<Apteryx>I'm trying to see how we're supposed to be able to use dhclient -4 and dhclient -6 at the same time. The problem is that they both overwrite /etc/resolv.conf <ajgrf>rekado_: i now think my problem might be from not generating a key. an older version of guix gave a more useful error message <ajgrf>running `guix archive --generate-key` doesn't put write anything to /etc/guix/ though, so i don't know how to fix that <ajgrf>hmm... it put it in /usr/local/etc/guix <Apteryx>ajgrf: If you used an older guix version in the past your key could be under /usr/local/guix or something like that <sirgazil>In The Future of Guix at FOSDEM, people were talking about "apt-get install guix"... <sirgazil>For what is worth, as a user of Guix on a foreign distro, I'd like to have "apt install guix", so that the congnitive load and time during the (un)installation process is reduced. Having this convinience could lower the barrier of adoption of Guix, and could also channel people to use GuixSD. <sirgazil>But with "apt install guix" I mean an easier way to (un)install (not necessarily Debian specific). <CharlieBrown>Parabola (and Arch) have packages in their repos for Guix. <Apteryx>How can I install the "dig" network analysis tool? <Apteryx>CharlieBrown: No. Apparently it's not packaged yet. <Apteryx>Ah, it must be part of the "bind" package. <Apteryx>doesn't seem to be part of the bind package we have... still digging ;) <mvb>it's in bind's "utils" output ***Piece_Maker is now known as Acou_Bass
<Apteryx>I'm confused by why our dhcp-client-service depends on the udev-service or user-processes-service. <Apteryx>How does Shepherd monitor processes? Does it look for that their pid is still up & running at some time interval? <clacke[m]>--> I'm trying to see how we're supposed to be able to use dhclient -4 and dhclient -6 at the same time. The problem is that they both overwrite /etc/resolv.conf <-- Apteryx <rekado>“As of 2017-02-20 we have fully reproducible builds on amd64 and sparc64” <rekado>I’d like to mark one day as a hack session to systematically fix reproducibility bugs in our packages. <efraim>sometimes I wonder if what's holding us back is the compile time <wingo>ACTION having problems building guix after a git pull; (gnu tests install) doesn't load and aborts the whole compile process <wingo>something about a waitpid that doesn't work <wingo>oddly if i comment out the contents of the file and then build, everything else builds; then if i uncomment the file and build, it builds too <wingo>now i am getting something where #f ends up being passed to cache! as the package argument :/ <wingo>ACTION starting to think that mark_weaver's "just use git pull" advice wasn't really for me :P <wingo>i update guix infrequently enough that things break :P <wingo>seems to be related to in-band-download <rekado>I have an ugly fix for it, but I’m not sure if we should commit it. <wingo>oddly url-fetch gets called twice in a row; one time "guile" is the right thing, one time it isn't <wingo>i think it might be bootstrap.scm:bootstrap-origin <wingo>which decorates normal "url-fetch" calls with an explicit #:guile argument <wingo>b/c #:guile isn't a usual kwarg for an origin-method <wingo>and bash/fixed uses url-fetch/reset-patch-level as its method <wingo>which has a guile keyword argument but which defaults to #f <wingo>i don't really understand grafts fwiw :) <alezost>Sleep_Walker: ahm, no problem, but thanks for what? <alezost>oh wow, I've just sent that message and you already checked it out! <alezost>that's not my wiki page, so thanks to rain1! <efraim>fontconfig FTBFS on core-updates <sneek>phant0mas, you have 1 message. <sneek>phant0mas, janneke says: yes get the same. tests/00-initial-env.test is fine, but adding tests/00-repl-server.test hangs for quite some time (1 minute?) and then prints lotsa In procedure fport_fill_input: Resource temporarily unavailable\\nscheme@(repl-server)> While reading expression:\\nERROR: In procedure fport_fill_input: Resource temporarily unavailable\\nscheme@(repl-server <phant0mas>janneke: we have the same error, different output <phant0mas>which version of gnumach and Hurd you are using? <wingo>holy recompilation, batman; this bash thing :P ***snape` is now known as snape
