IRC channel logs


back to list of logs

<buenouanq>as far as I know, I have no access to any efi bios or it"s features
<adfeno>I just found some guide, here, but I don't know if the software recommend here is free/libre,
<mekeor>anyone else experiencing frequent icecat crashes?
<adfeno>Me too.
<mekeor>is the reason known yet?
<adfeno>I would try disabling all extensions/add-ons.
<buenouanq>the fix is very simple
<cbaines>Have you seen this?
<cbaines>I think that helped the situation for me
<buenouanq>change and from cairo to skia
<buenouanq>icecat hasn't crashed on me since
<mekeor>cool, thanks :)
<adfeno>buenouanq: What does it do in contrast?
<buenouanq>crashes -> not crashes
<buenouanq>I don't even know what cairo or skia are - This was just the fix someone gave me a few months ago and it works.
<buenouanq>no idea why
<adfeno>Oh... OK :)
<lfam>They are 2d graphics libraries.
<buenouanq>why does cairo cause icecat to crash constantly, and why hasn't someone made skia the default in our repos?
<lfam>1) We don't know
<lfam>2) I looked into it but couldn't find a compile time option. Also see
<buenouanq>would should add a note about it in the package description or something then
<lfam>Search the mailing list archives; there's been a lot of discussion about thi
<lfam>I'm curious, do you notice the performance regression mentioned in that bug report?
<adfeno>Hm... I wouldn't enable GPU acceleration by default though (per Mozilla's bug tracker referece)
<adfeno>Supose someone has no GPU acceleration... how could they start IceCat for first time?
<lfam>I didn't suggest we do that
<adfeno>lfam: Not us/we, but the bug reference asks for it.
<adfeno>civodul and buenouanq: I found this:
<adfeno>It seems one might be able to make his own signed UEFI compliant copies of GRUB2.
<adfeno>efitools is available in Parabola repositories.
<adfeno>(not in Trisquel, currently)
<adfeno>Although it doesn seem odd that most guides I find online simply tell to use grub with special UEFI/EFI like parameters and they don't exactly mention how they make sure that GRUB2 is signed.
<adfeno>Because if I understand, UEFI, if implemented correctly, should allow the end-user to add his set of trusted keys to the first boot loader.
<adfeno>And in this case the first one would come prior to signed GRUB2, to check if this one wasn't tempered with.
<adfeno>But the guides I find rarely mention how to add such custom trust keys.
<buenouanq>I'm just not going to have any way to get to the efi boot or shell when this is done installing.
<buenouanq>So it'll all be installed properly, but I won't be able to do that last thing it needs.
<mekeor>in the `Networking Services' section of the guix manual, network-manager-service is listed but it doesn't exist (in the code)
<mekeor>there is network-manager-service-type but there is no network-manager-service
<myglc2>Hi Guix! Anyone know where the doc is for allowable kernel-arguments?
<kyamashita>myglc2: This is a good start:
<kyamashita>I reckon anything the Linux kernel accepts is valid for Guix's Linux-libre kernel.
<myglc2>kyamashita: Super! thank you.
<kyamashita>You're welcome. :)
<myglc2>going off to play with the kernel ... ;=)
<happy_gnu>wanna see a windos 10 user going crazy
<buenouanq>happy_gnu: they're using windows, they're already crazy
<happy_gnu>buenouanq: fair enough
<happy_gnu>still the video us really funny
<buenouanq>I just don't get even though that same thing is happening all over every single day to so many people why there aren't more people tuning in to the wonderful world of free software.
<happy_gnu>buenouanq: people is afraid of change
<happy_gnu>and people hate to learn
<lfam>It's very hard to change your habits, especially when they relate to tools you rely on
<buenouanq>but why then can someone like me upon learning about free software see that is it so obviously the only way to be and start moving over?
<kyamashita>I found the problem with packaging Free Pascal. Everything compiles statically, and I don't know how to fix it. :(
<buenouanq>`Wrong type argument in position 1 (expecting struct): "grub-efi"'
<buenouanq>what did I do wrong?
<buenouanq>should it not be in quotes?
<buenouanq>`grub (default: grub) The GRUB package to use.'
<buenouanq>I'm unsure how I'm supposed to change this to grub-efi.
