks like at somepoint gcc finished building on powerpc and now I think I'm on guile-final <raghavgururajan>sneek, later tell apteryx: I have sent v2 to #40657. No propagation required. :-) <nagamalli>while I packaged rxnat , I checked with guix lint for errors it was showing below 2 errors <nagamalli>r-rxnat@1.0.8: use @code or similar ornament instead of quotes <nagamalli>r-rxnat@1.0.8: synopsis should be less than 80 characters long <nagamalli>I checked at the given line numbers I didn't understand about the @code and for the synopsis how can i minimize the length? <cbaines>nagamalli, what's the current synopsis? <nagamalli>cbaines : (synopsis "Queries and Extracts Images from Extensible Neuroimaging <efraim>I'd drop 'Public/Private Datasets' and see if it's still too long <efraim>the @code is probably from the description <efraim>it means to use @code{foo} instead of 'foo' or `foo` or similar <nagamalli>efraim: <I'd drop 'Public/Private Datasets' and see if it's still too long> <nagamalli>If I remove that Is there any problem in the future? <efraim>I'd probably also change the wording to 'Query and extract' and only capitalize the first word <efraim>nagamalli: shouldn't be a problem, the synopsis and description are mostly there for us humans <nagamalli> "Allows communication with Extensible Neuroimaging Archive Toolkit <cbaines>The line numbers just relate to where the field starts <cbaines>The quotes in the description don't look to mean anything, so I'd remove them. <Kimapr>i'm experiencing the issue with SSL not working in some apps again. I remember there was a patch somewhere, but lost the link (and also forgot how to use it) <Kimapr>why isn't the patch merged to upstream? <nagamalli>cbaines:yeah I checked actually for the line number , But I didn't undertsood where the error is? <nagamalli>guix/gnu/packages/cran.scm:21121:6: r-rxnat@1.0.8: use @code or similar ornament instead of quotes <efraim>it doesn't like the quotes around Rxnat and XNAT <cbaines>So that relates to the quotes around 'Rxnat' and 'XNAT' I believe <nagamalli>cbaines:Are u talking about the quotes, shall i remove them? <cbaines>nagamalli, Yes, I'd remove the quotes around Rxnat and XNAT <nagamalli>Yeah, I removed them now ..I will check with guix lint again <Kimapr>ah, i didn't finish profile update ) <Kimapr>(accidendial ')') will test after <nagamalli>Thanks cbaines and efraim , It's fixed actually! <janneke>hmm core-updates fails: make check TESTS=tests/print *janneke goes to check master <cbaines>Ah, and i586-gnu is a system now! Very exciting <cbaines>I need to fix the hard coded filter... ***apteryx is now known as Guest26536
***apteryx_ is now known as apteryx
<janneke>cbaines: i read your email with great interest and i wanted to share some encouragement <cbaines>ah, that one, I've sent out a few recently! <cbaines>(or rather these pages, there's one for each revision) <janneke>cbaines: what does "Unknown" mean, or eh; what can we do to learn more about "the unknown"? <janneke>oh, also on master: make check TESTS=tests/print <cbaines>janneke, it means that there's 0 or 1 narinfo files known for that output <cbaines>If you click on Unknown in the key to the right on x86_64-linux, you'll see the outputs, and what data is known about them <cbaines>but that's what I'm thinking about as well. There's some support for looking at build reproducibility in the Guix Data Service, but at the moment it's lacking data <cbaines>I need to see if I can get bayfront to build more reliably <janneke>cbaines: ah; so i see at least two reasons: it hasn't been built at all (why?) or only on one machine <cbaines>I think this is representative of what's actually been built, at least for recent revisions <janneke>it would also be interesting to know if some core set were 100% reproducible <cbaines>although there could be problems with how I'm querying for substitutes/builds <leoprikler>I have a question regarding GWL: It appears GWL installs guix in the user profile. Won't that cause problems? <cbaines>janneke, it would be nice to break up the problem a bit. I'm not sure along which lines to break it up though <cbaines>by module might be interesting, although I'm not sure how useful that would be <efraim>uh oh, something I'm packaging wants tensorflow-1.14.0 <janneke>cbaines: i guess what i'm saying is that this looks like very important and thankless (lonely even?) work <janneke>so these graphs/visuals are very nice; they could really help to get people involved <rekado_>leoprikler: it uses Guix as a library. If the user has ~/.config/guix/current/bin first on PATH then