<nckx>civodul: Is it acceptable to start a description with an uppercase @acronym{}? I say ‘of course’, guix lint says boo hoo. <NieDzejkob>ngz: It's not X.Y vs X, it's 0.X vs X, because that's how rust-semver does compatibility <ngz>NieDzejkob: Does that explain why there is rust-serde-json-1.0 and rust-clap-2 for example? <chromebooksla>Installing Guix on a small capacity disk, is there a way to save space? <pkill9>regularly deleting old generations and running `guix gc` ***lispmacs` is now known as lispmacs
<lispmacs>does anybody know which package exactly contain the gio-launch-desktop program? <brettgilio>Goodnight, Guix! I worked a double shift today, so I am tired! <str1ngs>lispmacs: umm I'm not 100% sure this. but I think that might program service run by dbus <str1ngs>lispmacs: basically it's a RPC call not a program <str1ngs>lispmacs: actually its found in glib:bin <efraim>NieDzejkob: I read the backlog about rust semver compatibility, should rust-serde-json-1.0 be rust-serde-json-1? (and the similar packages) <hexagona`>one quick question - what does one do about #! in a file? <hexagona`>I have a shell script with #!/bin/bash but when I run it, it says that /bin/bash can't be found. <madage>change it for #!/usr/bin/env bash <leoprikler>if you're packaging it, this is also done by patch-shebangs (or rather, it hardcodes bash) <eiro>also: from the doc: The sha256 checksum of the requested file. This is essential to ensure the source is not corrupted. Note that Guix works with base32 strings, hence the call to the base32 function. <leoprikler>fwiw english still works, has the cookbook been translated yet? <eiro>which means: sha256sum work? <leoprikler>sha256sum may work, but it's probably better to do `guix download` or `guix hash` <nckx>eiro: We use standard sha256 but not standard base32 (different characters), so use the two commands above. <7GHAAGGRV>If I want to add a package, but I need to write two additional definitions cause it's dependencies don't exist in Guix, do I send each package separately in its own patch, or do I bundle the three commits into a single patch? <eiro>guix download and guix hash while guix edit ... i have to revert things <nckx>eiro: Both guix hash & guix download take an argument of the thing to [download and] hash; they don't look at the current Guix package set. Unless your edit left the ./pre-inst-env guix in a broken state, then you can just ‘git stash’, run guix, ‘git stash pop’. There should be no need to revert. <eiro>damned! i screw the file up while using guix edit, i guess <eiro>hope guix pull will save me <eiro>ok nckx. thanks for the details <leoprikler>Why does this sound like you're not working on a local git checkout at all? <eiro>oh ... i forgot this ... actually it was a https:// link. i was wondering if i would be able tu push <eiro>maybe i have to create an account on savanah <leoprikler>you won't be able to push your commits to savannah either way, patches go through the ML 😉️ <NieDzejkob>If I reference packaging from other distros, where do I put the credit? <NieDzejkob>the build commands used. PKGBUILDs, Nix recipes (is that the word?) <leoprikler>there should usually not be a need to reference them at all, but if you do have some very exotic setup, putting a link to an explanation directly in the comment is not a bad idea <leoprikler>that said, the link should explain WHY some other distro does this, not what said distro does <str1ngs>generally unless if it uses standard upstream build steps there is no need to mention the other distro. if though you need to do technical changes in order to make the package work, or with with guix specifically. just comment what those technical changes do. IMHO <valignatev>Hey #guix, just noticed that ripgrep has landed with all its dependencies to upstream. This is super-awesome and encouraging, thanks a lot to all involved! I'm trying to package alacritty and ripgrep example should be super helpful for this. <sneek>Welcome back valignatev, you have 1 message. <efraim>valignatev: when it comes time tag me with the patch also. I can't promise anything but i'll try to review it <sneek>Welcome back civodul, you have 1 message. <sneek>civodul, wingo says: maybe we can add a third option to GUILE_INSTALL_LOCALE, to install without warnings, and make that option the default <civodul>nckx: it's great that adding support was (apparently) quite smooth <Parra>how is it possible that guix deletes npm logs? <Parra>in theory they are written to /tmp/.npm/_logs <Parra>but after the build, I go there and there is nothing <janneke>Parra: if you use `guix build', that happens in a container <janneke>so guix does not delete anything, they were never in /tmp <Parra>how is that possible? npm says the opposite <Parra>in fact I'm using guix inside a container right now <bavier>The /tmp dir is bind-mounted somewhere else inside builds <Parra>how can I find the real route then? <janneke>Parra: i have no ready-to-use recipe ***ng0_ is now known as ng0
