IRC channel logs

<notadrop>I cannot boot the GuixSD USB install image at all. GRUB tells me the FS is unrecognised.

<notadrop>Can anyone help with this?

<notadrop>If you have any ideas and I'm not online, send me a Memo. Thanks.

<davexunit>you guys need to add Guile to the environment

<davexunit>if you don't add Guile then you will never have the right load path

<davexunit>see native-search-paths in the Guile package

<davexunit>Guix environment gathers up all the search path definitions from the packages specified and sets environment variables based on that

<davexunit>so just specifying an ad hoc environment with a Guile library isn't sufficient to set the Guile env vars

<kristofer>davexunit, artanis-0.1.2 is located in /gnu/store/rg5x61x2v0k6nz25ck7vh1bnxpam0dmq-artanis-0.1.2/gnu/store/rg5x61x2v0k6nz25ck7vh1bnxpam0dmq-artanis-0.1.2 on my system

<kristofer>it's nested twice, not sure how that happens

<lfam>Gah, look at all these patches for libxml2: https://git.gnome.org/browse/libxml2/log/

<lfam>I wonder if 2.9.4 will work when grafted...

<lfam>What can I build that can serve as a good test of whether the interface is the same?

<lfam>kristofer: Could you try building artanis after removing the line that sets DESTDIR?

<lfam>Or fiddling with the 'substitute-root-dir' phase?

<kristofer>lfam, looking at it I wondered if the strings were appended weirdly accidentally

<lfam>Yeah, something like that. I don't see any obvious mistakes but I only read it once

<lfam>The really weird thing is that it works for some people

<Emacsite>notadrop: What's a "Memo"?

<kristofer>artanis 0.1.0 works for me :)

<Emacsite>kristofer: no. you shouldnt exist. :o T_T

<kristofer>time to diff

<kristofer>Emacsite, are you looking for a web framework in guile?

<Emacsite>kristofer: duh

<kristofer>Emacsite, after some tinkering with artanis and guix-web, I've started building my toy web app on guix-web

<kristofer>although I really like how artanis dispatches requests to handers with the regex

<lfam>Does anyone know some software that uses XML for its configuration file? That would be an easy way to test the libxml2 update...

<Emacsite>lfam: No XML, SXML!

<kristofer>guix-web isn't a framework, it's just a web frontend for the guix package manager, but it's seriously very easy to build off of

<lfam>Emacsite: You have some program that uses SXML? What is it? Maybe it uses libxml2 under the hood

<Emacsite>kristofer: no, i want to play with artanis

<Emacsite>guix has artanis in the repos and i want to use it

<kristofer>I understand. lfam is on the right track I believe, with adjusting the package definition to fix the path, but the bug isn't jumping out at me. davexunit suggests including guile in the environment to fix the load paths; maybe I'm doing it wrong, but I think it's a problem with the package. gotta get some sleep, long day tmrw. good luck :)

<lfam>To clarify, I'm not on any track :) That is, I'm not working on fixing that package right now. Help wanted!

<rekado_>what surprised me at the Guix workshop was that people in the audience brought up the issue of exporting package closures that include non-free software.

<rekado_>so I got a very good opportunity to make clear that Guix itself only contains free software, but that some of the software that we were asked to add has restrictions on redistribution and/or use.

<rekado_>I advised them to do something about this and stop using non-free software and contact the vendors about this issue.

<rekado_>do we have an automatic way to convert a profile's contents into a manifest?

<rekado_>because so many users here need Python stuff and propagated-inputs make it very easy to accidentally create conflicts (simply by installing one thing, updating Guix, installing another thing that happens to pull in a subset of the dependencies of the first thing), I told them to use manifests instead.

<rekado_>using variable names isn't very convenient as they don't seem to be shown in the output of "guix package --show".

<rekado_>(I seem to remember that there was a procedure to turn package names into variables that we could use in a manifest file)

<efraim>awk?

<efraim>guix package -I doesn't print which file it's in so awk wouldn't help as much

<alezost>rekado_: how could it be even possible to turn a package into a variable? For example, consider (define v1 "foo") and (define v2 v1). To what variable "foo" should refer to? Moreover, how can you define a variable by its value?

<alezost>So AFAIU it's impossible to convert a profile's contents into a manifest

<civodul>alezost: a profile contains a manifest, though

<civodul>like ~/.guix-profile/manifest

<alezost>civodul: I meant a manifest of package variables taken by "guix package --manifest"

<alezost>i.e., a manifest created with "packages->manifest" procedure

<alezost>I think it's impossible to use ~/.guix-profile/manifest for "guix package --manifest"

<civodul>ah yes, we cannot convert an instantiated profile into source (where "source" is the thing that one passes to --manifest)

<civodul>but we could/should have an option to allow:

<civodul>guix package -X ~someone/.guix-profile/manifest

<civodul>guix package -Y /gnu/store/...-profile.drv

<civodul>so people could simply carry over these things

<civodul>also: guix package -W /gnu/store/...-profile

<civodul>all these are quite easy to implement

<efraim>mpc123 is missing upstream

<efraim>they took down the sourceforge files and didn't do a good copy over to github

<efraim>oh debian's orig.tar.gz actually has the same hash

<civodul>ACTION tries: "guix build -S mpc123 --no-substitutes"

<civodul>hopefully tarballs.nixos.org has it?