this shouldn’t be problem <janneke>cbaines: however, when i see "58% reproducible" ... that's pretty discouraging <cbaines>janneke, yeah, I want to get more data soonish <janneke>what i would need is a clear target/goal to work on (oops, "core set X" dropped below 100%) or, ... oh, 89%...we're getting there :-) <efraim>I wonder if we could sort it based on the number of dependant packages <cbaines>coupled with the work Ludo has done on guix challenge, I think it's becoming easier to investigate reproducibility issues. Although I haven't been doing this work at all. <efraim>I assume packages that fail to build aren't marked as reproducably failing to build :P <cbaines>efraim, the data I'm looking at is subsitutes/narinfos, so for build failures, there's no output <janneke>master: b44e19acd1a0d449427cea0333f02ac98605be1c <rbonnal>hello everyone. I used guix a long time ago, more than 2.5 years ago for experimenting. Right now I am using the GUIX OS inside a VM. I have created my personal profiles and then cloned the git repository because I want to update some package ( just for me now ). It is not clear to me, from the documentation, if I need to run the `./pre-inst-env guix-daemon` so having 2 daemons running at the same time in on my host. <cbaines>rbonnal, you don't, one guix-daemon process is fine <rbonnal>but `./pre-inst-env` is required for building a custom packages ? I am working in bioinformatics and I am going to update some package there. <cbaines>rbonnal, yeah, prefix commands with ./pre-inst-env so that you're using the code in the Git repository <janneke>=> 1475d45112243a37ae921df41977c4acf3064a6d is the first bad commit <janneke>rekado_: this commit "1475d45112 import/print: Return license with prefix." breaks: make check TESTS=tests/print <janneke>for me...it looks pretty harmless, can you have a look? <cbaines>efraim, there's a process running under your user on bayfront that's clocked up 238 hours of CPU time, is that something gone rogue? <efraim>cbaines: I killed it, not sure where it came from <janneke>rekado_: the "breakage" is merely in the resulting diff: (license gpl3+) => (license license:gpl3+) *janneke supposes that the baseline should be updated? <rbonnal>if I want to compute the sha256 for a package I need to specify the uri, but if the uri in the `.scm` is defined by pypi-uri package_name version; how do I type that in the cli ? <janneke>i found that in tests/print: + #:use-module ((guix licenses) #:prefix license:) <janneke>but i'm not sure which way you want it to go :-) <janneke>and then using: (license license:gpl3+), of course, twice <rekado_>pushed a fix. Thanks for letting me know! <janneke>rekado_: beautiful, thanks for the quick fix! <cbaines>I'm trying out the 1.1 installer, but I'm getting an error from grub-install <cbaines>Something like: this GPT partition label contains no BIOS Boot partition: embedding won't be possible <efraim>9.5 hours in so far on guile-3 on powerpc <cbaines>I left the partitioning as the defaults <mothacehe>cbaines: That could be because the EFI partition is not mounted. Could you jump to a shell (ALT-F3) before starting the install and report what 'mount' says? <cbaines>mothacehe, I've still got the failure up, can I switch to the console now, or should I restart? <cbaines>hmm, I'm not sure that key combination is working in the web console thing I'm using <cbaines>^[OR appears on the screen, but I guess that's not supposed to happen <mothacehe>The goal is to switch to TTY3 (if there's a dedicated button) <cbaines>ctrl-alt-f3 switches on my host system <cbaines>still no luck switching tty, but I have got back to the partitioning screen <mothacehe>What's the proposed layout, is there a small partition with bios_grub flag? <mothacehe>Ok thanks. Then the installer is locating an existing ESP partition (the first one). <cbaines>This is a hosted virtual server, the disk will have probalby had Ubuntu installed when I started the installer <mothacehe>Now, given the error message you had, it must fail to mount it for whatever. <cbaines>because I'm using the installer through a webpage, that just changes the tty on my machine <cbaines>To work around this issue, if I delete all the partitions, then go back in to the partitioning phase, will that help? <mothacehe>Ok I see. Do you want to keep the installed Ubuntu system. If not you can try the "Manual partitioning mode". <mothacehe>You can create a first partition of 600M with ESP flag on. <mothacehe>And create an other partition mounted on "/". <nagamalli_>Hi , I actually got some doubt while generating package definitions for cran , while some packages like poio and silicon there are other defintions also generating along with the asked packages <nagamalli_>when i copied only one package defintion of whatever i want and running guix lint and guix build , it was showing poio package was not found like that. <nagamalli_>I tried with all package definitions they have given while i was imported , them also it was showing like that. <nagamalli_>can someone help me to understand why this was happening? <cbaines>mothacehe, the guix-daemon and cow-store services were restarted <cbaines>mothacehe, and now I've got an odd error: guix system: error: opening file '/gnu/store/...