<slyfox>how does 'guix build --target=' populate enviromnet with host and taret compilers? <slyfox>'guix build re2c --target=alpha-unknown-linux-gnu' does the right thing as it uses proper ./configure detection <slyfox>'guix build re2 --target=alpha-unknown-linux-gnu' doesn't becauses it just uses g++ <jmd>Perhaps the inputs for that package are wrong? <jmd>Or perhaps the package uses a carelessly written Makefile <slyfox>no, it's a broken guix package definition <slyfox>in both cases there is no explicit inputs. both are '(build-system gnu-build-system)' <jmd>Then it's not a broken package definition. <slyfox>i would need to pass a CXX= into makefile but i don't understand why g++ is not already a ${CTARGET}-g++ <jmd>Normally gnu-build-system will do the right thing. <slyfox>while package definition treats it as such <slyfox>"(delete 'configure)" is quite a strong hint <jmd>In other words the package is broken, and the guix definition hasn't done enough to work around its brokenness. <slyfox>if your definition of broken is "not using autotools", then yes <wingo>ACTION pushed a guile-next update <jmd>"Not conforming to all conventional norms" <ng0>by your experience, where do fonts with .bfd go to? 'share/.../local' seems wrong <janneke>phant0mas: i only sent a small bit of the output... <janneke>phant0mas: GNU debian 0.9 GNU-Mach 1.8+git20170102-486/Hurd-0.9 i686-AT386 GNU <quigonjinn>how can one build a package with a different gcc? '--with-input=gcc=gcc@6' doesn't work <braunr>off topic question but somewhat relevant because you're probably affected too : do the autotools provide an easy way to detect compiler option availability ? <slyfox>you'll need to write your own test for it in m4 <thomasd>"IPython must be installed to build the Matplotlib docs" @!*?#$ <slyfox>as i understand there is no simple distinction between "good to have but cound be ignored" options and "need that to be present to work at all" case <ng0>get rich or stop typing ;) <thomasd>the scientific python people really love them some cycles <thomasd>used to be, numpy documentation depended on matplotlib (which depends on numpy) <thomasd>now, it seems matplotlib documentation also depends on ipython (which depends on numpy and matplotlib <ng0>with david leaving the project, is someone working on rust circular dependencies or is it now up to me to pick up the rust? <ng0>or maybe I got names mixed up :) <braunr>slyfox: do you happen to know why there is no native way to do it yet ? <wingo>ACTION grumbles about cert warnings <Amynka>slyfox: did you paste the guile bug to guile devs? <mbuf>ng0, who left which project? <jmd>ng0: Unfortunately not. <ng0>why, including the license file is no problem from my pov <jmd>That licence does not give permission to modify. <thomasd>(jmd: my comment yesterday about guix system reconfigure not honouring GUIX_PACKAGE_PATH was indeed a mistake. Classic mistake of sudo not passing environment variables...) <ng0>in another font, maybe I should pretend that the .bfd font files are "truetype" and copy them to ... truetype. local doesn't seem to be picked up <jmd>thomasd: Oh. I'm glad you found the problem. <jmd>ng0: It couldn't do any harm to write to the author and ask him to relicence the work. <ng0>2 out of 3 cases when I ask about features/changes they are happy to see systems have interest in their work :) <snape>ng0: I answered your email, although I forgot to cc you <ng0>hm.. moving fonts to truetype didn't fix it. is there some fontconfig command I need to run to refresh this? I know there is.. <ng0>fcache refresh or something <ng0>maybe we don't build support for this font extension.. which would be weird. rebooting ***jonsger1 is now known as jonsger
