IRC channel logs

2019-01-01.log

back to list of logs

<apteryx>arg... first time with CVS. How does this keep the global state of a project? The equivalent of a git-ref?
<apteryx>(i'm trying to fix the build of emacs-w3m)
<apteryx>I'm looking at `cvs rlog` now, I see "revision" strings such as 1.2342, would that be safe to use?
<apteryx>also, it's not obvious to me how to find the latest one (short of some shell scripting)
<janneke>apteryx: i don't think cvs has a global state, it's just revisions per-file
<apteryx>ouch.
<apteryx>our tidy package suggests I might be able to use a date as a revision
<apteryx>that would correspond to a snapshot of the project for that commit/date.
<janneke>yes, that might work
<apteryx>ok, I'll try this, thanks
<reepca>The store has to be on a single filesystem, right? Are there filesystems that can exist across multiple devices? What happens when the store can't fit on a single storage device anymore?
<THFKA4>your builds will start failing
<THFKA4>lvm can create logical volumes across multiple physical drives
<THFKA4>some file systems support it too (i think btrfs and zfs)
<reepca>THFKA4: good to know, thanks!
<THFKA4>l/buffer 11
<THFKA4>whoops
<efraim>what do we have left holding back a staging merge?
<efraim>when I'm looking at a CVS repository, how do I know what revision to check out?
<ennoausberlin>Hello guix.
<ennoausberlin>Can someone confirm, that chicken-install readline fails due to read-only file system? How can I prevent this?
<wigust->ennoausberlin: Maybe ‘export CHICKEN_REPOSITORY=~/.eggs/lib/chicken/8’ and ‘export CHICKEN_DOC_REPOSITORY=$HOME/.eggs/share/chicken-doc’?
<reepca>Hm, my guix package -u seems to be stuck just after it finished downloading texlive... strace says it's stuck on a "read(14," and that's all it has to report.
<rk4>reepca: you might be able to hunt down what it's trying to read via /proc/$x/fd
<reepca>well, it's got fd 14 open, but it looks like an empty file to me, and the corresponding /proc/fdinfo/14 doesn't have much useful information.
<rk4>and ls -l /proc/pid/fd/14?
<reepca>hm, links to 'socket:[236273393]'
<roptat>Hi guix!
<janneke>Hi roptat!
<roptat>So I kind of worked around my parser issue
<janneke>nice
<roptat>And I'll try to compile only a subset of the files in the sources of scala with my bootstrap compiler
<roptat>Hopefully, I won't have to support xml vblues
<roptat>I also found that the grammar in their documentation is incomplete
<roptat>Scala is much more complex than what I initially thought
<roptat>So i'm trying to compile the files and fix every issue I find that way, until my parser can parse everything
<roptat>Then I'll work on the typer which is going to be hard
<roptat>I'm having some troubles with guix build -f
<janneke>...much more complex than what I initially thought -- a good thing you already started before you knew that ;-)
<roptat>I use (source (local-file ".")) and guix complains "regular file expected"
<reepca>local-file has a #:recursive? flag for non-regular files
<roptat>Ah, I'll try that
<janneke>i found that gawk-3.0.0 can be configured without gawk (unless modern versions), but it sed features that are not currently supported by Gash
<roptat>build failed: illegal name: `.'
<roptat>I must be doing something wrong
<reepca>seems it doesn't like relative paths
<roptat>or maybe local-file is for files, not directories
<reepca>no, it definitely works with directories
<roptat>ok
<wigust->roptat: Maybe ‘(dirname (current-filename))’ instead of ‘.’?
<roptat>that's it, thanks!
<Laalf>does someone use a channel with cuirass? for example mbakke made https://gitlab.com/mbakke/guix-chromium but i wasnt able to set it up because "unsupported URL protocol". if someone has an example for me that would be great.