<Parra>I think guix is great but if you go outside C/C++ it's a nightmare <janneke>you may have to modify the package build recipe to convince npm to store its logs somewhere else <Parra>or maybe I'm too noob, I don't know <janneke>yeah, guix wants to be pure and that can feel very annoying at times <janneke>if things are annoying enough, chance is someone will create a solution <janneke>Parra: possibly add something like: (setenv "TMPDIR" (getcwd)) to a 'setenv stage in your the package build, use the --keep-failed flag and ^C the build before it ends <janneke>ah, just missed that you were gone... <Parra>in any case, I need to see that logs <Parra>(it was my internet connection) <janneke>i tried to say : Parra: possibly add something like: (setenv "TMPDIR" (getcwd)) to a 'setenv stage in your the package build, use the --keep-failed flag and ^C the build before it ends <Parra>uff the ^C will be difficult because it fails and ends almost at same time <janneke>if the build fails, no need to ^C; just use --keep-failed <Parra>I dont really care about reproductibility <Parra>I'm only interested by the self contained and packaging methodology of guix <Parra>if it isn't reproductible, I will solve.that in the future <Parra>have you ever tried to build a npm package with guix? <Parra>maybe it would be better to download the code and paste it there <Parra>it's failing to run node commands <Parra>they are just commands that are js scripts located in node_modules <janneke>have you managed to inspect the log/stop the build? if so you can try to run those commands by hand <janneke>i only ever built very simple npm packages; working on improving the npm build system by jlicht inspired me to move away from node and npm to guile a couple of years ago <janneke>things have improved since then, and i welcome progress in that area <nckx>civodul: ikr. It was almost disappointingly easy. Almost 😉 *nckx happily invites similar disappointment with bcachefs, but adding multi-device support will be hairy. <NieDzejkob>Does anyone here using termite know how to disable scrollwheel escape sequences? <NieDzejkob>Alternatively, does anybody know a terminal that allows it? <janneke>emacs shell allows it, but that's hardly a terminal; probably emacs eterm? *janneke usually disables all mousy things ***lispmacs` is now known as lispmacs
<brown121407>Is there any work currently being done towards packaging new versions of Rust? I see our latest is 1.37.0, but 1.40.0 came out on Dec. 19 ***lispmacs is now known as lispmacs[work]
<janneke>brown121407: i don't think so; there are bugs open about rust building reproducibly and about cleaning up or simplifying the rust bootstrap path <kirisime`>patchelf: getting info about '--print-interpeter': No such file or directory <nckx>kirisime`: Spot the typo. <nckx>(Hint: his name is Peter.) <nckx>Not that the error message isn't misleading. <nckx>Who's patchelf upstream? Nix? <kirisime`>No such file or directory is what you get when you run a binary with the wrong interpreter. <bavier>patchelf should warn about an unknown option, not treat the misspelled option as a file argument <kirisime`>It should also have those autocomplete things for options. <nckx>Bash-completion is a separate (but valid) feature request. <nckx>Just more work than fixing the misleading error. <apteryx>hello Guix! Would someone know of a way to create an inferior for the Guix currently used in a profile? <apteryx>To add some context; I'm attempting something a bit strange; defining a new Guix package that reuses the inputs of the Guix that is used to build it. <raghav-gururajan>I am getting a build error "ERROR:mock-httpd.c:117:govirt_httpd_run: assertion failed: (error == NULL)". <apteryx>some test failed, and one side of the assertion is not expected <apteryx>raghav-gururajan: yes, but this disables every tests. To disable a particular test, it depends of the test framework being used. Sometimes you can define a command line option to have them skipped, sometimes there's no other way than editing the source itself to