<sirgazil>This Weechat on foreign distro bug is marked as solved, but I still get the same error. <sirgazil>Should I report a new bug or is it possible to reopen it? <katco>gnu_guix: guix package -s <regex> <katco>gnu_guix: or, guix package -A <regex> <gnu_guix>Can I install Arabic and Persian fonts on Guix? <gnu_guix>Can I install Arabic and Persian fonts on Guix? <lfam>gnu_guix: It depends on whether anyone has packaged Arabic or Persian fonts yet. At least, we have GNU Unifont, which should include them. <efraim>which is one of the packages that we need the special code to recognize mips because the config tools they use are incredibly old? <efraim>libelf won't recognize aarch64, and i can't just toss in my isl patch <quiliro>guix/packages.scm:871:27: In procedure thunk: <quiliro>guix/packages.scm:871:27: In procedure struct_vtable: Wrong type argument in position 1 (expecting struct): #f <quiliro>should i report to the mailing list? <efraim>it sounds like your guix daemon is old <lfam>This bug has exposed a secondary bug: many users were never updating root's packages :( <efraim>orpheus had one I could copy, libelf now works for me <lfam>All of these users were vulnerable to a large number of serious security vulnerabilities <lfam>We need some sort of scary security announcement system <lfam>Even a mailing list that automatically forwards commits that include 'CVE' in the commit message <catonano>it seems like thhe tests attempt to open a display <catonano>and that isn't available, I suppose, in the build environment <catonano>should I write on the help mailing list ? <efraim>so far on core-updates I've found fontconfig and eudev FTBFS <paroneayea>would be nice to have a "How Debian and Guix can work together" talk at debconf <lfam>catonano: Try starting a virtual frame-buffer, as in commit fc119619c16931eb3f2ca4fffcc90d478ba1433e <lfam>efraim: Any ideas about how to fix them yet? <efraim>fontconfig is already newest, was about to check eudev <lfam>paroneayea: Indeed, "someone" should make it happen :) <lfam>efraim: Got the build logs handy? <lfam>I'm still building GCC locally <efraim>right now i'm focusing on finding packages that FTBFS on aarch64 so I don't end up with low level issues when its time to merge <lfam>It's okay, I'm reconfiguring my system here on x86_64 <quiliro>lfam: Even though I have GuixSD, I am following the steps on "For those of you who are using Guix on top of a foreign distribution" because I do not know how to apply the patch suggested. <lfam>I'll get to those packages eventually <lfam>quiliro: You don't need to apply the patch. I meant the instructions after the patch, starting with "# become root" <lfam>quiliro: Skip the instructions about the systemd service <efraim>eudev looks like an incompatibility with glibc-2.25 <efraim>i'll try bumping eudev from 3.2 to 3.2.1 <lfam>efraim: Might as well push it directly to core-updates, right? <lfam>quiliro: On GuixSD, these are the steps I recommend: <lfam>3) guix system reconfigure ... <lfam>4) herd restart guix-daemon <lfam>5) Let us know if it worked <thomasd>Hi Guix, git clone is complaining about SSL certs, did I miss something? :) I'm on GuixSD, have installed nss-certs $GIT_SSL_CAINFO set to its default value <efraim>no go on updating eudev. patch applied cleanly, failed to build <ng0>efraim: now what's left for GuixSD on aarch64, just having all tools for the SD build and proven to work? <quiliro>lfam: or guix system reconfigure ... <quiliro>lfam: with or without ... it would not work <lfam>quiliro: The '...' was a place-holder. You should reconfigure in the normal way, giving the path to your system configuration file and any other options you want to use <lfam>thomasd: What is the value of the variable? <quiliro>lfam: i do not remember what was the name ...i used the installation instructions by the book exactly though <lfam>quiliro: Is it found at /etc/config.scm? <quiliro>lfam: i think it was in /etc when it was mounted in /mnt <thomasd>/home/thomas/.guix-profile/etc/ssl/certs/ca-certificates.crt:/etc/ssl/certs/ca-certificates.crt <lfam>thomasd: It needs to have only one element instead of a list of elements. Try removing the second part <lfam>quiliro: Yes, see the manual section 'Invoking guix system' <lfam>quiliro: The specific case of what? <thomasd>lfam: thanks. weird though, wonder how that setting came to be <lfam>thomasd: I *think* there's a bug report for this. At least, it's come up before. All the Git "search paths" are actually not search paths, but must point to single directory or file. <lfam>Can you check and file a bug if there is not one? <ng0>efraim: was that last 'i hope' a reply to my question? <lfam>quiliro: The case of how to work around this bug? <lfam>quiliro: The bug is new and we are still deciding how to solve