<adfeno>Hi all
<amz3>happy new year all
<adfeno>Hi all, again! :D
<adfeno>Does anyone have issues with IceCat choosing the wrong default font for the GUI ? (Note that this doesn't talk about webpage text, although these *are* incorrect also, I want to address the GUI font first)
<wi24rd>I used to use icecat installed by guix, but the fonts are all square...so I quit
<adfeno>I think they were all square beucase it seems to be picking the wrong default
<wi24rd>Yes, I do think so, however I doesn't have time to handle this
<adfeno>So in your case it was probably some font which was supposed to replace some non Latin characters, but only had placeholders at the time
<roptat>someone talked about it yesterday I think
<adfeno>Oh, interesting
<adfeno>I'll try to debug the issue the best I can
<adfeno>Using `strace icecat` gave no useful results so far
<roptat>I think the solution was to set XDG_CONFIG_DIRS=~/.guix-profile/etc/xdg and XDG_DATA_DIRS=~/.guix-profile/share
<adfeno>I do see that the wrong font is picked up, but I don't know why yet
<adfeno>Hm... will try this now, just a second
<adfeno>It is somewhat strange that there is no ~.guix-profile/etc/xdg
<adfeno>~/.guix-profile/etc/xdg
<adfeno>BRB
<adfeno>OK, it didn't work
<apteryx>wi24rd: this happened to be some locale/graft issue in my case when I had the problems.
<apteryx>Try upgrading your GuixSD system and updating your whole profile too
<apteryx>s/happened/appeared/
<apteryx>in my case it went away by itself following a complete update IIRC.
<apteryx>emacs-w3m build fixed on master
<adfeno>Using the "Refresh Firefox" from about:support doesn't work.
<roptat>adfeno, this is the thread on the mailing: https://lists.gnu.org/archive/html/guix-devel/2018-12/msg00560.html
<roptat>so, can you also try export GIO_EXTRA_MODULES="$HOME/.guix-profile/lib/gio/modules${GIO_EXTRA_MODULES:+:}$GIO_EXTRA_MODULES" ?
<adfeno>Hm.... didn't work, even after exporting GIO_EXTRA_MODULES as you did
<adfeno>I'll check my system's configuration again (Trisquel 8.0)
<roptat>sorry I can't help you more than that :/
<roptat>maybe try to run fc-cache -fv?
<adfeno>OK
<roptat>or maybe install fonts with guix
<apteryx>why does downloading substitutes report on new lines instead of modifying the current line? I find this output unnecessary verbose.
<janneke>samplet: i found why I couldn't bootstrap bash in Guix; Gash's bash-without-bash bootstrap test was leaking Gawk into the build process -- we don't have that yet :-(
<janneke>however, i finally managed to build gawk-3.0.0 and added a test for that in Gash
<samplet>janneke: Oh! I could have told you that. Sorry.
<samplet>You can build Gawk? Amazing!
<samplet>janneke: Your strategy is to basically re-implement what Autotools would do in Guile/Guix, right?
<janneke>samplet: np! yes, i had to enhance sed support, add (a very naive) diff and cmp...
<janneke>samplet: oh yes, but that would be a far end-goal
<janneke>for now, i'm just trying to bootstrap Guix without any dependencies besides Guile, Gash and Mes
<samplet>I mean that you manually add what the Autotools results would have been, had you been able to run Autotools.
<samplet>Not that you can run all the checks and everything.
<janneke>oh, yes I've been doing that as a shortcut on my wip-bootstrap branch
<janneke>however, i'd like to remove most of that and try to ./configure; make real soon
<janneke>thanks to the new Gash we should always be able to run ./configure and i can remove the scripted/pre-cooked configure output hacks
<janneke>gnu make-3.80 supports building without make or gawk, then we build gawk using gash+make, then hopefully bash -- we're getting pretty close
<samplet>Right. That seems reasonable enough.
<janneke>make's configure needed: ls -t, sleep and it also needs job backgrounding (which historical gash has...) but i patched it out of configure for now :-)
<janneke>hmm, and it also needs: SED: unsupported function (n)
<samplet>janneke: Background jobs, eh? I can take care of that sooner or later.
<samplet>Fortunately, sed is pretty easy to extend.
<adfeno>roptat: Also, I have fc-match to fonts different than the one IceCat uses as default.
<janneke>samplet: yes, esp. after all the work you already did -- that's great!
<janneke>samplet: it does: ( sleep 1 ) & ac_foo=$! .. etc
<janneke>hoping you can steal some code/ideas from old gash' script.scm/pipe.scm/jobs.scm
<samplet>janneke: I’m sure I will. Having a head start with the data structures (at least) will be nice.