delete or otherwise disable a given test. <nckx>raghav-gururajan: It depends on the package and its build system. Often you add a phase (before 'check …) that either patches a file to remove a test from the list, or sets an environment variable to the same effect, or…. <nckx>There's no on-size-fits-all answer but many examples in gnu/packages. <pkill9>It's good to have another one of us. *raghav-gururajan is on a roll today. <grillon>it works on my foreign distro but on guixsd I have an error like symlink already exist <nckx>sneek: later tell grillon: That's a know bug with old versions of Guix (and, unfortunately, 1.0.1 is ’old’). The solution is to simply delete the offending symlink and pull again. <sneek>Welcome back grillon, you have 1 message. <sneek>grillon, nckx says: That's a know bug with old versions of Guix (and, unfortunately, 1.0.1 is ’old’). The solution is to simply delete the offending symlink and pull again. <nckx>Great. Yes, there have been many installer fixes since 1.0.1 and everyone agrees there should be a new release in the next few weeks, but a system installed from 1.0.1 is still 100% fresh once you've pulled, reconfigured & upgraded. <NieDzejkob>documentation fixes are also important since they will land on the website <grillon>it's doing a kind of big upgrade, is that normal? because of new version? <grillon>not so big and guix pull is ...empty :/ <nckx>bandali: Your year-old system will not be 100% fresh as long as it's a year old, that is true. <nckx>We should probably push mandatory updates like other OSes for the good of all users. <nckx>Unfortunately this security feature will only be available to paying subscribers to Pull Plus™. <grillon>what's the best way to have other input? ibus? <nckx>grillon: --news is only for major announcements & breaking changes, not package updates. <nckx>grillon: ibus is probably the best supported. I've heard reports of it being broken with latest GNOME. <grillon>nckx: on my foreign distro guix pull --news was huge <nckx>bandali: PM me your bank details & root passwd kthx. <nckx>grillon: Then I guess your guix installation was (much) older 🤷 <bandali>nckx, sure, one moment. anything else i should include while i’m at it? <grillon>nckx: may be, I guess it's not a good idea to do a guix install while a guix reconfigure? <nckx>grillon: That's fine. They will not conflict. <nckx>guix pull --news lists all the new things you last *pulled*, not ones that you've ‘installed’. <nckx>bandali: No, we take your privacy very seriously. <nckx>I'll change your root password for security. <oriansj>nckx: yeah set it after your first born kid abcde ***sneek_ is now known as sneek
<bandali>nckx, ah okay good. don’t forget to *not* tell me afterwards, i clearly can’t be trusted with it <nckx>bandali: Thank you for your message. Unfortunately, you've used up your free support credit for 2020. Please upgrade your plan to continue or legal action will be taken. Happy holidays! <lispmacs[work]>does Pull Plus have any modern features? I feel like with a modern OS updates should be randomized according to a cryptographically secure algorithm <lispmacs[work]>so nobody, least of all the user, should know when to expect them, or when their computer will suddenly reboot <bandali>i wish you a *very* happy holidays too <lispmacs[work]>preferably using entropy pulled from my TPM temperature flucuations <nckx>bandali: Hostility detected. A crack team of systemd units has been dispatched to your location. Please remain calm. <nckx>lispmacs[work]: We seed your TPM with 0 for reproducibility tho'. <drakonis>i have a craving for guix after running nixos for a while <nckx>drakonis: Welcome! (back) 🙂 <drakonis>nix is pretty good but it has a lot of historical baggage and cruft <drakonis>a long way before it is caught up with guix in the ways it is ahead <mehlon>I really can't switch to Guix as my primary system until they fix the slowness of Guile Scheme <mehlon>even with substitutes available it can take at least a few minutes to install a single package <KE0VVT>mehlon: I think it's slow because every package has its own version of everything. <mehlon>I don't think that's the cause of the problem since Nix has this too <lekzikon>The documentation of "guile-build-system" says it installs the scheme files in the right place, and also the documentation. Is "documentation" texinfo documentation or just the README? <lekzikon>I ask because I added texinfo documentation to the example library I packaged, guile-inutil, but "info guile-inutil" says