it. What I suggested for you is a temporary work-around until we find a real solution <lfam>The only difference between the instructions I gave you and the typical method of updating GuixSD is that you are `guix pull`-ing a specific commit. <quiliro>lfam: ok then, please tell me if i can help out with something <lfam>quiliro: Please try the instructions I gave and let me know if you still have the problem "guix/packages.scm:871:27: In procedure struct_vtable: Wrong type argument in position 1 (expecting struct): #f" <quiliro>lfam: i though guix reconfigure was never needed after installation <lfam>quiliro: `guix system reconfigure` is the way that you update GuixSD <quiliro>lfam: i am still running reconfigure <paroneayea>lfam: ugh I can't trick anyone else to do it? ;) <lfam>paroneayea: Well... I think we shouldn't make a big presentation about it if none of us are actually interested in doing the work :) <thomasd>lfam: I checked. It's bug 25422 (you filed it!), and there appears to be a patch circulating. (also, setting the variable to just /etc/ssl/certs/ca-certificates.crt works) <quiliro>lfam: so guix package --update will not update guixSD? <lfam>quiliro: `guix package --upgrade` only updates the user's packages, but it doesn't update the system <roptat>hi, I get "guix lint: warning: TLS certificate error: ERROR: X.509 certificate of 'static.nvd.nist.gov' could not be verified:", how do I fix that? <ng0>quiliro: reconfigure does <lfam>paroneayea: I got a reply from serveraptor. They are going to test the image this week and let me know how it goes. <lfam>paroneayea: I'm hoping we can make it work, because I'd love to ditch my other VPS providers and just use GuixSD <quiliro>lfam: so reconfigure should be run once in awhile? <ng0>reconfigfure updates kernel and all system stuff, yes <lfam>quiliro: Yes, you should run it as often as you like. If you don't ever run it, your operating system will contain security vulnerabilities <lfam>I'm going to re-read the section of the manual 'GNU Distribution' from beginning to end to see if we can make this more clear <quiliro>is there a package in guixsd for dental practice? <lfam>quiliro: Sure, let's both do it :) <ng0>no idea.. catonano and myself are somewhat working on GNU Health.. or at least that'S whati figured from my mails.. i domn't know how much health covers though <thomasd>is there a short way to simply checkout a git repository in the store, without building a full package around it? <quiliro>ng0: you mean gnu health could be used for dental practice? <ng0>idk.. i'm just trying to make its build system more useful <ng0>that didn't require runningf or reading docs other than code so far <lfam>That example could use git-fetch instead of url-fetch <lfam>thomasd: Is that the sort of thing you had in mind? <thomasd>probably it will get shot down when I submit the patch, but it's a start ;-) <lfam>At least you know how to do it now :) <thomasd>python people are little too creative in their packaging habits ;) <ng0>no.. the problem is just the interactive bash script <ng0>i had offlist conversation with the maintainer. nbow i just need to learn more hg usage and write the additions <roptat>hi, I get "guix lint: warning: TLS certificate error: ERROR: X.509 certificate of 'static.nvd.nist.gov' could not be verified:", how do I fix that? <thomasd>roptat: do you have nss-certs installed? <roptat>yes, for my user and in my system configuration <ng0>complain to the admins or try if what you wants works otherwise without tls <ng0>if it doesn't, find a mirror <roptat>hum... I'm complaining about guix lint ;) <lfam>roptat: You need gnutls-guile available. Try `guix environment --ad-hoc gnutls` <lfam>That assumes you've followed the instructions about setting SSL_CERT_DIR and SSL_CERT_FILE, found in the manual section X.509 Certificates <lfam>Urgh, we missed the latest version of GnuTLS, which brings support for internationalized domain names <efraim>i didn't check it, just updated and built it <catonano>quiliro: don't wait for GNUHealth on Guix. We were just bouncing ideas among ourselves <catonano>I have no iidea when and if it will ever be on Guix <efraim>now I need to figure out signing commits from the console and I can push commits from my aarch64 board too <lfam>efraim: Hm, it wants libidn2 for IDN support <roptat>ok, I didn't set those correctly it seems <ng0>catonano: i didn't quiet get you in the conversation though.. is tryton seaprate? falcon told me it isn't <ng0>if it isn't, I could help you out and pass you the previous conversation/info with falcon <catonano>Tryton is a framework that GNUHealt uses. <ng0>the source includes tryton though <roptat>Throw to key `gnutls-error' with args `(#<gnutls-error-enum Error while reading file.