-loadkeys-static-2.0.4.drv: No such file or directory (I've omitted the hash) <cbaines>I guess something might have got a bit confused, shall I just restart the installation? <cbaines>I'm also happy to try and debug the original problem more, I just need to figure out how to change tty or otherwise get to a console <mothacehe>If you can find a way, that would be really nice :) I can also try to create a layout that looks like the one you sent and see if I reproduce in a VM. <cbaines>mothacehe, I wonder if there is a way to add additional key combinations for changing tty <mothacehe>In QEMU there is (it's CTRL-ALT-2, then sendkey ctrl-alt-f3), maybe there's a similar mechanism. <karmayogi>Can I launch applications in guix in separate container with separate network namespace <cbaines>karmayogi, I don't think there's any reason why not :) <karmayogi>What I want to achieve is use internal ip range to run multiple instances of nginx and run them on port 80 with different IP address <karmayogi>I did find the documentation of running application in containers, but didn't see any help on how I can run this container and assign it a different IP address to run daemons on port 80 for multiple application in same guix system. <cbaines>karmayogi, as you say, there is some "conatiner" support, but I've got no idea how to link up the network namespace for a service with the networking config <cbaines>nagamalli_, I guess you're using guix import cran. Are you trying to add the package definitions to the modules in the Guix git repository? <cbaines>karmayogi, guix system container like guix environment can do two network namespace related things. Either use a new network namespace for isolation, or not. <cbaines>karmayogi, have you looked at libvirt networking stuff? <cbaines>mothacehe, I've got the same error somehow: this GPT partition label contains no BIOS Boot Partition; embedding won't be possible <leoprikler>why the *beep* hardcodes gwl/cache.scm ISO-8859-1 in workflow->data-hashes? <cbaines>nagamalli, did you see my message earlier? <nagamalli>I just lost my internet connection? Did I miss anything before? <cbaines>nagamalli, all I said was: I guess you're using guix import cran. Are you trying to add the package definitions to the modules in the Guix git repository? <nagamalli>I am doing it locally and adding to git after the completion <cbaines>Ok, great, and what's the current error you've got? <nagamalli>When i am trying to import package defintions i am gettig multiple definitions of different packages also along with that <cbaines>I'm not familiar with the cran importer, which package are you trying to import? <nagamalli>suppose I did for silicate package , I got silicate package definition along with other <cbaines>When I run guix import cran silicate, I just see one package definition <nagamalli>My problem is when i am running guix lint after adding it to cran , it was showing package was not found <cbaines>Ok, that makes sense, I guess some dependencies are missing <efraim>aww, karmayogi left, I had code examples for them <cbaines>efraim, I'd be interested to see, you could also leave a message with sneek <nagamalli>It's occuring for other packages also,But i think it's not be a problem , It's occured for poio package also <cbaines>nagamalli, what I believe you'll need to do is use the importer to generate definitions for the missing packages too <nagamalli>cbaines:how can i know the missing packages? I am just beginner into this,That's why i am getting doubts like this. <efraim>cbaines: BNW is run as a guix system container on top of debian, managed using systemd <cbaines>nagamalli, I'd try building the packages, and look out for where Guix says something is undefined <bnw>~_~ Got mentioned a few times <nagamalli>yeah, I Didn't understood why it wa coming like that , That's why I tried with one another pkg also but it was same as previous <cbaines>nagamalli, when you try and build the package