<efraim>uri->string on nix tarballs backtraced for me

<efraim>now and then I check https://www.gnu.org/software/guix/packages/issues.html for TODOs

<efraim>there's not much on the github page, but their sourceforge page says that's the new home so i'm going to update that too

<rekado_>alezost, civodul: what I remembered was this message: http://lists.gnu.org/archive/html/guix-devel/2015-05/msg00244.html

<civodul>efraim: yeah i just noticed the backtrace

<civodul>also the long timeout when establishing connection to broken SF mirrors

<civodul>ACTION prepares a fix

<alezost>rekado_: as civodul said several packages can match a given name, so if it's ok for you, then... OK :-)

<civodul>efraim: fixed :-)

<efraim>yay!

<emyles>Has just started again with guix and 'guix pull' stalls (overnight) at 'building packages'.

<emyles>What could the problem be?

<emyles>ps shows these processes:

<emyles>guix-daemon --build-users-group=guixbuild

<emyles>guile --no-auto-compile /gnu/store/blahblah-guix-0.10.0/bin/.guix-real pull --verbose

<emyles>guile --no-auto-compile /gnu/store/blahblah-guix-0.10.0/bin/.guix-real substitute --query

<emyles>and two other guix-daemons

<civodul>emyles: currently running "guix pull" and it works fine

<civodul>could you paste the full log?

<emyles>civodul: thanks for having a look

<emyles>civodul: the terminal output is here: http://paste.lisp.org/+6SIU I'll have a look for the log (should be in systemd's journal?)

<civodul>emyles: nothing happens after the last line?

<emyles>civodul: yep, nothing, no excessive cpu activity or anything

<civodul>can you try with --substitute-urls=https://mirror.hydra.gnu.org ?

<civodul>in case it's due to hydra.gnu.org not responding quickly enough

<wingo>ACTION wants to package docker

<wingo>i guess there is some kind of service that would need packaging

<civodul>wingo: the main difficulty is to finish packaging Go

<civodul>there are partial patches on the ML

<emyles>civodul: yes! that is building and loading things now, thank you

<civodul>cool!

<wingo>ah i see

<wingo>maybe it is not a morning project then

<civodul>esp. at 11:43 ;-)

<civodul>but i don't know exactly what's going on with Go

<wingo>i am in spain right now, the morning goes all the way to 14h ;)

<wingo>it sounds like it's just related to how go finds libgcc_s.so.1

<wingo>so it should be not bad to fix, dunno

<civodul>yeah

<civodul>we hardcode a -L... in gcc for that

<davexunit>wingo: the other problem with Docker (and all Go projects AFAICT) is that they bundle *everything*

<davexunit>so we need some kind of solution there that will mitigate that particular security nightmare

<ifur>i thought most people had moved to rocket by now?

<ifur>not sure if attracting current docker users is worthwhile actually.

<ifur>or is that still in development?

<ifur>anyhow, getting the coreos people on-board sounds more tempting :)

<ifur> https://github.com/coreos/rkt

<davexunit>ifur: rkt has the same problem: it's 80% Go code.

<ifur>ah :/

<davexunit>I don't know when Go became *the* hip language to write ops stuff in, but I very much don't like it.

<ifur>go is shorthand for get off my lawn, right? hehe :P

<piyo>Does "guix build emacs --with-source=.local/emacs-25.0.94.tar.gz" really work? http://emacs.stackexchange.com/a/16932 What is the next step I need to do to install it?

<davexunit>piyo: guix package -i emacs --with-source=.local/emacs-25.0.94.tar.gz

<piyo>! gonna try that now.

<davexunit>I've never used this feature, but that's what the docs say

<rekado_>piyo: if you want to enable certain features (such as embedding widgets) you'd have to write a package variant.

<piyo>rekado: thanks for that answer post

<davexunit>ACTION tweaks emacs build system to handle packages that are a single .el file

<piyo>ok the package command finally finished and emacs --version is the expected version, sweet!

<adfeno>Great! :D

<civodul>neat, i'm glad to see that feature put to good use :-)

<piyo>guix package -I "emacs 24.5 out /gnu/store/crkqj15j5c7cw35a0063iirjj3vrw4kw-emacs-25.0.94"

<civodul>piyo: oh, does it get the version number wrong?

<piyo>I don't think its a bug.

<civodul>normally "guix package -I emacs" should show emacs 25.0.94

<piyo>I assume I am reusing the 24.5 recipe

<civodul>yes, but still

<efraim>i used --with-source=http://software/next-version.tar.gz on a machine that I didn't have the git repo on but I don't remember what version it said when it finished

<efraim>when we use grafting it shows the wrong version number, i assume this is similar, it uses the emacs-24.5 package but with a different source

<rekado_>does anyone here have some thoughts about the mechanism I used to implement the recursive CRAN importer?

<rekado_>is it okay to abuse "(values ...)" like that?

<davexunit>rekado_: I need to take a look at it

<davexunit>so, it looks I can make the emacs-build-system do the right thing with sources that are a single .el file (not in a tarball)

<davexunit>but

<davexunit>it involves processing the store file name such that something like "xxx-better-defaults-0.1.3.el" is copied to the source directory as "better-defaults.el"

<davexunit>strip-store-file-name can easily get rid of the hash, but getting rid of the version number is problematic because that part of the store file name is arbitrary.

<civodul>piyo: ok, i see how to fix it :-)

<civodul>rekado_: i've been meaning to look into it; soon!

<piyo>civodul: please retain the behavior because it breaks my Workflow ;-) https://xkcd.com/1172/

<efraim>i love that one

<civodul>heheh :-)

<civodul>hello lxo, nice to see you here!