> set-certificate-credentials-x509-trust-file!)'. <catonano>Now the lack of separation comes in how it uses it <lfam>roptat: Did you make gnutls-guile available as I suggested? <catonano>exactly. They just copied the Tryton repo and added their own modules thhere <catonano>sort of what I did with amz3's graph db to populate it with jquery deps <ng0>where tryton contains 'everything' needed for Health <catonano>Tryton is teh only heath dep, as far as I understand <ng0>there are some python deps <catonano>ok, I have a question so naive that it boders on lameness <catonano>lfam gave me a commit id. How do I find such commit ? Do I have to checkout to that ? Or is there a way to see that speciic commit through the web or through any other mean ? <ng0>i'm considering either a makefile to call all other setup.py which are present in the subdirs already, or one setup.py to rule them all, i only did have discussions about it so far.. now I need to think :) <efraim>how do I check that my commits are signed? <efraim>oh, regarding the aarch64 jobs, I don't know that armhf or i686 can crossbuild for 64-bit architectures <lfam>efraim: You can pass any Git revision selector to `git verify-commit`. For example, HEAD, deadbeef, v0.12.0..HEAD <lfam>efraim: You can also do `git log --show-signature`, the lazy way ;) <lfam>They both will verify the signature, which is what you want <lfam>efraim: I was wrong about `git verify-commit`. It takes a single commit. But you can give it a list like this: `git verify-commit $(git rev-list $range)`, where $range is any Git revision selector <lfam>efraim: You should just use the pre-push signature verification hook :) <efraim>i normally sign everything now, i just didn't have it set up on the aarch64 board before :) <thomasd>what should I do with a bundled jquery (and possibly others) in the html docs of a package? (python-numpy) <thomasd>I see that it wasn't copied until now, but I'm not sure if this is on purpose. <lfam>We *should* delete it and replace it with a jquery package. Is that possible? I don't know <thomasd>it's probably possible, but I'll stick to just updating the version for now :) <thomasd>I'd rather fix the texinfo documentation then (which currently fails to build) <quiliro>uix/packages.scm:871:27: In procedure thunk: <quiliro>guix/packages.scm:871:27: In procedure struct_vtable: Wrong type argument in position 1 (expecting struct): #f <lfam>quiliro: Please walk me through exactly what you did <quiliro>excep when going to the user and typing guix pull <lfam>quiliro: So, steps 1 through 4 worked? <quiliro>guix system reconfigure /etc/config.scm <lfam>quiliro: Do you have the pgrep command? <catonano>_XSERVTransmkdir: ERROR: euid != 0,directory /tmp/.X11-unix will not be created. <lfam>quiliro: Okay, what is the result of this command (run it as root): `realpath /proc/$(pgrep guix-daemon | head -n1)/exe` <quiliro>/gnu/store/9nw4zglfl1dc8zc00rikkvln4kgjify9-guix-0.11.0/bin/guix-daemon <lfam>Hm, you're still using the old guix-daemon <lfam>I'm not completely sure, but I recommend doing this as root: `guix package -u . --fallback && guix system reconfigure /etc/config.scm && herd restart guix-daemon`, and then checking again <lfam>You could even limit it to `guix package -u guix --fallback && guix system reconfigure /etc/config.scm && herd restart guix-daemon` <lfam>You don't have to update all of root's packages to check this <lfam>catonano: It seems to fail because it expects to the root user (euid != 0) <lfam>catonano: I mean, it expects to be the root user <alezost>quiliro: was "guix system reconfigure" successful? look at the output of "realpath /var/guix/profiles/system/profile/bin/guix" <catonano>lfam: ok. Is thehre anything I can do about that ? <quiliro>alezost: i am running guix pull as root <quiliro>but on another terminal i ran the command you told me <quiliro>/gnu/store/c77gfhkmnwjvgmnsafjxf1g9kkpiilbh-guix-0.12.0-4.d9da/bin/guix <alezost>quiliro: that's the latest guix, so I think after reboot the right guix-daemon will be loaded. I'm not sure why something is being compiled for you since you already did "guix pull" several minutes ago <alezost>quiliro: if I were you, I would just reboot and look if "it works" <quiliro>alezost: i did guix pull as user not as root <quiliro>alezost: i am waiting the end of guix pull as root, then i will reboot <alezost>quiliro: you answered to lfam that you did it as root, and you surely did since the latest system has the latest guix <quiliro>should i reboot now or test as user first <catonano>ok, now I'm gonna write on the help mailing list about this <quiliro>alezost: i did all the other commands as root <alezost>quiliro: reboot if it's not a problem for you just in case <quiliro>guix pull has same problem as user, no problem as root <tmstmp>Hey, just installed guixSD and trying to have most of my packages in my configuration file. But for some reason some packages it won't recognize, I keep getting "... Unbound variable: emacs-flycheck" even though I have emacs in my use-package-modules and emacs-ag and emacs-neotree works fine. This also happens with guile and guile-emacs. <lfam>quiliro: Awesome! I'm glad it's working. I'm sorry about this bug... it's really a pain for all of us <quiliro>lfam: is there anything i can help withZ <lfam>quiliro: I don't think there is anything for you to help with at the moment. The person who fully understands the code that has the bug is on vacation, so the rest of us are learning and trying to find a good solution <lfam>tmstmp: Try using 'flycheck' instead of 'emacs-flycheck'. <quiliro>lfam: you are talking about civodul? ***snape` is now known as snape
<ng0>snape: what email? i didn't see any response <snape>ng0: yes because I forgot to CC you <ng0>no, i see every response in the threads, there was none to my rust question if that's what you mean <ng0>I will test this tomorrow, i've read the subsystem patch.. it's more than I thought of. thanks :) <snape>yeah, and the Match thing is tough to implement <snape>its a bit like the Virtualhost thing, in the Prosody service <ng0>if it works, i can submit my sshfs patch <snape>I hope you don't need Chroot and Match <ng0>well sshfs works, one option just needs the subsystem <tmstmp>lfam: Allright, that solves flycheck. don't understand how to solve the other ones though. Maybe it's a versioning problem? Seeing as I don't specify versions in my programs list... <lfam>tmstmp: That list of packages you are putting together is a list of variable names, rather than package names. <snape>ng0: the patch seems big, just because of indentation though. <lfam>tmstmp: So, take a look in 'gnu/packages/guile.scm' and find the Guile variable you want to use. I'm not sure about guile-emacs <tmstmp>lfam: Aha, awesome. thanks. A bit strange the names from define-public aren't put out there somehow though. <lfam>tmstmp: They are almost always the same as the package name, so it's usually not an issue. But I agree that it's annoying in cases like this. <lfam>There are very few cases where you need to know the variable name but you are not working on Guix itself <snape>quiliro: sudo herd start mysql <snape>but it should be already started if you follow the manual and reconfigure <quiliro>snape, lfam: it is not possible to run mysql as user? <lfam>quiliro: I don't know, sorry <quiliro>it would be great to be able to download the complete repository to be able to install totally offline <quiliro>even if it were by compiling, the best would be the substitutes because the usual case is old machines offline <snape>quiliro: by default, mysql is ran with user 'mysql' <lfam>quiliro: I agree that we should offer some kind of offline archive. But nobody has made it happen yet. Some Chinese users are thinking of mirroring everything in China because they can't access the rest of the world's internet. I think their work might give us some ideas about how to make an offline archive <snape>and if mysql is run by the service (shepherd), it is not "installed", by "installed" I mean it is not of in user/root $PATH <snape>and you can install it manually with 'guix package -i mysql' and then run it from command line <kevinfish>How long does the minimal graphics version of SD take to install anyway? I have an approx 1mb/SEC connection with a 2.3 GHz CPU and 4 GB RAM and it's been running for several hours now <lfam>kevinfish: What's it doing? <lfam>In that case, I guess you are limited by the 1 megabit download speed <lfam>I'm not sure, but I'd guess it would take less than 1 hour assuming you didn't have to build anything. But, I haven't installed a graphical system in a while, so I'm not sure <lfam>I'll try it on my 25 megabit / sec connect and tell you <kevinfish>I guess this thing could use an ncurses installer with a gas gauge <lfam>There is a work-in-progress interactive installer <lfam>Honestly, I don't recommend you install a graphical system. The system initialization is the most brittle part of GuixSD. I recommend users install the bare-bones system and then reconfigure into a graphical system <lfam>Of course, we should make the process more robust but, in the meantime, that's my recommendation <kevinfish>Probably the manual should say something like that is highly recommend <lfam>Maybe it wouldn't matter as much if you had a fast connection, but for 1 megabit per second, I think it could matter <lfam>There are many gigabytes to download to install a graphical system <kevinfish>I just want bare bones GUI and tiling WM plus my dev tools (emacs, GCC, scheme, etc...) <kevinfish>Another thing that would be nice is if there were an option on the web page for a pure HTML responsive version as my eyesight isn't what it once was so I have to zoom in and it's a PITA to have to scroll back and forth on every line <lfam>Agreed. Can you send a message to <bug-guix@gnu.org> describing this limitation and any details you might have about a possible solution? <lfam>kevinfish: You are using the lightweight-desktop template, right? <lfam>If you aren't planning to use xmonad, you should remove it from (packages). It pulls in a bunch of Haskell dependencies, which are huge. <lfam>You should also remove ratpoison, although its dependency graph is much smaller. <lfam>And since you mentioned i3, I recommend adding dmenu and i3-status. It's not really a complete i3 environment without those, IMO. <kevinfish>Ah, yes. I'd like to see 2wm be the included wm. It has the essence of xmonad and it's dinky and in C <lfam>I don't think anyone has packaged that yet. <lfam>Anyways, those templates are just examples. Users are expected to customize them <lfam>They are meant to illustrate how to use the configuration system <kevinfish>The page that hosts 2wm has several equally light weight utils <lfam>We'd certainly welcome a package for it! <kevinfish>Yeah, well IMO, it would be one thing to have some samples and another to get you up quickly with a workable system (e.g. not command line so you can actually get on the web and chats such as this to figure out how you want to enhance it) <lfam>I think that if you decide to install a graphical system without xmonad, it should not take more than a couple hours on your connection. <lfam>The interactive installer will serve in that role <lfam>There's no template we can offer that will not require customization, because users at least need to set up their user account and filesystem configuration <kevinfish>Yeah, I should have suspected, but I thought if came precomposed it might not be so fat <lfam>Also, our manual is in the installation image on tty2 <lfam>And you can get on the chat room with a terminal-based IRC client. For example, with `guix environment --ad-hoc ircii` <kevinfish>Yeah, also too small for me to easily read and idk how to change the screen geometry. There was no VGA option in 'e' from the grub splash <lfam>Sorry. Maybe your bug report should be more about general accessibility for people with low vision <kevinfish>If I get a chance I'll package up an ultralight GUI sample for inclusion on the distro <kevinfish>It comes up in like 80x24 but then halfway thru the boot switches to something much smaller <snape>I'm not sure what you mean by ultralight, but ratpoison is quite light, and written in C <lfam>I