<buenouanq>I did (grub "grub-efi") in the config, but that was obviously wrong.
<kyamashita>buenouanq: I don't think it needs quotes. Just checked out the documentation. I guess if the packages in (packages* ...) don't need quotes, neither should this.
<buenouanq>throws an unbound variable then
<kyamashita>ACTION apologizes for the possible gross violations on English grammar
<buenouanq>that was perfect english actually
<kyamashita>Have you included the grub package module, i.e. (use-package-modules ... grub ...)?
<buenouanq>ah, no I have not
<kyamashita>I guess I don't speak enough to know. :P
<buenouanq>I seriously never would have guess you were not a native speaker - Have more confidence.
<kyamashita>LOL, I am a native speaker. X-D
<buenouanq>wait what
<buenouanq>then why would you
<kyamashita>The inverse of the Dunning-Kruger effect.
<buenouanq>well, now I've started it building again - Does it really have to compile everything all over?
<kyamashita>That's the current procedure. I'm not sure if restarting the process mid-procedure is theoretically possible or not.
<lfam>It makes me happy when I read the security advisories from the big distros and find that we have already patched against most of them
<kyamashita>Yes, that makes me feel good about using my GuixSD machine. :-)
<lfam>ACTION builds the icedtea@1 update
<buenouanq>grub-install: error: /gnu/store/.../ doesn't exist. Please specify --target or --directory.
<efraim>sneek: later tell civodul on mips tar takes 70 minutes, and 90 on armhf, I'll try extending the Timeout. I assume those machines arent running from an sd card
<sneek>Will do.
<efraim>sneek: botsnack
<efraim>(properties '((timeout . 72000))) and `guix build foo --timeout=72000' don't do the same thing, needs to be `guix build foo --max-silent-time=72000'
<lfam>That sounds like an annoying thing to discover :/
<efraim>turns out tar needed about 65 minutes of silence instead of 60 for that test to pass, not sure that's really worth writing a patch
<cbaines>sneek, later tell mekeor if you want to use the network-manager service, you write something like this (service network-manager-service-type (network-manager-configuration))
<sneek>Will do.
<civodul>Hello Guix!
<sneek>Welcome back civodul, you have 1 message.
<sneek>civodul, efraim says: on mips tar takes 70 minutes, and 90 on armhf, I'll try extending the Timeout. I assume those machines arent running from an sd card
<efraim>took about 65 minutes in the end
<civodul>the test suite?
<efraim>it took ~72 minutes to build tar, xz and hello, and about 7 when i disabled the test suite on tar
<efraim>it kept on timing out on the one test, so that's where the majority of the time was spent
<civodul>i don't remember that it took so long
<efraim>i think on hydra on x86_64 the whole thing was 10 minutes
<rekado>kyamashita: how do you build Free Pascal without relying on their pre-built Pascal compiler?
<efraim>lfam: on your khal commit you wrote "add it" for dist_patch_DATA for removing the patch ;)
<efraim>is there a maximum size for mail to the mailinglists? I want to send in my patches for aarch64 but the bootstrap one is 2.4 Mb
<civodul>efraim: for the big one, it's probably better to upload it somewhere
<civodul>it's just binaries, right?
<efraim>the bootstrap binaries are the big part of it, I can make a fake version that replaces them with a text file pointing to the static tar/xz/bash/guile
<civodul>can be cross-build those bootstrap binaries from current core-updates?
<efraim>er, tar/xz/bash/mkdir
<efraim>probably, I haven't tried recently
<civodul>when we add the binaries, it'd be great to write in the commit log that they were built from commit XYZ
<efraim>i'll have to check which version of grep we have, the egrep/fgrep problem was fixed iirc
<efraim>... which makes me feel special that I updated grep to 2.26 and then to 2.27 and didn't check the extra phase
<efraim>i'll go ahead and rebuild the bootstrap binaries, and while i'm at it i'll see if make test passes on aarch64
<efraim>gcc has lib and lib64 in the search path, so I assume I don't need to force aarch64 to use lib, right?
