<lfam>morgansmith: Yay at having gotten it working! I'm not really qualified to say if it's done "right" or not but I'm celebrating with you <civodul>like i wrote we'll probably remove the 'store' argument of 'options->transformation' soonish and make it "official" <civodul>it's an interesting use case that you have <civodul>BTW i'd recommend removing "guix" from your manifest <morgansmith>civodul: thanks! I removed guix. it seems to cause trouble with emacs-guix maybe? I have a really hard time getting emacs-guix to work. Is there ever a good reason to install the guix package? <civodul>morgansmith: no, guix is "installed" via 'guix pull', so in a separate profile <civodul>emacs-guix is currently broken, there's an issue reported upstream i think ***apetresc is now known as apetresc[m]
***apetresc[m] is now known as apetresc
<mroh>not enough cookies, I guess. <maav>well, gonna sleep, bye bye :) <xelxebar>Hrm. Build on my offload machine is failing with "no space left on device" but I have about 5GB left on /. <xelxebar>Also, /tmp isn't mounted tmpfs, it's just a normal directory. /dev/shm is mounted with 5GB as well. <mroh>maybe the client is out of space? Does building on the offload machine work, w/o offloading? <xelxebar>mroh: Well, my local machine has a ~50GB on / and 1.8 on /tmp. It looks like the output isn't in the store on the build machine, so I assume that it's actually failing there and not just failing to copy over to my local machine. <xelxebar>Wait... the build log is filled with a bunch of progress bar output labeled something like "copying 520 store items..." <xelxebar>Hrm... I'm building a disk-image. Perhaps the image is too small for some reason? <xelxebar>Is there a pre-existing way to set the number of guixbuilderN users in config.scm? Or do we just need to create the accounts manually? <jonsger>oh our online manual looks amazing now <jonsger>those variable "definitions" are much better to read :) <vits-test>xelxebar: maybe with the 'guix-service-type ? <xelxebar>vits-test, jonsger: Oh... it's right there in the manual... *xelxebar buys a thousand guix for jonsger and vits-test :) <xelxebar>Man, having so many problems trying to build this disk-image. <xelxebar>It's failing to build the partition.img with this error: mke2fs: Ext2 inode is not a directory while populating file system <xelxebar>Any idea at all what could be causing this? <jonsger>mothacehe: somehow cuirass forgets to set the status of a build. So the build is set to /path/to/log.drv.bz2 but `status` is still at -2 <vits-test>xelxebar: IDK. Wonder why it is "ext2 inode". <xelxebar>vits-test: Ran with -K and noticed that the /tmp/guix-build-partition.img.drv-0/temp-root used to populate the image just *barely* fit on the drive. <xelxebar>After expanding the partition, the build completed just fine <mothacehe>jonsger: what do you mean by the build is set to /path/to/...? <jonsger>mothacehe: not the build, the log. So there was an actual build of this package resulting in the log <mothacehe>Cuirass is only aware of the build it started itself <maav>mothacehe: i wanted to talk to you, but i killed sneek trying... :( <maav>i spent yesterday banging my head against walls because I couldn't make work any encrypted-os test... because it was creating luks2 instead of luks1 <maav>i'm going to check when did that happened, but wanted to check with you, mothacehe, if you were running the tests lately <maav>grub doesn't support it for boot yet <mothacehe>the CI is running the installation tests on every Guix commit <maav>mothacehe: that's what i was looking for, thanks :) <jonsger>6|/gnu/store/wm5ymcrm6z14hg95xamasxb799kldl1q-foo-5.9.1.drv|1|foo-5.9.1.x86_64-linux|x86_64-linux|foo-5.9.1|/var/log/guix/drvs/wm/5ymcrm6z14hg95xamasxb799kldl1q-foo-5.9.1.drv.bz2|-2|1603707427|0|0 <maav>i'm creating a new test for /boot not encrypted and / encrypted <maav>as it isn't implemented... yet (see 25305) <jonsger>is the db entry, but doesn't make sense. If there is a build log, and the build passed. `status` should not stay at `-2` <mothacehe>select count(*) from Builds where status=-2 and log is not "" <jonsger>mothacehe: ah yes, this reproducer is maybe a better explanation :) <xelxebar>So, when building gnu/system/install.scm, I get an image that's roughly 3GB; however, the installation os iso provided system install iso provided guix.gnu.org is about 550MB. <morgansmith>so I'm trying to install texlive-texmf but it's a 2.61GiB download from ci.guix.gnu.org that keeps failing. The download from upstream is much more reliable. Cracking open texlive-texmf, I see we set: #:substitutable? #f for this exact issue but it doesn't seem to be working? <morgansmith>for the record I'm actually installing texlive and texlive-texmf is an input, but it's texlive-texmf that's giving me grief <ryanprior>I've been cleaning up some more of my golang packages to make another round of submissions, but a couple rounds of past submissions are still unreviewed and I don't want to pile on where there's no bandwidth <ryanprior>And if there's anything I can do to make these packages easier to review and process I'm all about doing the work up front to make things easy for y'all so let me know <bluekeys_>Got exwm working with multiple windows, nothing was wrong with it, it just works better if you tell it the correct displayport. You guys now have your own screen. <morgansmith>bluekeys_: multiple windows? Do you mean multiple monitors? <bluekeys_>morgansmith, yes. At that level of attention to detail is exactly why it took my 2 weeks to figure out what I was doing wrong! <morgansmith>exwm is a bit of a nightmare :P but I absolutely love it <bluekeys_>I'm enjoying it at the moment, haven't hit any problems. Where is it a nightmare? <morgansmith>I first really learned elisp by making a function that loops through all my monitors and adds them to exwm. it also checks the geometry so that they are mapped in order (although looking at it now I did that wrong). I kinda felt like that should've been built in. <morgansmith>plus exwm freezes when emacs freezes (oops. I opened gnus. Guess I'll go get a coffee cuz I can't use my computer right now) <morgansmith>also it's fun when you have a window open in exwm line mode and then you can't use C-g to stop emacs from doing a thing so you have to jump to another tty and just kill emacs :P <ryanprior>I've heard of people using emacs-in-emacs with exwm, which sounds absolutely necessary to me for the reasons you cite <morgansmith>I only emacs-in-emacs for certain tasks (like gnus) but I should probably emacs-in-emacs for everything really <morgansmith>Also, pro tip after you run exwm-enable run this: (shell-command "gpg-connect-agent \"UPDATESTARTUPTTY\" /bye") <jlicht>sending USR2 to emacs gets it to unfreeze gets in almost all situations ;-) <jlicht>usually works better than flat out killing emacs in my experience, if you care about any of the stuff you had open <dannym>How do I find out, given a file in a profile, like /home/dannym/.guix-profile/lib/libcairo-gobject.so, why it was put there? I tried removing the package "cairo" from the profile, but it's not in there in the first place. So I take it it was propagated? How do I find out by what? <leoprikler>You could `guix graph`, but I also suggest taking a look at $GUIX_PROFILE/manifest <leoprikler>cairo, glib, etc. get propagated through various libraries and I would assume even applications through pkg-config <ani>some bash commands don't work in guix environment <ani>which parameter shall I pass to this command ? <ani>guix environment guix --pure --ad-hoc help2man git strace <ani>Okay, like clear terminal screen: "clear" Doesn't work. <ani>nope its not working <cbaines>I don't think clear is built in to bash <ani>wait a moment, I am in ./bootstrap phase. <efraim>janneke: 5319f716cf575321bef4c501d7aebdc7220d69cf on core-updates doesn't actually add the patch for python <ani>Can we add tools in guix environment without exiting it? <ani>I want to add vi and vim. <cbaines>You can nest guix environment, but that might not always have predictable behaviour <cbaines>So if you want to "add" vim, then you can run guix environment --ad-hoc vim <cbaines>But in the general case, you're better off amending the guix environment command <ani>so exiting environment then again entering it with --ad-hoc parameter is good option? <civodul>mothacehe: yay for the