<samplet>janneke: Is that filling in for some other attempt at randomness? Like is it a fallback for “mktmp” or something?
<samplet>*mktemp
<Laalf>what does guix gc: error: build failed: executing SQLite statement: FOREIGN KEY constraint failed mean? and what did i do?
<janneke>samplet: oh, i'm not sure -- i never know how to read configure scripts
<pkill9>Laalf: I and someone else get that error
<pkill9>i think they reported it to the bug mailing list
<Laalf>hm... alright thanks
<danielszmulewicz>Hi, I'm following the installation guidelies for guix as documented in the manual, everything went fine until 2.6 Application Setup, where "$ guix package -i glibc-locales" gives an error: "guix package: error: build failed: opening directory: File name too long"
<danielszmulewicz>I'm on Android, which is probably key.
<danielszmulewicz>Internal partition.
<danielszmulewicz>Running atop Arch via the Userland app.
<danielszmulewicz>Bit of a crazy setup, admittedly.
<danielszmulewicz>getconf NAME_MAX / outputs 255
<danielszmulewicz>getconf PATH_MAX / outputs 4096
<pkill9>Laalf: there's a couple bug reports, http://lists.gnu.org/archive/html/bug-guix/2018-12/msg00333.html and http://lists.gnu.org/archive/html/bug-guix/2018-08/msg00036.html
<janneke>yay, Gash hash configure generate make's build.sh!
<samplet>\o/
<janneke>it shouts about `date' and `uniq' but it seems we don't really need those yet
<lsl88>hi guix!
<wigust->hi lsl88
<janneke>hmm, guess we need uniq -- oh well
<samplet>janneke: Like sort, I bet (hope) it is just the simplest case.
<lsl88>sorry for the delay. I need help :) I installed shellcheck to make my scripts compatible with sh
<lsl88>I have one script that sources another one
<lsl88>that is in a subdirectory
<lsl88>my tree is: in the root I have myScript.sh and in that script I source (.) another script I have under the scripts/ subdir
<lsl88>that script creates a file in the root dir
<janneke>samplet: yes, most probably -- i'm only adding -c
<janneke>i would like it if we were to create a real nice shelly line-based programming interface for these; still cutting corners right now
<janneke>something like substitute* in Guix
<lsl88>if I cd scripts subdir and run the subScript.sh, it works fine, and it does what it has to do (create a file),. but in the big one it doesn't. I am sourcing ./scripts/subScript.sh
<lsl88>wigust-: hi!
<samplet>janneke: It’s getting harder to balance our desire for nice interfaces with the build scripts’ desire for completeness. (At least I think so.)
<janneke>samplet: i've let the nice DSL/library go for the short term and am going for bootstrapping first
<janneke>we will want to see how easy it is to use Guile 2.0
<samplet>I’m on the same page.
<janneke>i think from where we are now, a grand bootstrapping success is the low hanging fruit
<samplet>Yeah, and that is quite incredible!
<janneke>we work to a 0.1 release, package it for Guix, replace Bournish and then we can hopefully recruite more hackers to help with the shelly DSL/library thing
<janneke>indeed!!
***rekado_ is now known as rekado
<janneke>samplet: isolated interesting bug that broke make's configure
<janneke>a trap set is triggered upon a subshell's exit
<janneke>adding a failing test
<samplet>janneke: Right. The spec. says to reset the traps, but I never implemented that....
<samplet>I think there’s a Guile procedure that would do it.
<janneke>it seems like such a detail...
<samplet>Wait, no. You would need to delete the custom “atexit” trap set up in the environment code, too.
<samplet>janneke: I will fix it later (in a few hours) if you like.
<janneke>until you find that confdefs.h accumulates #define's and then gets reset, etc.
<janneke>oh, that's great! i have a nice test and an ugly workaround so i can continue with the make/gawk bootstrap :-)
<samplet>Great! I’ll email you when I push a fix.
<joshuaBPMan>Hello! After a lot of trial and error, I successfully installed guixSD on a Macbook 7,1! I am having a weird error where the touchpad seems to only let my mouse move up and down. That's a bit odd, but other than that, it seems to be working just fine.
<joshuaBPMan>And I'm surprized at how well Icecat works (when my mouse works...).