<civodul>and we should add the aarch64 target for Hydra, like we did for PPC and GNU/Hurd
<civodul>efraim: few packages install to lib64/
<civodul>and really, none of them should IMO
<efraim>ok, so i'll leave my "force /lib" patch in for now
<efraim>as expected perl built fine, now i'm building to python and python-2
<braunr>what are the maint targets of guix ? i see you talking a lot about aarch64
<efraim>guixsd works on i686 and x64, guix also runs on armhf and mips64el, and i'm working on getting it running on aarch64
<braunr>i see
<efraim>wow systemd bug, creating files in /run/systemd/show-status with 07777
<buenouanq>expect nothing less
<civodul>braunr: there's also WIP for ppc32 and GNU/Hurd :-)
<civodul>there's a high demand for these ;-)
<braunr>i'm sure :)
<pareidolia>Guix is quite verbose
<civodul>pareidolia: yeah there's an open bug on this topic:
<civodul>i'd like to close it ASAP but "it's complicated"
<efraim>just checked grep on core-updates, its correct. my original bootstrap-tarballs were after the 2.25 update but before the egrep/fgrep fix
<cbaines>Does anyone know of any packages that spin up a database server to use as part of the builds?
<cbaines>e.g. to use when running the test suite
<civodul>hmm no
<civodul>presumably mysql and postgresql themselves do that
<cbaines>I've been thinking about how this could be done, do you think the services in Guix could be used?
<cbaines>So that you can add services to a package, and they will be started up at the beginning of the build process, and shutdown at the end
<civodul>no, within the package recipe you'd have to simply start mysqld or whatever
<civodul>several packages start Xvfb to test their graphical stuff, for instance
<civodul>that would be similar
<civodul>and the daemon takes care of terminating those processes at the end
<cbaines>It would be quite a nice interface if you could just use the existing service definitions
<civodul>yeah, that's right
<cbaines>I also want to try and write some packages that perform transformations on database dumps at some point
<cbaines>and having the framework to spin up and shut down databases as part of the package builder would be most of the work
<civodul>maybe what you need is the ability to run things in a one-off GuixSD instance?
<cbaines>I'm mostly thinking in terms of packages here, as its a good way of taking some inputs, and getting an output
<cbaines>Am I correct in guessing that the system tests for Guix are run as a derivation?
<civodul>right, everything is a derivation
<civodul>a derivation is what you say: a process that takes inputs and produces outputs
<civodul>packages are higher-level
<efraim> from python3 fails on aarch64
<divansantana>Quick Q. During guixsd install, I'm having issues with grub and filesystem configs. I want to boot up the install iso, change my config.scm and reconfigure system. How do I do this without redownloading everything?
<divansantana>If I mount all etc then run guix system init /mnt/etc/config.scm it redownloads all. When in fact all I changed is something in the filesystems declaration
***Piece_Maker is now known as Acou_Bass
***ShalokShalom_ is now known as ShalokShalom
<divansantana>I would think the other option would be to boot an install image, chroot into /mnt , then my paths are not set though. How do I correct that?
<adfeno>... Do you have actual access to the installed system partition? (did you mount it successfully)?
<divansantana>adfeno: yes
<divansantana>adfeno: /mnt/gnu/store exists.
<divansantana>how does one chroot in? I see there is no dev proc sys under /mnt should there be? normally one uses that to chroot
<divansantana>I think the issue is /mnt/run is empty, so I can't chroot in and run guix system reconfigure, because /mnt/run/current-system/... doesnt exist
<adfeno>I faced a similar issue in first attempt
<adfeno>I tried doing mount procedure but failled due to absense of these paths.
<adfeno>Although I'm new to these advanced usages, so don't take my experience foregranted.
<jlicht>hello guix
<adfeno>Hi jlicht :)
<thomasd>hi jlicht, and also hello guix, by the way :)
<adfeno>Reminder: does anyone know how to mount ab already installed GuixSD, assuming that one wants to start that copy to recover/redo something that went wrong? I and divansantana are puzzled with this challenge.
<adfeno>It seems we're somewhat lost because we don't know how to populate some important paths that Guix expects to find.
<adfeno>See logs of this channel for more details:
<rekado>adfeno: you can bind mount directories to the chroot location.
<adfeno>divansantana: ↑
<adfeno>rekado: But, in the case of GuixSD, which ones need to be bound, and from where to where?