think we should just remove xmonad from the example <lfam>It will be hard to make the template any lighter after that <lfam>BTW, I decided not finish downloading the system installation dependencies. I don't have space on my disk at the moment <kevinfish>Yeah, sounds like at least anything more than the pure binaries of xmonad and lite are oxymorons <quiliro>ERROR 2002 (HY000): Can't connect to local MySQL server through socket '/run/mysqld/mysqld.sock' (2 "No such file or directory") <lfam>kevinish: I'm not sure what you mean. The binaries required for xmonad at least several hundred megabytes (I didn't let them finish downloading). That alone is too large IMO <snape>quiliro: do you use the GuixSD service? <quiliro>it seems running mysql is done with mysqld <kevinfish>Don't get me wrong, xmonad sounds cool but the average user will get almost as much out of 2wm in just a couple dozen k <quiliro>but i do not want to run mysql as root <snape>using the service doesn't mean you will run mysql as root <quiliro>and running mysqld as user gives errors <kevinfish>I thought Haskell binaries were supposed to be quite tight <snape>well, as said lfam, you should first have a look at the documentation, concerning services <snape>you edit your config.scm, add (mysql-service) in the 'services' field <snape>and then, you do 'guix system reconfigure config.scm', as root <snape>but that does not mean services are run as root <snape>yes, but the shepherd knows how to switch to the mysql user <snape>it is specified in the service code <snape>so for example, mysql service is in gnu/services/databases.scm <snape>but you are right, documentation could be improved <quiliro>but i cannot still understand well yet <snape>then, after the 'reconfigure' step, mysql should be started <snape>you can read it to understand better, but I agree documention ought to be enough <quiliro>(services (cons* (gnome-desktop-service) <quiliro>is it there where i should include mysql-service ? <quiliro>(services (cons* (gnome-desktop-service) <quiliro>guix system reconfigure /etc/config.scm --fallback <snape>--fallback should not be mandatory <quiliro>i do not understand why a user should ask root to run a service for him <snape>usually, two users cannot run the same service twice <snape>add (use-service-modules databases) on the top of your file <snape>because services listen on sockets <snape>and only one program can listen on a socket <snape>so it does not really make sense to have, say, two instances of a service running at the same time <quiliro>i should add that text before ;; This is an operating system configuration template <snape>you can just add 'databases' to the list of service modules <snape>and you still need (mysql-service) <quiliro>(use-service-modules (desktop) (databases)) <snape>(use-service-modules desktop databases) <quiliro>the previous text was (use-service-modules desktop) <snape>you just add databases to the list of service modules to load <snape>in that module is 'mysql-service' <snape>so the service manager needs to be root, because it has to start some services as root too <quiliro>snape: the users should not need permission from root to run a service as user <snape>but your user is not the mysql user, right? <snape>we are talking about running mysqld as 'mysql', not as 'you' <snape>each service is run with a different user <quiliro>snape: so mysql can only be run as user mysql? <snape>it's the best way to use mysql <snape>because then it doesn't have access to your home, say <snape>(although I just realized my home is 755, which is weird) <buenouan1>most distros"s default home permissions have always confused/bugged me <lfam>I mean, I think it's common for home directories to be created 755 <lfam>I agree that it's a relic and they should be 700 <buenouan1>/home should be 711 and all ~/ should be 700 <lfam>My Debian systems create home directores as 755 <lfam>I think it's a hold-over from when Unix systems where truly multi-user and everyone was friends and we hadn't started thinking about security in a serious way <mekeor>where can i find what you could call libc6-dev-i386? <rekado>mekeor: you would install a gcc-toolchain for system i686. <mekeor>ACTION tries `guix build --system=i686-linux gcc-toolchain`