guile-avahi release! :-) <mothacehe>civodul: thanks for the lightning fast review! <cbaines>ani, yeah, exiting, amending the command, and then running it again is a good option *janneke pushes follow-up commit <civodul>mothacehe: now i'm curious what your plans are :-) <mothacehe>civodul: hehe you guessed I have an ulterior motive :p <mothacehe>yes I'm searching on savannah how to upload release tarball <civodul>ah yeah, the upload process is weird *vagrantc is surprised at the lack of signed release tags for guix's guile-* build dependencies <mothacehe>oh nice! I'll take care of that and send the announcement by mail <civodul>guile-gcrypt has signed tags but maybe not all the other deps <vagrantc>i've found all combinations of signed tag, no signed tarball; signed tarball, no signed tag; and signatures, what are those? <civodul>i guess we could ping the individual maintainers, some of which might be reading this :-) <civodul>in the same vein, it'd be interesting to use "guix git authenticate" in other projects out there <efraim>having written the blog post I feel weird asking, does anyone have an example of using a custom kernel using the extra-options flag? <vagrantc>hmmm... quick review ... guile-git (no signed tag), guile-json (signed tag, key nowhere to be found) ... <vagrantc>civodul: yeah, figured this would be a good place to start asking :) <civodul>vagrantc: for guile-json the key is probably on savannah (there are tarballs there as well) <vagrantc>civodul: the key from that maintainer on sv is not the key used to sign <vagrantc>there was also at least one project that didn't use tags at all ... <civodul>it could be that the person forgot to push the tags (that happened to me in the past) <vagrantc>it had some very interesting release branches that seemed to be used consistantly in lieu of tags <vagrantc>civodul: i'll try and do a more systematic audit of all the permutations at some point <vagrantc>seems guile-zlib and guile-lzlib didn't sign the 0.0.1 releases... <mothacehe>vagrantc: hehe, I didn't even know signed tags were a thing, sorry. <vagrantc>mothacehe: oh, well, i've done some good now :) <mothacehe>civodul: I uploaded Guile-Avahi 0.4 tarball + signature (without signed tag) and sent the announcement *vagrantc wonders about re-signing tags <vagrantc>oh, re-signing tags don't propegate if people have old tags, i think... <vagrantc>there's no "tag has changed upstream" detection or something like that <mothacehe>vagrantc: it probably means removing the existing tags first which is tricky <vagrantc>i guess you could also just push new tags ... e.g. 0.0.1-signed or something like that <rekado_>there’s so much arcane knowledge surrounding TeX and the various systems built on top of it. <rekado_>I’m determined to figure out the cause of the last big issue with our modular TeX Live packages <rekado_>it’s a variant of ‘why are we using bitmap fonts?’ and associated questions <atw>I get a segfault when trying to start icecat 78.4.0esr. Have others experienced this? <rekado_>if you have additional observations to share please comment there <rekado_>adding texlive-cm-super to the texlive-union avoids the use of bitmap fonts <atw>thanks rekado_, seems like I might be able to assist if I could find the minimal set of things to delete out of my apparently-broken user profile <rekado_>atw: I think a segfault is not acceptable, even with an old profile <rekado_>there should at least be some mitigation, an error message, and a proper exit <rekado_>perhaps you could try starting without extensions first <atw>is starting without extensions safe mode? If so, "icecat -safe-mode" also segfaults <zimoun>rekado_: Cool about TeX. I have not yet tried your patch with the French accent in the toc, but I guess it will be fine. :-) The last error seems weird… <divoplade>Is there someone working on adding cross-compilation to the meson build system? <rekado_>zimoun: yes, I don’t know which of the many messages corresponds to the error <rekado_>at first I thought it might be bitmap font generation, but that’s not it <rekado_>I then thought it might be ‘Unicode char @u8:в not defined for Texinfo’ and the like, but we also see similar messages with the