<thomasd>Having broken my /boot (or at least grub) on an otherwise fine GuixSD installation, I'd like to know, too ;)
<adfeno>Oh! So we are three. ;)
<adfeno>Seeking enlightment :)
<efraim>We have enlightenment! In enlightenment.scm ;)
<jlicht> wow
<efraim>I would guess following the install instructions in terms of mounting and cow store might get you part of the way there
<thomasd>efraim: that's what I'll try tonight, but I wonder if it will detect that the whole system is actually already in /mnt/gnu/store store.
<thomasd>I thought that's where the bind mounts would come in (but have 0 experience)
<civodul>davexunit: what about having 'guix environment -C' & co. populate ~/.cache/guix/containers, which would list the PIDs of running containers along with metadata such as the container's profile (where applicable)
<civodul>that would make 'guix container exec' easier to use
<davexunit>civodul: ooh
<davexunit>that is a good idea
<davexunit>I like that a lot
<divansantana>doing the following in the chroot `/gnu/store/...system/bin/ln -s /gnu/store/...system /run/current-system` solves the path issue
<divansantana>but then I hit another error.
<divansantana>Which I don't have now. :)
<adfeno>divansantana: Oh... That's awesome! :)
<divansantana>another q. Howcome during the install, it seems to compile some packages? (I don't think it's compiling everything, just some.)
<adfeno>divansantana: Do you care to document your progress overtime?
<divansantana>adfeno: yeah, I'm typically quite good at documenting all etc. But for the last 3 months I've very spaz"
<divansantana>lol (proofing my point too earlier)
<divansantana>"spaz" typing, due to some rsi or carpal tunnel issues. So I'm typing less and without index fingers
<divansantana>both are a bit screwy at mo. And my thumbs a bit too.
<divansantana>/proofing/proving... err
<divansantana>adfeno: ill drop it to the ML if/when I figure it out.
<divansantana>not trying it at the mo tho
<thomasd>divansantana: compiling some packages during install is not unusual,
<jlicht>How should I approach the license field of a package if several files are gpl2, others gpl2+ and yet some other files "MIT" license. Should I list them all, including all respective files they apply to?
<thomasd>it depends on whether substitutes for your package are available from hydra. Depending on whether your local guix is very recent or very old, substitutes might not yet be available, or not available anymore
<jmi2k>jlicht: here you have an example:
<jlicht>jmi2k: Ah, we just add a comment explaining which licenses apply where. Thanks, that was all I needed to know :-)
<divansantana>thomasd: ah. ok.
<efraim>After removing the win32 test both python3 and python2 build
<efraim>After seeing if aarch64 also fails hard on gstreamer plugins I'll check out cmake
<efraim>Hopefully nss wont take 60 hours
<civodul>is anyone with an arm(32) box able to test this: ?
<civodul>this is a possible fix for <>
<GNUnymous_IRC>Are GuixSD packages newest? I want distro with newest version of packages?
<GNUnymous_IRC>I want free distro, of course.
<adfeno>GNUnymous_IRC: Well, we try to update recipes when we can.
<GNUnymous_IRC>Why some packages are outdated?
<GNUnymous_IRC>For example, ManaPlus client must be updated otherwise it cannot be used to play.
<GNUnymous_IRC>I meant play on many servers.
<ng0>you can contribute to updating :)
<civodul>TLDR: "Nix and Guix are as imperative as Debian"
<GNUnymous_IRC>Why some packages are outdated?
<GNUnymous_IRC>For example, ManaPlus client must be updated otherwise it cannot be used to play on many servers.
<GNUnymous_IRC>Why some packages are outdated?
<GNUnymous_IRC>For example, ManaPlus client must be updated otherwise it cannot be used to play on many servers.
<civodul>GNUnymous_IRC: please stop repeating
<civodul>it's a volunteer community where anyone, you included, can help update packages
<GNUnymous_IRC>Are you received messages?
<GNUnymous_IRC>Can I check is *something* GNU package using Guix?
<adfeno>GNUnymous_IRC: If you want to, we can update the package, but please give us sometimes.
<GNUnymous_IRC>This version is almost useless I think.
<GNUnymous_IRC>It is 1.6 but they are already updated it to 1.7.