you're trying to add, what error do you get? <nagamalli>I have taken only silicate package definition , and added to cran.scm file <nagamalli>after that I made the correct format and saved that and runned guix lint for error checking <nagamalli>It's ran upto sometime and after that finally it was stopped by saying guix lint: error: r-silicate: unknown package <nagamalli>cabaines: I just impoted the pkg definition by using $guix import cran -r <package> <nagamalli>cbaines:sry, I was troubling with my internet issues <cbaines>nagamalli, when I try adding r-silicate, I had to add a prefix to the license, but after that I get: error: r-crsmeta: unbound variable when trying to build the package. <cbaines>I can't see a package for r-crsmeta, so I guess that needs importing too <nagamalli>Yeah I am using this command only guix import cran -r <package> <nagamalli>I am getting also , And after adding to cran.scm file I just ran guix build using ./pre-inst-env guix build r-silicate <nagamalli>guix build: error: r-silicate: unknown package <nagamalli>This is hapening everytime, I am unable to understand why it was coming like that? <rekado_>in cran.scm all license values need to be prefixed with “license:” <rekado_>that’s because at the module declaration cran.scm imports (guix licenses) with that prefix <rekado_>the importer doesn’t and cannot know this <rekado_>when adding things to cran.scm you’ll have to add that prefix manually <rekado_>so instead of (license gpl3+) you’d write (license license:gpl3+) <nagamalli>By referreing the previous packages formats! <rekado_>you probably get more output than just that one error <nagamalli>By the way , I was done with 2 patches actually and this is the third one for today ! <rekado_>(sorry, getting used to this new keyboard…) <nagamalli>rest , It's too long actually , Can i paste it here? <rekado_>please use a paste service like paste.debian.net or similar <rekado_>got to go now, but I’m sure someone here can help you with more context *janneke finally produces an image and pushes updated wip-hurd-vm <cbaines>nagamalli, that seems OK, can you create a new paste with the full output of ./pre-inst-env guix build r-silicate <mothacehe>cbaines: too bad, sorry about that :( Could you file a bug report? I'll see if I can manage to reproduce it. <cbaines>so running make will hopefully help with the ;;; note: source file ... messages <cbaines>There's this message in the output: /home/nagamalli/Documents/outreachy/Guix/guix/gnu/packages/cran.scm:21088:2: warning: invalid field specifier <cbaines>Ok, so maybe one of the fields in that package is invalid <nagamalli>I had pushed that package into guix already..! <cbaines>What package definition do you see around line 21088? <cbaines>Ok, and does git diff show any uncommitted changes around there? <nagamalli>cbaines:I don't know exactly once check in that paste? <cbaines>nagamalli, that seems OK, I'm not sure what to suggest. I'm guessing there's something somewhere in one of the package defintions that's preventing Guix from seeing r-silicate <emys>hi, I tried the systems installation on a brand-new mini PC. Burnt the ISO image to a DVD and now booting from an external USB drive. Grub comes up and my process starts, but I get the following errors and the boot process stalls. <cbaines>if you're still having problems, could you share everything from 21086 until the end of cran.scm <emys>after that, the system seems to idle /hang <emys>"[0....0...] kmscon Revision UnkownRevision Jan 1 1970 00:00:01 (log_print_init() ...)" <cbaines>emys, so is this the installer booting, or the installed system? <emys>cbaines, the installer booting <emys>from an external DVD drive <cbaines>emys, unfortunately I'm not sure what to make of that output. What graphics hardware does the system have? <nagamalli>cbaines: Have you seen that paste of git diff? <cbaines>nagamalli, yeah, nothing was obviously wrong to me <cbaines>nagamalli, if you're still having problems, could you share everything from 21086 until the end of cran.scm <rbonnal>rekado_: I would like to try to build the lastet scanpy from github because they fixed EBI downloads. Unfortunately they did not yet release the version on pip. <efraim>bnw: sorry I didn't mean to mention you, I had a package named bnw so your name came up a couple of times. sorry <rbonnal>more in general it they install the software using `pip install -e` from source, how can I do that from guix ? <rekado_>rbonnal: you don’t use pip install from guix <rekado_>but you can simply fetch the sources from git <rekado_>so you’d override the “source” field and use “git-reference” as the “uri” of the “origin“, and use ‘git-fetch’ instead of ‘url-fetch’ for ‘method’. <rbonnal>rekado_: ok, put in this way, building the python-scanpy package fetching the latest commit from git I get a LookupError from setuptools-scm that was unable to detect the version <rekado_>rbonnal: can you show me the full error message, please? <MYMYRSRS>Can I share /gnu folder with multiple machines? <rekado_>if it’s a static setup you can also export /gnu over NFS and mount it read-only on other hosts. ***rekado_ is now known as rekado