<piyo>the latest one is pretty good and relevant to this channel, maybe. http://xkcd.com/1685/

<efraim>every time I look at vigra to try to fix it I read "viagra" instead of "vigra"

<efraim>oh, x.org is mostly "hidden" behind https mirrors, and the perl "all versions of everything" mirror is also https

<adfeno>Hi lxo/ Oi lxo. :D

<efraim>so thats why most of them can't download the source

<rekado_>civodul, davexunit: no hurry. I just wondered if the silence means that you're preparing a long critique of my poor style ;)

<davexunit>rekado_: heh, nope. just haven't read it.

<civodul>rekado_: :-)

<davexunit>once we agree on implementation, I'd like to mimick it for ruby

<rekado_>cool cool

<bavier>and me for cpan/hackage

<davexunit>wee, looks like I've got the emacs build system doing what I wnat

<davexunit>want*

<lxo>civodul, adfeno, hello there!

<adfeno>lxo: Nice to see you here in #guix. :D

<lxo>(sorry, I'm a bit slow here; I woke up to find my primary gateway down; ceph cluster is now recovering)

<lxo>:-)

<lxo>thanks

<adfeno>lxo: Well... I'm also very slow when doing daily tasks and personal computing. My real strength is giving ideas and debating on matters regarding the free software philosophy.

<adfeno>And I am a really... really slow coder. :D

<adfeno>lxo: If I recall correctly... Today is Free Software Directory Meeting...

<lxo>I wouldn't know

<adfeno>... Perhaps you have some suggestions for the directory there... I currently have a prposal there...

<adfeno>lxo: Perhaps we can submit Makadu's apparently-non-free smartphone application for review....

<lxo>don't you already know it's non-free?

<adfeno>(The application used during FISL16. I think you have mentioned it briefely during a talk)

<lxo>oh

<lxo>that one

<lxo>I was later told it was Free Software, after all. though nobody knew where the source code was

<adfeno>I said "apparently-non-free" because I don't know if it was submited for review there (in the Free Software Directory).

<efraim>definate mistake searching for "scrot" online without the family-filter on

<efraim>looks like linuxbrit.co.uk is down

<bavier>ergh, just saw the patch submitted for clang-3.8

<bavier>I've been working on that for the last few days

<bavier>I don't think the patch even works, at least not given what I've seen. The clang-libc-search patch doesn't apply to clang-3.8

<efraim>i was wondering why not just move clang to 3.8

<efraim>i think i tried 3.7 but the linking at the end used all my ram and swap before failing spectacularly

<bavier>efraim: yeah, llvm et all just need to move to 3.8

<bavier>I was planning to leave 3.5 and 3.6 and add 3.7

<bavier>3.7 seemed to build fine for me

<efraim>i found where rdup is hiding now

<adfeno>ACTION trembles everytime he sees or hears "Clang" or "LLVM"...

<efraim>how much ram do you have?

<efraim>also I may have been doing it wrong

<bavier>efraim: this machine has 8G RAM

<bavier>I wasn't paying attention enough during the build to get an idea of the high-water mark

<efraim>same, so I must've been doing something wrong when I tried

<efraim>if anyone wants to take rdup, the new homepage is https://github.com/miekg/rdup

<bavier>efraim: I can take rdup; I added it in the first place

<lfam>civodul: Should I revert my changes on core-updates-next and do them on core-updates?

<efraim>lfam: don't think you'd need to revert them if they're applied to both

<efraim>oh wait, I don't know what the question references

<lfam>efraim: You're probably right.

<lfam>I made changes on core-updates-next because I thought core-updates was frozen, but those changes would be nice to have on core-updates, and it's not frozen after all

<lfam>An "ungraft" and an update to libxslt

<efraim>apparently gd has more CVEs

<efraim> https://lists.debian.org/debian-security-announce/2016/msg00164.html

<lfam>efraim: I can look into it unless you've already started

<efraim>i havent started

<lfam>I don't know how to interpret this line of the DSA: For the unstable distribution (sid), these problems have been fixed in

<lfam>version 2.2.1-1 or earlier.

<lfam>Or earlier? Is that a typo?

<efraim>probably a typo

<lfam>BTW, if anyone ever complains about how stringent we are about commit messages, send them here: http://git.imagemagick.org/repos/ImageMagick/commits/ImageMagick-6

<jlicht>lfam: that seems uninformative indeed :-)

<jlicht>does the recent policy change regarding signed commits affect contributors that want to create patches, or does it only affect people making the actual commits to the repo?

<lfam>jlicht: We are signing commits rather than patches, so it only applies to those who push the commit to Savannah. But, everyone else should feel free to sign the mail that the patch is in

<lfam>To clarify, we are signing the Git commits themselves, not the commit messages

<jlicht>lfam: I might need to set up some signing keys then, as there is no harm in signing mails that patches I guess. Is GuixSD as of now usable for generating keys? (referring to the /dev/random discussion on the ML)

<lfam>jlicht: That is still WIP (if you can help me, please do!)

<lfam>I don't know how gnupg generates keys (random vs urandom, etc). If you are paranoid, use a more established distro

<lfam>to generate the keys, that is :)

<lfam>And, why not verify their urandom seeding mechanism while you're at it? ;)

<kevinaucitron>hi there : )

<jlicht>lfam: Not really the right person for the job, as I only know what the fsf email self defense page taught me ;).