<thomasd>about the reproducible-builds discussion: The word "pure" has a normative ring which is probably not helping :)
<civodul>thomasd: i agree, and i rarely use it for that reason :-)
<civodul>my message which was considered "propaganda" is here:
<nliadm>I still don't understand why people want to make a special kind of equality to talk about compiler outputs
<civodul>oh i did use "pure", but in a math sense
<civodul>yeah, not sure
<bavier`>doesn't read much like propaganda to me; more just a "hey since we're talking about this, just thought I'd mention since not everyone might know"
<thomasd>I thought the rant was more of a reaction to John Gilmore's message.
<GNUnymous_IRC>Can I use older package version package?
<GNUnymous_IRC>I cannot use SuperTuxKart 0.9.2 without bugs with textures bug 0.8.1 is working fine.
<thomasd>GNunymous_IRC: Guix as-is provides only a single version, but can create a package for an old version yourself quite easily.
<bavier`>GNUnymous_IRC: you can try building/installing using '--with-source=...' option pointing to the 0.8.1 source tarball
<thomasd>oh, didn't even know that :0)
<GNUnymous_IRC>Can Guix compile source code?
<rekado>civodul: after reading that email I scratched my head, paused for a minute, then shrugged.
<rekado>civodul: I also find the overreaction (rewrite compilers in pure languages...) to be ... revealing.
<nckx>So I was having a pretty good day until I clicked on that thread. :-(
<davexunit>civodul: wow just read that message you linked.
<nckx>Thanks for linking to it.
<davexunit>this is the same argument made by people all the time that don't understand functional programming
<davexunit>that it's the model, not the implementation, that makes something functional or not
<davexunit>at the end of the day we're mutating program counters and registers, so I guess nothing is functional!
<davexunit>does anyone know what the formal name for this fallacy is?
<davexunit>the same thing happens when people try to put down free software
<civodul>the world is not functional
<GNUnymous_IRC>Can Guix compile source code?
<davexunit>by saying that nothing is *truly* free
<civodul>it's even dysfunctional at times!
<rekado>GNUnymous_IRC: I don't know what the question means. Guix includes compilers.
<davexunit>you can see "I run a free OS" and they will retort "yeah but to use the internet you have to interact with proprietary switches, etc."
<clacke[m]>I just came back from a meetup where I presented Nix and Guix and I said just that. It was a functional programming group, so I said we all know this is all lies, we just need to choose at what level and to what degree we try to make them true. :-)
<GNUnymous_IRC>Can Guix compile package from source code?
<nckx>GNUnymous_IRC: stop it please.
<davexunit>GNUnymous_IRC: yes
<civodul>clacke[m]: indeed :-)
<clacke[m]>And then I pick up my phone and see this link :-)
<davexunit>so yeah, I'm seeing this general fallacy where someone keeps taking things down an abstraction level until they can make their point
<davexunit>it's "moving the goalposts" I guess.
<davexunit>a: "I run a free OS" b: "why even bother unless you have completely free hardware that you made yourself" a: "uhhh"
<nckx>davexunit: or a variant of the ‘no true Scotsman’, though neither sound 100% right.
<davexunit>surely there's a precise name!
<rekado>BTW similar objections are made when talking about a vegetarian diet.
<rekado>"you know, plants are also alive!"
<davexunit>rekado: another good example
<bavier`>"plants can feel! they cry when you harvest them!"
<thomasd>actually, there are communities which avoid many plants for similar reasons, e.g. Jain :)
<davexunit>reality isn't perfect, therefore you are wrong
<pareidolia>davexunit: I guess reductionism
<rekado>maybe it really is just a strawman; the goal is misrepresented as its most extreme version and then argued against using practical limitations.
<nckx>By that logic, isn't running advanced AI research also unethical?
<pareidolia>davexunit: My father likes to say "money rules in the end" when it comes to politics for example
<davexunit>yeah, reductionism or straw man sound like good fits
<pareidolia>rekado: Vegetarian?
<pareidolia>Someone is having trouble staying connected from Ukraine
<davexunit>can't tell if trolling or bad connection
<pareidolia>How can I patch KDE2 on FreeBSD?
<pareidolia>In general these discussions can be evaded by fixing clear practical goals, such as: 1) I want to avoid requiring dead mammals for nourishment. 2) I want to control which programs I run, and change them, easily and at-will.
<pareidolia>Chromium gives me headaches, since changing it requires a huge checkout and long compilation times.
<pareidolia>Is there a specific reason why neither Firefox nor Chromium are in the packages?