<MYMYRSRS>I installed Guix on the server system, and then installed Guix in my lxc container. Can they share a / gnu? <MYMYRSRS>Because there are two Guix, my server disk is not enough <rekado>you can mount /gnu into the container, no? <MYMYRSRS>I was thinking, will it conflict? For example, guix gc destroys another system <rekado>mbakke: oh, didn’t know there was an actual bug report *civodul is in a shepherd cleanup mood <roptat>civodul, I think I understand what happened <roptat>%host-type is "arm-unknown-linux-gnueabihf" <roptat>so linux-gnu is not a suffix, so it chooses the hurd variant <civodul>so we need string-contains, not string-suffix? <cbaines>any objections to adding the prometheus-node-exporter service to bayfront? <roptat>wouldn't that mean rebuilding the world? <civodul>cbaines: is it any useful on its own? <roptat>isn't that file on the build side? <cbaines>civodul, great :) and yeah, not really, normally you have prometheus scrape the data frequently <civodul>roptat: it's not used by packages, so we're fine <roptat>yes, I noticed it yesterday, and sent a bug report too <cbaines>(with a Guix package and service, I haven't handled the go vendor mess yet though...) <cbaines>I've only just started collecting data though <cbaines>and that Prometheus instance is scraping the data every 15 seconds <cbaines>and I can now see the disk space on the graph I shared previously <cbaines>mothacehe, hey, I managed to make it though the installation process eventually! <cbaines>mothacehe, I blame Hetzner a bit, as I tried a different datacenter, and that worked better. I also asked for a Debian 9 VM, and installed over that, I think that meant the installer used BIOS rather than EFI. <nagamalli>cbaines: HI , Here I am sharing from the warning line where it was showing, I don't know why this was happening like that <nagamalli>I hav tried 3 more packages , but still same is going on and previously i did more tha 6-7 packages and successfully pused to guix. <cbaines>nagamalli, this line looks problomatic: <<<<<<< HEAD <cbaines>That looks like git was trying to resolve conflicts? <cbaines>this line as well: >>>>>>> a8f01c45c627f92f9fa1ab8b0c8de722ecd1e4df <nagamalli> Yes, I have solved it by using one command actually! <roptat>oh well I can't find the bug I sent yesterday anymore... was it all a dream? <mothacehe>cbaines: ok, then maybe we were in strange setup were you had an ESP partition but no EFI support. <nagamalli><this line looks problomatic: <<<<<<< HEAD> If I remove that Is there any problem further? <civodul>really helpful for the person who writes NEWS <cbaines>nagamalli, I'm guessing you might have committed those lines, so that might need addressing <cbaines>nagamalli, but I'd start by removing them <civodul>ah ah, it's a terrible threat, isn't it? :-) <nagamalli><nagamalli, this line looks problomatic: <<<<<<< HEAD <nagamalli>this line as well: >>>>>>> a8f01c45c627f92f9fa1ab8b0c8de722ecd1e4df> <nagamalli>Yeah Thanks , I am going to remove those now , I will try again using guix build! <nagamalli>cbaines: I tried guix build by removing those all,But it remains same error..! <cbaines>nagamalli, the same can't find the package error? <nagamalli>cbaines: yes , guix build: error: tea: unknown package <cbaines>nagamalli, actually, the error was: guix build: error: r-silicate: unknown package before right? <cbaines>Does it actually say "tea: unknown package" ? <nagamalli>cabines: I Have tried with different packages , to understand the error , but same was occuring evary time , as i told before <cbaines>Ok, I'd have another look at the package definitions and see if you can spot anything wrong <cbaines>If you can't, share the bottom on cran.scm like before <thecom>hey, can somebody help a n00b? i fail to find the guix package providing libstdc++.so.6. if i search in places like /gnu/store/4sqps8dczv3g7rwbdibfz6rf5jlk7w90-gcc-5.5.0-lib/lib/libstdc++.so.6 but i guess, this is obsolete? <cbaines>nagamalli, do you see anything like: warning: duplicate field initializer <thvk-ivgf>thecom: if that is inside of gcc-xyz, then you most probably want gcc-toolchain (default version is 7, you may want gcc-toolchain@9) <cbaines>nagamalli, because the r-asd package definition is invalid, you're declaring several fields twice <cbaines>nagamalli, look at each field, and check if it appears again in the package definition <cbaines>this will have happened when resolving (or not resolving in this case) the Git conflict <sirmacik>hey guix! can someone remind me how to define kernel version to install in config.scm? <sneek>Welcome back sirmacik, you have 1 message! <nagamalli>cabines, Yes It's repeated twice I will remodify that, thanks! *ecbrown goes bonkers and guix import cran's everything, discovers Too many heap sections: Increase MAXHINCR or MAX_HEAP_SECTS <nagamalli>cbaines, Thank you so much for your patience through out helping me into this <nagamalli>it was solving I think, I am not getting any error like before <cbaines>You're welcome, glad you got it working :) <ecbrown>civodul: added a "my-cran.scm" to local.mk, `make' <ecbrown>(inside ~/src/guix, which is git clone of guix source) <nagamalli>cbaines , can u suggest me what would be better if i remove in the description , when i used guix lint i got it was 92 characters long <nagamalli> "Different approaches for selecting the threshold in generalized Pareto distributions. <nagamalli>Most of them are based on minimizing the AMSE-criterion or at least by reducing <nagamalli>the bias of the assumed GPD-model. Others are heuristically motivated by searching <nagamalli>for stable sample paths, i.e. a nearly constant region of the tail index estimator <cbaines>nagamalli, for the description, that'll be a formatting issue. It's line length, not the overall length. <cbaines>nagamalli, for the synopsis, the overall length matters, as it's meant to be short. <rekado>rbonnal: oh, I’ve seen this before. That’s setuptools-scm complaining about not working with a real git repository, because we delete the .git directory for reproducibility reasons. <rekado>you probably need to patch scanpy to tell it not to try to use git information. *rekado made some progress with guile-xapian for indexing the debbugs data directly <asmodehn>hi everyone, I was wondering if there is a way to produce a config.scm file from a running GuixSD system ? <civodul>however, since 1.1.0, there's /run/current-system/configuration.scm <civodul>that is, "guix system reconfigure" and "init" save your config file <asmodehn>guix system describe doesnt give me the location of the config file... should it ? <ecbrown>civodul: env GC_INITIAL_HEAP_SIZE=48G make powers through it. 50G memory high-water mark. (probably need a new strategy ;-) <cbaines>ecbrown, could you have loops in the package graph? <jayspeer>asmodehn: running guix system describe gives me in the last line: "configuration file: /gnu/store/17g0f9rbjmf2ka5yp4c3xv9mivrbfja6-configuration.scm" <ecbrown>cbaines: just a naive loop over "guix import cran ${i}" to a .scm file, and a little touchup. <cbaines>ecbrown, guix will just use more and more memory if you have a loop <cbaines>ecbrown, then at least you'll know if you've got loops or not <rekado>cbaines: I’d like this loop detection to be part of Guix <rekado>ecbrown: r-knitr is probably the reason <jayspeer>once again I face this issue - my user on foreign distro debian is missing the "current-guix" but guix-profile is there <rekado>a lot of packages will want to use r-knitr for vignettes, but r-knitr may depend on these packages <rekado>jayspeer: did that user not run ‘guix pull’/ <cbaines>rekado, not that I'm aware of. I don't think it's very neat, but I do find it useful occasionally <cbaines>it would be good to look at again soon <rekado>jayspeer: current-guix is a profile managed by ‘guix pull’. <jayspeer>so let's get this clear - on foreign distro every user has to run guix pull? <rekado>if they want to manage their updates by themselves they’ll have to run ‘guix pull’ <jayspeer>ok, so if guix-daemon is running under root on debian, both me and root has to run pull <rekado>this is independent of ‘guix-daemon’ <rekado>users can also share the same Guix, e.g. the one maintained by the root user. <cbaines>jayspeer, the init script might reference root's Guix, but you can change that if you'd like <jayspeer>but if I run guix pull then guix-daemon can gc my stuff, isn't that true? <rekado>for ‘guix gc’ to destroy anything it first needs to be marked as unused <jayspeer>I shortcutted what I've wanted to say, let me rephrase that <jayspeer>s1 told me on this very channel that if guix-daemon runs