<jlicht>Once I have some more progress and clarity on npm-land, I might spend some evenings on settings things up \\o/

<lfam>Hey, everyone starts somewhere!

<lfam>Hi kevinaucitron!

<jlicht>Hi there kevinaucitron

<kevinaucitron>i will probably do another try on guix next week. i don't like to fail using a distro' X)

<lfam>You tried to install GuixSD and it didn't work?

<kevinaucitron>lfam, yep

<lfam>What happened?

<kevinaucitron>i can't remember : /

<lfam>Okay, hard to help you then ;)

<kevinaucitron>but we helped me to install it and i works, so i know that i can have help here : )

<kevinaucitron>i think it buggued at boot time.

<lfam>efraim: For libgd, do you think we should keep source links to both BitBucket and GitHub (see the comment in the gd package definition)

<lfam>kevinaucitron: We're happy to help!

<unknownbuckethea>Hello, is there any plan to release Geary in packages?

<davexunit>unknownbuckethea: if someone were to send us a patch for it

<unknownbuckethea>davexunit: hmm, thanks.

<davexunit>contributors package that software that interests them. so, it's up to someone who wants to use Geary to package it.

<davexunit>package software*

<avoine>just to be sure if I run sudo ./pre-inst-env guix-daemon --build-users-group=guixbuild and build something it will end up in /gnu/store/ but not in my .guix-profile right?

<unknownbuckethea>I understood, I'm not an expert for kind of thing. Then I should try by myself.

<davexunit>avoine: I don't understand the question

<davexunit>that just starts the daemon

<davexunit>~/.guix-profile is just a symlink to a place in the store

<jlicht>avoine: if you then ./pre-inst-env guix build, that would only build packages yes (and put them in the store)

<avoine>davexunit: I wan't to test my local patches to the guix source code without modifying my user profile

<avoine>and I did guix environment guix before that

<davexunit>avoine: you don't need to start the daemon

<davexunit>ACTION is confused

<avoine>but I get an error: guix build: error: failed to connect to `/usr/local/var/guix/daemon-socket/socket':

<lfam>You need to do `./configure --localstatedir=/var` when building Guix from your Git checkout

<lfam>Otherwise it defaults to /usr/local, and that's not the right place if you've installed Guix "system-wide" as described in the manual under 2.1 Binary Installation

<lfam>So, try reconfiguring your Git checkout of Guix and then running `make` again

<avoine>ok

<lfam>I want to know if a particular Git commit was included in a particular tag. Does anyone know an easy way to do that?

<jlicht>lfam: you could checkout the tag, and then log for the commit-sha in the git log

<jlicht>*first log = check

<lfam>jlicht: That's exactly what I just did :) There must be some Git tool designed for this, however

<lfam>Interestingly, when we package the latest version of libgd, it doesn't seem to run the test case for the most recently fixed CVE bug

<lfam>I wonder why that is

<lfam>Maybe I'm reading the log wrong. It doesn't mention any skipped or failed tests in the summary

<lfam>FYI there's probably going to be some updates to GraphicsMagick soon: http://seclists.org/oss-sec/2016/q2/422

<lfam> http://lwn.net/Articles/688401/

<lfam>"An attacker with control on the image input could, with the privileges of the user running the application, execute code"

<lfam>efraim: gd is updated to the latest

<jlicht>just found out that guile 2.0.11 has a bug in the (web uri) module regarding the parsing of fragments in the string->uri function. It seems this is fixed in guile 2.1, but I'd actually really like to use this in guix

<jlicht>Would it be an issue to copy the (fixed) function defintion, with a comment stating that this is a temporary workaround?

<civodul>lfam: i think you can just cherry-pick them

<lfam>civodul: Okay, I will do that

<civodul>ok!

<civodul>BTW, there's a lot of 2013-2014 CVEs if you run 'guix lint -c cve' now :-)

<civodul>ACTION has to go

<civodul>later!

<lfam>I noticed you updated the checker to look back several years...

<lfam>bb!

<davexunit>has anyone ever run a build farm for another distro?

<davexunit>interested in how to run something like what debian has

<davexunit>guix makes things too easy...

<bavier>efraim: it might have been disk space you were running out of building llvm/clang

<lfam>Can anybody apply this patch? http://lists.gnu.org/archive/html/guix-devel/2016-05/msg00765.html

<lfam>Even when I trim the email down, I still can't make it apply to maths.scm

<bavier>lfam: not related to the patch applying, but note that it doesn't add the new patch to dist_patch_DATA and the commit log should be adjusted

<lfam>True, I hadn't even read the commit log yet

<bavier>I'm not convinced that the second half of the hdf4-config-settings.patch is necessary

<lfam>Okay, can you reply to the list?

<bavier>sure

<bavier>lfam: got it to apply. seems like the email client mangled the patch a bit

<lfam>My client, presumably ;)

<bavier>I pulled the patch off the archive, so that might have been an issue too :)

<kyamashita>2016-05-27 11:23 *** kyamashita JOIN

<kyamashita>2016-05-27 11:23 *** TOPIC GNU Guix | https://gnu.org/s/guix/ |

<kyamashita> videos:

<kyamashita> https://gnu.org/s/guix/help/#talks |

<kyamashita> bugs:

<kyamashita>I pushed the wrong button...

<notadrop>Is anyone else finding it rather impossible to boot from the GuixSD installer USB/flash drive image?

<lfam>No, that's how I've done it

<lfam>How does it fail for you?