there is no such thing. <oriansj>Open call to all scheme developers: I need help with getting tests for how this: https://github.com/oriansj/mes-m2 is not guile; so that I can make it behave exactly like guile and we can use it as a drop in replacement for guile and make history. <mehlon>there was a scheme to wasm compilery thing as well <mehlon>although guile bytecode may or may not be comparable to wasm in performance <oriansj>mehlon: honestly mes-m2 is expected to be dog slow, it is just a bootstrap scheme <mehlon>maybe we can get funding to build modern lisp machines to solve the problem <mehlon>make it real popular and hyped like bitcoin so NVidia comes out with their LPUs <oriansj>NieDzejkob: a bad idea wrapped in an expensive mistake <mehlon>NieDzejkob: it's when your computer has an LPU <mehlon>maybe we just have to wait a few decades until we can all program in Quantum Lisp <oriansj>mehlon: so not only can most people not understand the code, it also is probablistic in behavior... <mehlon>actually I think there is already a quantum assembly language that looks like lisp <oriansj>mehlon: actually no because it doesn't support reader macros <oriansj>but it is very s-expressiony and does support function macros <mehlon>this is one of those times I wish I actually knew how to program lisp <oriansj>mehlon: no one knows how to program in lisp; they only know an approximation of how to write lisp <mehlon>maybe everyone who programs in lisp only pretends they understand what they're writing and the compiler just does black magic that only makes them think the code works <oriansj>Its greatest strength is also its greatest weakness; it like mud can be molded to look like anything you want. <mehlon>I just realized that you can make a dialect of Scheme that also has something to do with the letter M and <oriansj>getting 2 scheme programmers on the same page is a miracle and when it happens; it is a world changing event *raghav-gururajan finally nailed the systemd back <lekzikon>I discovered something nice about epiphany. If you are editing an HTML document locally and load it in epiphany, the document is reloaded automatically in epiphany everytime you save. <lekzikon>This is useful when editing static websites. <lekzikon>You can have the editor on the left side of the screen, the browser to the right, and see the changes load as you work. <raghav-gururajan>Folks! If cmake-build-system throws this error "Error opening include file 'c-int-source-funcs.inc'". What is it mean? <leoprikler>nckx: regarding ibus on GNOME, I plan on having ibus installed by default with my super gnome package (read: gnome + extras), but the way ibus works is currently not really documented well in Guix <nckx>raghav-gururajan: I'm guessing this package is called libcoverart? It would mean that ./src is not in the include search path of the compiler. Adding -Isrc (that's an I as in Iovis) could help. We'd need more output to say more. <leoprikler>specifically the need to set GUIX_GTK{2,3}_IM_MODULE_FILE <nckx>raghav-gururajan: Ah, then you need to tell libmusicbrainz where to find that file, which I think is provided by libcoverart. <nckx>(This all based on GitHub Web pages.) <raghav-gururajan>I thought I need older version because sushi which depends on it givers error "Could not generate cargs for libmusicbrainz". <leoprikler>is there a way of handling this by default elsewhere, e.g. /etc/profile? <nckx>Something like (string-append "CFLAGS=-I" (assoc-ref %build-inputs libcoverart) "/include") assuming that .inc file is actually installed. Otherwise you need to use (package-source libmusicbrainz) "/src" or something. <nckx>leoprikler: Oh boy, you and your blasts from the pasts. I'd plain forgot about that hack. <leoprikler>Heh, sorry. I really think ibus should work out of the box however 😉️ <leoprikler>Though I'd also accept a separate ibus-service-type <nckx>raghav-gururajan: Well, you didn't give me more to work on than what I can duckduckgo. But OK, libmusicbrainz has that file too, it seems, as src/c-int-source-funcs.inc. Could you paste the full build output? <nckx>leoprikler: I agree. raghav-gururajan: I'm afraid I have to go now. <leoprikler>which file should i put this def into? gnome.scm? <raghav-gururajan>My confusion is why I get "Dependency libmusicbrainz5 found: NO" when I have that as input. <raghav-gururajan>I usually put everything in input first and then move to native based on guix lint suggestions. <leoprikler>yeah, gstreamer itself is required as far as i can see