<rekado>pareidolia: yes, I don't eat animals.
<rekado>pareidolia: Firefox is available as icecat.
<davexunit>but mushrooms, watch your back.
<pareidolia>rekado: Salaton, samideano
<adfeno>Aw... these "moving the goals post" argumentations are a pain in the neck for me
<rekado>pareidolia: (but I don't speak Esperanto :))
<adfeno>I see this quite frequently with some people from Brazil who call themselves "free/libre software activists."
<civodul>rekado: yet! ;-)
<adfeno>... and yet, they recommend/share/sell/install/teach-usage of non-free distros. :(
<rekado>civodul: maybe!
<pareidolia>And the uncertainty wrt. to whether someone is being deliberately disingenous (for e.g. political reasons) or simply incapable/untrained in precise thought
<rekado>pareidolia: chromium bundles a lot of things IIRC, so packaging it properly is hard.
<CcxWrk>ACTION avoids eating organisms with centralized nervous system
<pareidolia>rekado: The Gentoo ebuild is 630 lines long :'(
<adfeno>cBesides Chromium has that freedomissue
<adfeno>Although It's not much as it used to be a year ago. Still, it's better to package Iridium instead.
<ng0>pareidolia: that's not even all of it. the ebuild needs the eclass, and the patches
<ng0>ACTION is currently balancing packaging palemoon, torbrowser, qupzilla
<ng0>funtimes with browsers x.x
<CcxWrk>Do you want start packaging Servo too? :]
<ng0>I would appreciate anyone who wants to take on what I am currently doing
<ng0>with notice of course :D
<ng0>for qupzilla, qtwebkit is the major thing
<pareidolia>adfeno: Other than the hotword debacle?
<adfeno>Speaking of QTWebEngine, it seems to be dependent on Chromium. See the folling discussions to know what's wron with Chromium and QTWebEngine:
<ng0>qtwebengine is parts of chromium for qt
<ng0>it's a separate project
<ng0>qupzilla is still okay. it's an engine, not a browser
<adfeno>ng0: Indeed, QupZilla seems to be OK for me too.
<adfeno>Also Iridium, as far as I 'm aware.
<adfeno>... is OK
<ng0>ok, thanks for the second link
<ng0>i will look into that
<adfeno>You're Welcome ;)
<ng0>but browsers are a long process.. I'm doing the full build first time and once I get to the functional state I start to unbundle.
***snape is now known as baley
<pareidolia>Is there a sort of policy document wrt. specific packages? Say, I'd undertake the translation of the Chromium nixexprs, but they wouldn't be accepted because of blobs etc.
<adfeno>I think there's a list in the LibrePlanet wiki
<adfeno>I'll give the link
<rekado>pareidolia: Guix follows the GNU FSDG.
<rekado>pareidolia: packages that violate the FSDG cannot be accepted.
<adfeno>And also this:
<adfeno>It's also good idea to look up in the issue/bug trackers of other free/libre distros.
<pareidolia>adfeno: I wish there was a Scheme version of that document
<pareidolia>I see some software in the list where only older versions are problematic
<pareidolia>It'd be nice to filter it
<adfeno>Perhaps one can use the Semantic MediaWiki Special:Ask page for that.
<adfeno>How would you like to receive the list?
<adfeno>In which format?
<pareidolia>It'd be nice to query with a guix plugin
<adfeno>Which formats guix plugins accept? CSV?
<adfeno>Let me check which exporters are available.
<pareidolia>Jumping to anchors seems broken here
<rekado>what are "guix plugins"?
<adfeno>Semanti MediaWiki outputs ask queries in: Feed, CSV, DSV (delimiter separeted values), JSON, RDF, table, ordered list, unordered list.
<adfeno>pareidolia: Indeed... I also noticed that, the anchor in the pages are indeed broken.
<pareidolia>I'm just musing here: it'd be interesting to have a "guix free <pkg>" command that a user can check when a package is missing
<pareidolia>But the list is actually not that long
<adfeno>We could perhaps query the Free Software Directory instead.
<adfeno>This way, we could have a more "pessimist" evaluation.
***baley is now known as snape
<ng0>so an lockscreen I am packaging tells me "Numlock?" (no real log entries anywhere, on screen display). I already wrapped libev in the path and adjusted the .pam file.. what more could a lockscreen need?