under other user, then running guix pull does not update the daemon <jayspeer>and so it can get rid of stuff I've pulled earlier <jayspeer>or does anyone can just run guix pull like it's nothing? will it update the daemon? <cbaines>what do you mean by "so it can get rid of stuff I've pulled earlier"? <cbaines>the garbage collection behaviour is independent of which user the daemon is running as, or where the guix-daemon binary is <jayspeer>this discussion was long time ago, so I don't remember the details. But the conclusion was: one has to run guix pull as the user who owns the daemon, usually the root (on foreign distro) <cbaines>I can see how that could be necessary to update the daemon, but it doesn't have to be <jayspeer>what was happening at the time on my one of my machines was: running guix --version returned different values depending on which user run this command <cbaines>that makes sense, different users can have different versions of Guix <jayspeer>well, it caused my guix install to malfunction at the time <jayspeer>but, scratch that, we can dig through logs and wonder what was wrong <jayspeer>I have an issue with .config/guix/current being non present for my user <cbaines>jayspeer, that would usually be a symlink. Is it a symlink, or completely missing? <jayspeer>I've tried deleting my user's guix stuff, and now it point to not existing file <jayspeer>to /var/guix/profiles/per-user/jayspeer/current-guix, but this is not present on my system <cbaines>OK, I think guix pull should manage that <cbaines>Have you run guix pull as your user? <jayspeer>I've tried but it just hangs without terminating for a long time <jayspeer>the guix process just hovers aroung 2% cpu usage <cbaines>without doing anything, or does it at least output something? <jayspeer>it prints "updating channel 'guix'..." line, but then it's nothing <jayspeer>usually after this message one can see what it updates <rbonnal>rekado: (setuptools-scm) I need to dig more into it, in any case thanks for the suggestion. In case you remember if you already pacthed some other package in the past, plese let me know. <cbaines>jayspeer, so the updating channel bit is when it updates the Git repository <cbaines>that could be slow, depending on your internet connection <jayspeer>this goes blazing fast as a root, and on my other machine <efraim>I was away from my computer so I didn't get the exact time, but guile-3.0 took more than 12 but less than 17 hours to compile on my ibook g4 <cbaines>jayspeer, perhaps try deleting .cache/guix/checkouts or .cache/guix then trying again <cbaines>jayspeer, also what do you get if you run guix describe as your user? <cbaines>hmm, OK, one approach here is to use another users guix to do the pull <cbaines>then you avoid using this guix of inteterminate origin <cbaines>jayspeer, if you run type -p guix as root, then run type -p on the filename that gives you, until you get to an item in the store <cbaines>you can then take that guix from the store, and as your user use it to run guix pull <jayspeer>I'll but it shouldn't help, because I was using symlink /usr/local/bin/guix -> /var/guix/profiles/per-user/root/current-guix/bin/guix to run guix <cbaines>ah, and are you using /usr/local/bin/guix as your user? <cbaines>or is guix maybe in your users profile? <jayspeer>I was using /usr/local/bin/guix because the one that should be in .config/guix/current/.../guix is not there <jayspeer>basically the "current-guix" is missing for my user <cbaines>running guix pull is the answer I believe <cbaines>did you try deleting the cache directory and running it again? <cbaines>it has to clone the repository, so it might take a while <jayspeer>I left for like half an hour, git clone is definatelly faster on my connection <alextee[m]>is there a way to use discord with a free client on guix? <alextee[m]>i saw a plugin for pidgin but i couldn't figure out how to package it <cbaines>I only suggested deleting the cache directories 10 minutes ago <jayspeer>I'll try once more -- I've cleared the cache and I've started guix pull <cbaines>I've deleted .cache/guix/checkouts/ and .cache/guix/pull on my system, and I'm running guix pull to see what behaviour I get <jayspeer>cbaines: thanks for that; guix pull still chugging on, cache is at ~32MB <cbaines>It's now started building things on my machine, and ~/.cache/guix/checkouts contains one directory which is 187MiB in size <jayspeer>mine is still only at 80MiB - seems weird for ~200mbit network <jayspeer>cache reached just under 200MB; under pulling started - seems I was a bit impatient, sorry for that cbaines <cbaines>jayspeer, no problem, I'm glad you've got it working :) <jayspeer>I wonder though what was going on that caused guix pull to get stuck before cleaning the cache - I'll need to brake it someday again <jayspeer>yeah! I've finally have guix from my guix-current in path, thanks again cbaines <cbaines>given the repository is cached, the next time you run guix pull as your user, it shouldn't take as long doing Git things <jayspeer>also, I think this is the best channel to connact guix staff -- download page still point to version 1.0.1 <lfam>Which URL jayspeer? I can't reproduce it. Maybe your browser has cached the old version? <jayspeer>of course it goes ok after few days, right after I mention it and embarras myself ;P <lfam>Did it lead you to download the wrong version? <jayspeer>I've been trying to build hurd image for some time now, but get stuck here and there. Is the blob post still up to date with the stuff in git? <jayspeer>I've tried building from git branch wip-hurd-vm but "./pre-inst-env guix build -f gnu/system/hurd.scm" fails with "Unrecognized keyword: #:file-system-options" <rekado>jayspeer: which commit are you on? <rekado>the branch is often recreated, so it’s a moving target <jayspeer>I'll trying rerunning ./bootstrap and ./configure first -- perhaps I've missed something <lfam>Make sure to run './configure --localstatedir=/var' <jayspeer>still same error :( I'm on HEAD: wip-hurd-vm <bricewge>Could I get a review for some of my patches? ***ChanServ sets mode: +o lfam
<bricewge>Both are straight forward and not controversial (I think) <jsoo>How many people here use fish for shell (login or not)? <James_Epp>Hey, folks. I've intermittently been using Linux through the years. I'm looking to get my daily system back onto a GNU/Linux system and have been playing with a bunch of different ones. For my needs, I believe I require the nvidia proprietary drivers - are these available in the Guix repositories? <James_Epp>That's what I expected. May not be worth my effort then. Thanks! <Blackbeard>James_Epp: you can always use guix as a package manager on top of another distribution if you want <raingloom>ughm, how do yall use gpg? it can't find pinentry on my machine. <jsoo>raingloom: i have both pinentry and gpg in my user profile. <raingloom>jsoo: then i guess it only picks it up if it's installed. thanks! <raingloom>btw i tried to use fish as a login shell for a bit, but realized it won't work with the generated profile scripts. <jsoo>raingloom: i use it as a login shell and it works quite well save for one outstanding bug with grafts <jsoo>it now uses fish-foreign-env to source the profile scripts. <jsoo>+1 raingloom it's working very well for me <jsoo>i think also in the future guix could easily be used as the package manager for fish <invalidlicensed>You are running non-genuine GNU/Linux copy. PM to purchase a license. ***invalidlicensed is now known as validlicensed
***validlicensed is now known as invalidlicensed
***lfam sets mode: +r
<Blukunfando>I’m pretty sure you can print as many copies of free-software licences as you like and sell them. <bricewge>That's sound like a good business plan! Where can I invest? <glat-licensed>For those who don't know, you have to purchase a license to use GNU/Linux. ***lfam sets mode: +b *!*@gateway/web/irccloud.com/x-vpbdwiyyybtxdvys
***ChanServ sets mode: -o lfam
<lfam>@bansearch glat-licensed *lfam testing the latest git <vagrantc>wow, i've been submitting patches to guix for almost two years now. didn't realise it has been that long <civodul>vagrantc: i hope it's been a pleasant journey! <vagrantc>not without it's moments of frustration, but overall, having a really good time with y'all. :) <janneke>vagrantc: has it been two years, amazing <vagrantc>i just noticed my earliest commit was april of 2018 *vagrantc tests the latest u-boot mainline patches for pinebook-pro-rk3399 <ecbrown>cbaines: i applied the patch and am recompiling `make' what should it do differently? <cbaines>ecbrown, nothing, just the same thing you did before <cbaines>if that code works, it should do quite quickly <ecbrown>ok. there's like 15,000 packages in there so i'm giving it time <ecbrown>rekado: thanks for the tip about knitr. wonder if that means to clear out r-knitr from any dependecies of r-knitr. <ecbrown>rekado: this could also be a fool's errand, all of cran :-) <rekado>I suggest to not add 15k packages at once <ecbrown>yeah, perhaps useful template and/or learning <ecbrown>cran is a black hole of interesting vignettes <ecbrown>thx for r-bridgesampling.. i'm going to work on brms and others. cheers