<lfam>And, how are you putting the installer on the USB flash drive?

<lfam>What computer are you trying to install it to?

<notadrop>lfam, a machine with Libreboot (fork of coreboot) as the firmware w/ GRUB payload is the target machine

<notadrop>lfam, I am creating the disk image as suggested by Guix docs using "dd"

<lfam>Ah, I remember much discussion of how to install to libre boot

<lfam>Libreboot installs its own GRUB, right?

<kyamashita>lfam: Yes.

<notadrop>lfam, the machine I'm on uses GPT/UEFI... should the partition table on the GuixSD installer flash drive be MBR or GPT?

<notadrop>that could very well be the cause of my problems!

<lfam>Please search the mailing list archives to see if somebody explained this already. That's all I'll be able to do, anyways

<notadrop>lfam, okay, I will :) thanks anyway

<kyamashita>notadrop: How are you attempting to boot the USB drive?

<lfam>People have installed GuixSD on libreboot, but it's not as straightforward as non-libreboot

<kyamashita>notadrop: I had to drop to GRUB's commandline, set the USB drive as root, and use the configfile command to load GuixSD's /boot/grub/grub.cfg file.

<kyamashita>notadrop: I also boot that way on my machine.

<notadrop>kyamashita, from the GRUB payload of libreboot. I've tried to run "ls (usb0)/" in the GRUB command line and it tells me it's an unsupported filesystem... and GRUB can read almost any file system, so it's definitely some kind of corruption going on AFAIK

<notadrop>yeah I was going to try that but GRUB can't even read the FS on the flash drive!

<notadrop>I'll try re-imaging it with an MBR table...

<notadrop>I think that might do the trick

<notadrop>I have a funny feeling that removable drives only support MBR

<notadrop>I mean flash drives, not disk drives or SSDs

<kyamashita>notadrop: And if you just run ls, GRUB doesn't list the USB drive as having partitions?

<notadrop>kyamashita, correct

<kyamashita>notadrop: If it didn't, then there's probably a corruption problem. It's worked every time I've done it.

<kyamashita>Hold on...

<kyamashita>ACTION is away

<notadrop>okay, thanks

<notadrop>one thing I could also try is loading a known working live image onto this flash drive. if that doesn't work, then it's clearly a hardware issue with the flash drive

<notadrop>This is ODD: If I give the device a new MS-DOS partition table, and create a volume spanning the entire disk with type Unformatted, GParted then shows me an Ext4 FS with label gnu-disk-image

<notadrop>I'm going to try booting this, hold on, brb

<notadrop_mobile>aha, progress. ls now lists (usb0,msdos1)

<notadrop_mobile>perfect.

<notadrop_mobile>how do I load another grub.cfg from the grub command line? which command do I use?

<notadrop_mobile>I hit tab and nothing is jumping out at me

<kyamashita>notadrop_mobile: configfile

<notadrop_mobile>thanks

<notadrop_mobile>SUCCESS thank you everyone

<notadrop_mobile>time to install Guix

<kyamashita>notadrop_mobile: Yeah! You're welcome

<notadrop_mobile>I ama GNoob... how do I write buffer to disk and exit in Zile?

<notadrop_mobile>am a*

<notadrop_mobile>oh nvm there are man pages :)

<notadrop_mobile>thank God nano is in this image

<notadrop_mobile>I have no idea how to use zile

<kyamashita>Zile comes by default as a very lightweight Emacs clone.

<notadrop_mobile>...and I have no idea how to use Emacs. haha

<kyamashita>notadrop_mobile: That would explain it. :-P

<notadrop>back, I got dc'd

<kyamashita>Does anyone else here use Guile WM?

<kyamashita>Guile WM being the Guile-based window manager packaged in guile-wm.scm.

<kyamashita>It crashes out of nowhere for me, and I've had to fallback to using Emacs as my WM. Has anyone had better luck?

<davexunit>I haven't used it in a long time

<jlicht>In the store monad, how can I get from a derivation to an actual executable package path?

<jlicht>I have a git package object, on which I can call package->derivation in the store monad. My goal is to have access to a git executable

<davexunit>jlicht: you need to use the outputs of that derivation

<davexunit>and then append "/bin/git" or whatever

<davexunit>keep in mind that the file name will not be valid until you have built the derivation

<davexunit>it sounds like you may want to use g-expressions

<jlicht>davexunit: makes sense, I'll be doing some reading then.

<notadrop>I'm installing the system right now and Guix's output looks... beautiful

<notadrop>I thought pacman was good but this is really nice

<notadrop>it tells me exactly what is going on

<kyamashita>notadrop: Ikr? I came from distro hopping and I'm staying here.

<notadrop>my system's firmware is GNU as well :)

<davexunit>:)

<notadrop>once GuixSD is installed my system will not only be 100% Foss but 100% GNU

<davexunit>glad you folks are liking it.

<notadrop>I'm going to make a little stencil to cover up the glowing Apple logo and turn it into a glowing Gnu lol

<jlicht>notadrop: You could still put the Hurd on there though ;-)

<jlicht>notadrop: someday, maybe

<notadrop>jlicht: when I'm sure it will be at least somewhat compatible with the hardware yeah

<kyamashita>jlicht: I'm going to run Hurd on my new old desktop.

<notadrop>I've wanted to be a Hurd developer since I found out about it

<kyamashita>notadrop: Same. I like the idea of microkernels.