<ng0>the troublemaker is this:
<adfeno>pareidolia: I'm looking for a way to export all the approved entries in the FSD.
<adfeno>pareidolia: Found it: :D
<adfeno>You can replace "csv" by any output formats supported by semantic mediawwiki.
<adfeno>This list is, of course, only for approved entries in the FSD.
<adfeno>Although the names aren't "package-like" names.
<ng0>PATH is just for binaries? What's used for data like images etc? "lxmenu-data{/etc,/share}" must be available but using PATH to wrap it doesn't do it
<nliadm>probably XDG something or other?
<nliadm>that sounds right
<ng0>there are multiple XDG_ dirs thoigh
<ng0>asd-substitutions uses the data one
<ng0>and probably I don't want to wrap
<ng0>I want native-search paths
<nliadm>so the HOME ones aren't it, and neither is RUNTIME_DIR
<ng0>thanks :)
<ng0>no, search-path sounds wrong
<ng0>this is a package which needs another package, which can't be referenced because it is just data (images, etc)
<ng0>gedit wrap looks helpful
<kyamashita>Is there a clever way of getting a statically-linked libc.a using Guix that I'm not aware of?
<kyamashita>It's necessary for a Free Pascal Compiler package in Guix as far as I'm aware.
<jmd>kyamashita: In principle there is no reason why not. I dunno if guix's libc package provides one.
<kyamashita>jmd: It doesn't at the moment.
<thomasd>ah, my GuixSD is restored...
<adfeno>thomasd: How was it? :)
<thomasd>Kids, don't close the terminal in the middle of guix system reconfigure... (or at least that's what I think went wrong... shouldn't transactions have saved me)
<adfeno>thomasd: Use GNU Screen. :)
<thomasd>I mean: shouldn't transactions have saved me? Not clear if it was a bug or my error
<jmd>thomasd: What happened when you did it?
<thomasd>so what i did was add my swap partition /dev/sda3 to the configuration (all other parts unchanged)
<thomasd>then, i believe I accidentally closed the terminal before guix system reconfigure returned,
<thomasd>next boot, grub tells me "terminal.mod not found", and I'm in a grub rescue terminal,
<lfam>efraim: Bah
<rekado>kyamashita: Could you please tell me if the Free Pascal compiler you work on requires a binary bootstrap pascal compiler?
<lfam>thomasd: I think you found a bug; can you send a report to <>?
<thomasd>lfam: ok, but the above is literally all the information I have :/
<lfam>thomasd: That's okay
<thomasd>and it's rather annoying to reproduce on purpose :)
<lfam>Reports on IRC can get lost more easily than in the bug tracker
<lfam>Sounds easy to reproduce in a VM :)
<lfam>Still annoying, but less annoying than on bare metal
<thomasd>lfam: I'm bugreporting as we speak
<lfam>Let's pool our energy to make GuixSD more reliable :)
<adfeno>Thank you thomasd :)
<thomasd>totally unrelated: how come the nice emacs splash screen isn't shown in the default emacs package?
<kyamashita>rekado: Yes. The Free Pascal Compiler can only be built by itself AFAIK.
<jmd>kyamashita: That is unfortunate :(
<kyamashita>jmd: Indeed it is. I just wanted to package jxself's text adventure game ports.
<ng0>isn't there some ancient pascal which can be used to go all the way up until now?
<kyamashita>GNU Pascal is unmaintained, but I'm no compiler wizard (yet).
<kyamashita>Last version supported GCC 2.95, I believe.
<lfam>Does anyone know where Ruby packages come from? There are no releases on the Minitar GitHub repo:
<ng0>so TROVE was some kind of CVE..
<ng0>thanks :)
<lfam>I guess it's some other numbering system? I wasn't sure
<lfam>And sorry everyone for the Minitar question. I already figured out the answer
<lfam>ng0: I guess it's useful to include those other security bug numbering systems if there is no CVE.
<kyamashita>ng0: The first version of FPC appears to have been written in Turbo Pascal.
<ng0>I'm not good in pascal history.. I know the name, that's all
<ng0>we can not compile from that.. right?
<pareidolia>I'd be surprised if there was TP for Linux
<kyamashita>I don't think so. We would need a free Turbo Pascal compiler, and that's bascially Free Pascal.