huge monolithic texlive package. <rekado_>it might be specific to the German translation <zimoun`>rekado_: nice for the french manual. <rekado_>I’m going through some other package definitions in (gnu packages tex) and check that they contain all necessary files. <rekado_>now that I know that cm-super is needed we can get rid of all of these (setenv "HOME" "/tmp") hacks that were introduced when we moved from ‘texlive’ to ‘texlive-union’. *rekado_ enjoys deleting code <jonsger>the declarative way of acl is nice :) <jonsger>less manual work after installation :) <jlicht>rekado_: you have a typo in 05352bf7c (argumenst instead of arguments), so you might want to rebase the entire thing before closing up shop <jlicht>you need to at least have some flaws, it keeps you relatable ;-) <jlicht>I can push my commit now (and then you can make yourself the committer /w the rebase), or I can wait on your signal to apply it myself rekado_. What would you prefer? <jlicht>rekado_: pushed, and thanks once more for working on all the tex stuff. <vagrantc>is there a "make dist" job on ci.guix.gnu.org ? <vagrantc>./doc/guix.de.texi:587: @pxref reference to nonexistent node `Transparente Emulation mit QEMU' ... and similar <civodul>vagrantc: i don't think there's a "make dist" job <vagrantc>how hard would it be to set up? might help detect these issues sooner, and more close to where they actually went wrong <civodul>then we'll need people to actually pay attention to it <civodul>IME that's the main difficulty with CI :-) <vagrantc>haven't tried it, but the installer images being made available make it a bit of a success ... as a consumer of guix-VERSION.tar.* i'd be happy to use them <jonsger>vagrantc: so you are building your own "release tarballs"? <vagrantc>jonsger: i'd like to test it before the release happens, so yeah, i've occasionally made a tarball <vagrantc>that way, i can try to get issues fixed before it's set in virtual stone <jonsger>vagrantc: that seems like a good idea :) <jonsger>I run often enough into errors with the openSUSE package... <vagrantc>i'm somewhat blocked on guile-gnutls in debian again ... or at least, that's what's blocking from being able to switch to guile 3 *vagrantc sees some new commits and tries again <maav>well, after two days tangling myself with the threads, I've created something like a marionette-test, yay! :) <maav>vagrantc: the error from makeinfo should be raised in the manual generation too, isn't it? <maav>i'm checking, my machine is slow :) <civodul>hi bluekeys_! if you're using %desktop-services (GNOME or similar), i think it just happens when you plug it in (?) <civodul>maav: tangling yourself with the marionette threads? :-) <maav>civodul: you have my first play in your inbox :) <maav>bluekeys_: i've just tested with gnome and it works, what desktop are you currently using? <nefix>hello! How can I add a package into my .local/share directory? <nefix>But I guess it can be in my GUIX_PROFILE or something like that, right? <bluekeys_>maav, exwm. I expect I have to mount it manually? I'll hit a search engine if there's nothing especially guixy I'm missing. <vagrantc>yay, my very own guix-1.0.1.23826-02c3c.tar.gz ! *vagrantc is used to projects where something like git archive ../foo.tar.gz works just fine <vagrantc>bluekeys_: it is a small accomplishment ... running make dist usually requires more than one try in my experience <vagrantc>from a clean checkout: ./bootstrap && ./configure --localstatedir=/var && make authenticate && make check-channel-news && make -j5 doc-pot-update && make -j5 dist <bluekeys_>Every accomplishment deserves a celebration. I will certainly celebrate when I eventually mount the usb holding all my passwords. Tips welcome! <vagrantc>seems a slightly short hash ... git describe --match='v*' gives me v1.0.1-23826-g02c3c51e0c <maav>vagrantc: i've just tried and just guix environment guix is not enough, hehehe, you're missing secret sauce ;) <maav>also, the version is not right, isn't it? <maav>bluekeys_: yep, you can mount them with mount, or create an emacs command to do it, of course <maav>i don't think any 'automounter' comes with our exwm installation out of the box, but I'm not sure