<notadrop>so do I

<notadrop>I can see why some Linux people like systemd. they are both monolithic programs

<notadrop>I dislike both

<kyamashita>notadrop: As long as they're both free software, it's a matter of design principles and preferences.

<notadrop>and I hope Hurd will have an emphasis on security and modularity

<notadrop>unlike Linux

<kyamashita>notadrop: Those are two consequences of a microkernel architecture IIRC.

<notadrop>which is why I also like the idea of having one :)

<notadrop>the only microkernel I know of is QNX which is totally commercial

<notadrop>is there a FOSS one in the modern world?

<notadrop>besides hurd

<kyamashita>notadrop: There's Minix!

<jlicht>nothing wrong with commercial software, notadrop ;-)

<notadrop>true! I do like Photoshop and the CS a lot

<jlicht>proprietary software is more the problem

<kyamashita>notadrop: QNX is proprietary, that's the deal breaker.

<notadrop>Creative Cloud is an abomination though

<notadrop>right. that's what I meant.

<notadrop>proprietary

<notadrop>The GuixSD install process is so wonderfully simple

<kyamashita>notadrop: Minix is a BSD-licensed microkernel. Back when it was for academic use only, Linus Torvalds wrote Linux in large part by referencing Minix's code.

<notadrop>Arch is just a major f***ing pain in the @$$

<notadrop>I'm glad to be done with arch

<notadrop>oh cool!

<notadrop>does MINIX support any modern hardware and/or software

<notadrop>I.e. is it still in active development by anyone

<kyamashita>notadrop: Yes. Some students and volunteers work on it. They might have very few paid developers, I'm not sure.

<kyamashita>notadrop: Minix supports very little hardware, and the BSD license allows proprietary forks, so I'm not terribly excited by it quite yet.

<notadrop>One thing I've wondered for a little while... why does GNOME now require systemd? systemd is contrary to a lot of the goals of the GNU project AFAIK and GNOME is a part of GNU

<kyamashita>*specifically 2- and 3-clause BSD licenses.

<notadrop>right

<notadrop>yeah I'm not a big fan of proprietary OSes, firmware or hardware

<kyamashita>notadrop: How is systemd contrary to GNU's goals?

<notadrop>if you want to run proprietary userspace applications on top then sure

<notadrop>kyamashita: perhaps not GNU's goals...

<notadrop>but the UNIX philosophy

<notadrop>I always undetstood GNU to be a more user friendly and Libre UNIX

<adfeno>notadrop: Free software isn't in favor of the "UNIX philosophy" and it isn't against also.

<notadrop>true.

<kyamashita>notadrop: I'm not too big on UNIX. Neither was RMS. He mainly designed GNU around UNIX for compatibility purposes.

<notadrop>I guess then systemd is against my philosophy and what I believe to be the benefits of *NIX operating systems

<adfeno>In fact... Historically, the free software movement was born from people who initially didn't use UNIX.... They used ITS

<adfeno>Which is somewhat the antitesis of UNIX's.

<notadrop>hm

<kyamashita>ACTION vigorously researches ITS

<notadrop>ACTION looks up ITS on wikipedia...

<notadrop>incompatible timesharing system?

<kyamashita>Yes.

<notadrop>Wow. It sounds like ITS was really a pioneering OS. Many of its features, which were new back in the 60s, are everywhere today

<kyamashita>TECO looks pretty cryptic at first glance.

<notadrop>And EMACS was originally written for it! coolc

<notadrop>cool. *

<adfeno>For story of the GNU project and it's relation to ITS, see: http://www.gnu.org/gnu/thegnuproject.en.html

<adfeno>Or http://www.gnu.org/gnu/thegnuproject.html for other languages.

<notadrop>thanks!

<kyamashita>adfeno: +1 free software history

<kyamashita>ACTION is away for a little

<adfeno>Hahahaha

<adfeno>I forgot "hi"... :D

<notadrop>I'm pretty damn excited that I'll soon be running GNU on this Macbook including firmware

<notadrop>I've liked apple hardware since I got a macbook a few years ago, but being a Linux used and later a FOSS enthusiast, OSX leaves much to be deaired

<notadrop>desiredb

<notadrop>fat fingers... on mobile while Guix installs lol

<kyamashita>notadrop: It's a pretty good retaliation for Apple's unfree software system.

<notadrop>Yeah. Pretty near perfect software on pretty near perfect hardware

<kyamashita>notadrop: OSX took what fun there was left in UNIX and destroyed it.

<kyamashita>ACTION shudders at OSX filesystem

<notadrop>yeah it's odd lol

<notadrop>not as bad as windows though

<notadrop>at least OS X is based on BSD

<kyamashita>When I "have to" run proprietary software for school, I use the Macs.

<notadrop>Yeah. Mac OS X n't as broken as Windows

<notadrop>isn't

<adfeno>notadrop: As a next challenge: give credit when its due. The system your using is only possible because of the efforts of the GNU project.

<notadrop>Damn straight!

<notadrop>GNU World Order :)

<kyamashita>They're both just as bad to me, but at least I can stop using the proprietary software quicker on a Mac because of its SUS verification.

<adfeno>notadrop: Just call it "GNU/Linux" (or "GNU with Linux", or "GNU+Linux"). That should be fine.