<kyamashita>This is quite the pickle. :/
<rekado>kyamashita: I tried building GNU Pascal, but earlier versions of GCC are somewhat difficult to build.
<rekado>kyamashita: I wanted FPC for the game hedgewars.
<lfam>Did you look at what Debian does? Do they accept a binary bootstrap for their hedgewars package?
<kyamashita>lfam: They have to. That's the only way Free Pascal builds. GNU Pascal was dropped a while back because no one was maintaining it.
<kyamashita>Even my WIP package used pre-built binaries.
<lfam>Too bad. Hedgewars is my favorite game
<adfeno>Mine too...
<adfeno>I love to play with people at #lgn.
<kyamashita>Wait, are we giving up on the FPC package?
<adfeno>Oh boy... This is really bad...
<kyamashita>adfeno: ?
<adfeno>I wish we could sort this FPC package thing.
<adfeno>Let's if I can give some contribution.
<lfam>I didn't suggest we give up on anything
<lfam>I don't have an opinion on whether or not we should package FPC
<lfam>Err... it seems that our ruby-build-system doesn't allow us to apply patches to the package
<lfam>No, it's just this particular package, ruby-minitar. How strange
<kyamashita>Huh. So there *is* a libc.a file in the glibc package. So why isn't FPC finding it when I try to build it?
<rekado>I’m currently too busy to continue working on packaging GNU Pascal, but I could share my progress.
<kyamashita>rekado: I'm not sure I could handle that level of software wizardry... What does your progress look like?
<kyamashita>P.S. Can't work at my machines right now as my home internet connection is down.
<adfeno>Share by email instead :)
<adfeno>Post to mailing list
<efraim>building bootstrap binaries for aarch64 gives the recursive install error
<ng0>I love building undocumented stuff in unknown languages which throws just this for error in test-suite: text-extents failed: pixel mismatch 0x0 not similar to 0xffffffff at 11 30
<ng0>and the only person to blame is myself if it will not build. fine :)
<ng0>because I've build the language before
<adfeno>pareidolia: Also, about packaging: I would advice against packaging Maze of Galious remakes (for example the one from Brain Games), see:
<adfeno>Although the bug report seems inactive, and the issue seems uncofirmed, I still sense it has importance for now.
<ng0>lfam: I'm reconfiguring another system with my new tor configure-flags right meow, can take a while until the patch will be send :)
<ng0>it's available in my branch tor/hardening
<ZombieChicken>ng0: What are you changing wrt Tor's config?
<ng0>not the config.
<ng0>configure flags
<ng0>different kind of thing
<ng0>ACTION afk
<ZombieChicken>like what it does and doesn't support? Or are we talking things like --funroll-loops?
<efraim>I always always read that as 'fun roll loops' and I cant figure out what kind of flag that could possibly be
<lfam>ZombieChicken: See:
<lfam>Yikes, good thing we don't build anything with setuid:
<ng0>By the wy I've build pcmanfm with gvfs and I don't get the comment
<ng0>why would gvfs chaneg anything?
<ng0>it did have exactly 0 effect
<ng0>it probably requires some pam things
<ng0>oh, for grsec parabola provides a good starting point for th libre part: though the gentoo hardened kernel also comes with the patchset which creates linux-libre, so both are good starts if anyone wants to take on this mighty task :)
<rekado>ng0: “I don't get the comment” <– what comment?
<ng0>oops, sorry context
<ng0>one moment
<rekado>ACTION –> zzZ
<ng0>in (pcmanfm):
<ng0>(inputs `(("gtk+" ,gtk+-2)
<ng0> ;; TODO: add ("gvfs" ,gvfs).
<ng0>i know it's about auto mounting, but adding gvfs alone solves nothing
<ng0>if I only could read the error message which appears for .0001 ms on screen when I try to start lxde
<ng0>i know my graphics card will soon no longer work, so its either failing card or something missing
<ng0>I know where I was stuck with uclibc-ng, at least one minor part: how do we name tripplets?
<ng0>$thing-$arch-$system ?
<ng0>ah, I got past that: -DUCLIBC_RUNTIME_PREFIX=\\"/usr/x86_64-linux-uclibc/\\"
<ng0>just symbol errors now