<notadrop>Yeah, I only accidentally use the term "Linux" to refer to my OS anymore v

<notadrop>of course when I refer to the kernel I call it Linux

<notadrop>even when talking to people about what I run on my machine I say "GNU Linux" now

<kyamashita>Same here

<adfeno>:D

<notadrop>hopefully in a few years, perhaps with a little beta testing or even contribution of code on my part, I can just tell people I run "GNU"

<notadrop>(Hurd kernel)

<adfeno>Well... That would be lovely... :D

<adfeno>... Although we do need to have coders in other areas.

<adfeno>Not on the kernel side for now.

<notadrop>I am choosing to pursue schooling and a career in development not only because I enjoy working with computers as a hobby (so why not get paid for it)

<adfeno>Freeing JavaScript is one of the critical areas for now.

<kyamashita>adfeno: Indeed.

<notadrop>but for philosophical reasons too

<notadrop>FOSS inspires me

<notadrop>I hope to make the world a little bit free-er with my work

<notadrop>yes free JS would be great

<notadrop>what are the issues with it currently? besides the huge privacy and security compromise one makes when executing JS

<adfeno>Of course, if notadrop wants to only work in Hurd because that's the only thing he can do, that's OK. But he also has to choose wisely where he can be of most help.

<notadrop>would it be that the user cannot audit the JS a website wishes to run on their machine

<kyamashita>notadrop: That's pretty much it. Plus the odd delivery of JS to the user.

<notadrop>adfeno: I have an eye for design and marketing as well and artsy things

<notadrop>I want to work in UX as a hobby or even a job

<notadrop>but low level things like kernels also interest me :)

<adfeno>notadrop: Idem (I also keep an eye on arts and marketing)

<adfeno>Also on public relations.

<notadrop>I like the GuixSD logo and branding a lot

<notadrop>kudos to whoever is doing that work

<kyamashita>adfeno: I've had many people ask me what's going on with my computer or why I won't use certain services.

<adfeno>kyamashita: Yeah... I also receive this questions.

<kyamashita>adfeno: Most I've talked to were genuinely interested, even it they didn't change their usage patterns.

<notadrop>hey the more of us who make people question, even a little, the status quo.... the better

<kyamashita>Yes. Let people know that these issues are important.

<notadrop>even I wasn't won over to the ideals of FOSS and GNU over night

<notadrop>I'd dabbled in GNU Linux for years before I really realised how important FOSS is

<adfeno>notadrop: Regarding non-free JS: the primary/immediate issue isn't privacy, isn't security, isn't transparency (well, it can involve part if it), isn't accessibility (in the "disability" sense)....

<kyamashita>notadrop: It took me about 2 years of serious flip-flopping between perceived convenience and freedom.

<notadrop>same here!

<notadrop>perhaps longer but, the same idea

<adfeno>... The real issue with non-free JS is related to which freedoms are made available for the visitors of the site that are obligated to execute the non-free JS.

<adfeno>"Pro-free-JS" free software activists don't go around criticizing *every site that uses one single line of JS*, they only criticize one single line of *non-free* JS.

<adfeno>If it weren't for JS to be client-side... There would be a bunch of services as software substitutes.... And then society would loose the *possibility* to control their own computing.

<notadrop>adfeno: very true.

<notadrop>I am glad I have the ability to run things like NoScript

<adfeno>notadrop: Make sure to try LibreJS also...

<adfeno>LibreJS blocks only non-free JS. Not all JS (which NoScript does).

<notadrop>oh nice :)

<notadrop>I will. I think IceCat comes with that.

<notadrop>there are some sites I don't want to run JS on my machine, period. noscript is good for blocking them

<davexunit>librejs blocks a lot of free javascript

<davexunit>because free javascript needs to be explicitly marked as such in a format that librejs can understand.

<notadrop>how does one distinguish between free and nonfree JS?

<davexunit>notadrop: you add metadata files to your website that librejs will read

<davexunit>that map js file to the license and corresponding source code

<davexunit>files*

<notadrop>aha

<notadrop>GuixSD does a LOT of checks

<notadrop>but once it's done, I'll have a distribution that's ready to rock and roll. So that's nice.

<notadrop>Looking back on my time as an Arch user... I did learn a lot, but that distro is just masochistic

<adfeno>Hahhahaha

<davexunit>notadrop: sounds like you are building everything from source. was that on purpose?

<notadrop>no, I used arch not gentoo...

<adfeno>davexunit: If he is really new to GuixSD, my guess is "yes".

<notadrop>yes it was

<davexunit>notadrop: so you disabled substitutes?

<notadrop>I wanted to learn how the software does things, under the hood, I wanted to do it myself

<notadrop>davexunit: substitutes?

<notadrop>what are we talking about

<adfeno>davexunit: At least 4m[onths] ago when I tried to install GuixSD, it didn't tell me to enable substitutes by default.

<adfeno>I had to check the manual and look for "substitute" to find out.

<notadrop>oh I have no idea what you guys are talking about

<notadrop>I just read the intro and the install guide from the GuixSD docs

<notadrop>I'm a former arch user. I'll figure out the rest as I go

<adfeno>(actually, I looked for "build" in the manual and found the instructions to enable substitutes).

<notadrop>is GuixSD a build from source distro...? please say no

<notadrop>I was under the impression it had packages

<adfeno>notadrop: It does.

<notadrop>okay, good. building from source on a C2D would take forever

<adfeno>The packages are called "[build] substitutes".

<adfeno>They are not enabled by default.

<notadrop>okay

<notadrop>um, did I have to enable that in the config file?

<notadrop>for the installer

<notadrop>if so this installation may be ready in about a week

<adfeno>notadrop: Actually, according to my past experience, you enable right after starting the live meidia.

<notadrop>oooooh boy

<adfeno>s/you enable/you can enable/

<notadrop>time to hold down power, reboot and reformat

<notadrop>installing the entire base system from source... on a nine year old laptop

<notadrop>lol

<notadrop>not a good idea

<notadrop>how big is the source of a base desktop install?

<notadrop>are we talking GB?

<adfeno>Let me make a quick check to see if the basic installation instructions don't mention this issue of "[build] substitutes". If they do not, then perhaps we need to change the documentation to aid users on how to do so.

<notadrop>they don't!

<notadrop>yes

<notadrop>please do

<notadrop>oh my God

<notadrop>it's done

<notadrop>it didn't even take an hour

<notadrop>wow

<adfeno>?

<adfeno>???

<notadrop>installing

<notadrop>apparently from source?

<notadrop>on a core 2 duo

<notadrop>with 4gb memory

<adfeno>notadrop: OK... calm down... I think we're seeing things.

<notadrop>I don't think I believe it. I must have enabled substitutes

<notadrop>well it just tried and failed to install grub. which is okay, I can do that manually

<notadrop>and dropped me back to a shell

<notadrop>sounds like installation is done to me.

<adfeno>Can you reboot and try the installed system?

<adfeno>Or do you need to leave IRC for that?

<notadrop>I'm on my phone right now

<notadrop>so no

<adfeno>OK, at least we can have feedback from you. :D

<notadrop>I have to generate a grub.cfg even if it complains that GRUB can't be installed

<daviid>i see someone just sent an email to package guile-gnome for guix, great!

<adfeno>daviid: Hahaha :D

<adfeno>That's great :D

<notadrop>adfeno: Then I can just insert the GRUB config for the GuixSD into my Libreboot ROM and re flash

<notadrop>I could always write my own grub configuration file too, modify the one on the GuixSD install image

<davexunit>adfeno: substitutes are enabled by default on the GuixSD isntaller

<davexunit>notadrop: guix is a source-based distro.

<notadrop>davexunit: yeah there is no way a core2duo compiled the entire base system in under an hour

<davexunit>but we also provide pre-built binaries.

<notadrop>davexunit: nice.

<davexunit>provided that substitutes are enabled and available.

<daviid>adfeno: is this you?

<adfeno>davexunit: Wow! It's strange that they are enabled by default. At least in 2015, when I tried it for the first time, it tried to build everything.

<davexunit>notadrop: since all builds are identified by a unique hash, we can check to see if a trusted remote server already has it built and just download that instead of building for ourselves.

<davexunit>adfeno: you probably were using something too out of date.

<notadrop>davexunit: for the record, I love the idea of a source based distro with substitutes. more power to the user

<daviid>because I have 2 quizz for that someone :)

<notadrop>davexunit: brilliant

<davexunit>adfeno: we only keep binaries for a short while due to space limitations.

<notadrop>I already love Guix

<davexunit>you should always run 'guix pull' before installing GuixSD

<davexunit>to get the latest version of guix that will have the greatest number of pre-built binaries available.

<notadrop>oh... is that in the quick install guide

<davexunit>I believe we mention it somewhere now.

<davexunit>not sure though

<notadrop>okay let me check

<adfeno>daviid: No, it wasn't me. :D

<notadrop>not in there

<adfeno>My current email is adfeno@openmailbox.org.

<notadrop>davexunit:

<notadrop>Guix pull is absent

<davexunit>the MOTD on the guixsd image should mention it, I think. I don't remember.

<notadrop>at least according to "Find on page"

<davexunit>if it's not there then someone should make a patch

<notadrop>um what motd

<davexunit>I've been asking someone to make a patch for this for ages

<davexunit>should just do it myself so I don't have to deal with this anymroe.

<adfeno>notadrop: When you start a terminal the first time you boot the system...

<notadrop>I'm new to this distro I'm not some 1337 power user... why do the docs assume I am

<notadrop>:/

<adfeno>... The message that appears is MOTD.

<notadrop>oh okay.

<notadrop>I don't think I read it. heh

<notadrop>that's my fault then.

<davexunit>notadrop: because it's a distro for people that are very comfortable with the command line and troubleshooting system problems

<notadrop>well that's me

<davexunit>notadrop: I don't actually know if its in the MOTD. it should be, or in the manual, and I thought that as of the last release something was added.

<davexunit>if not, then it needs to get added.

<notadrop>I just can't run a command I don't know exists. if I am supposed to read the man pages for everything first then please say that some where

<davexunit>notadrop: I just told you its an omission

<davexunit>a bug, something to fix, something that isn't right, etc.

<notadrop>ok yeah

<notadrop>I appreciate you looking into it

<davexunit>people come in here all the time wondering why they are building everything from source and its always because they didn't run 'guix pull'

<notadrop>I can always just do an update from the installed system. it'll just take a little more time than starting with the newest versions but whatever

<notadrop>actually I installed all the packages from the mirrors

<davexunit>then you were never building from source

<notadrop>exactly... yeah. because it checked the hashes

<notadrop>so it must've been downloading packages not source

<davexunit>if you saw lots of download progress bars

<notadrop>yeah

<notadrop>okay

<bavier>there's a note in the "Invoking guix system" section

<davexunit>and little to no compilation logs

<notadrop>makes sense

<notadrop>yep no compilation logs, but lots of output I'd never seen before so I wasn't sure

<davexunit>okay