IRC channel logs


back to list of logs

<lyr3`>does guix accepts any argument to use more cpu processors...It seems as I am installing it with -j1 hahahahahaaha
<mange>lyr3`: There's --max-jobs, which defaults to 1. I often run --max-jobs=4 when I'm doing things.
<lyr3`>--cores=n ?????
<mange>Yeah, apparently cores defaults to "0", which means it can use as many as it wants. You can use --cores to restrict how many cores Guix uses.
<lyr3`>oh, my bad. I must to pass qemu some argument to use more cores! lmaaaaaaaooooooooooooooo
<lyr3`>oh... -enable-kvm
<civodul>Hello Guix!
<pkill9>hi civodul
<pkill9>How do you suspend the system from the comman line?
<rekado>echo mem | sudo tee /sys/power/state
<vagrantcish>hrm. python-qt is failing for me ...
<vagrantcish>ACTION tries reverting the update to 5.11.0
<efraim>Unfortunately updating pyqt doesn't cause it to build
<civodul>pkill9: FWIW i just close the lid of my laptop
<rekado>I’m now using EXWM and it doesn’t reliably suspend the system when I close the lid.
<pkill9>how do i put xcbgen (from xcb-proto package) in python's path so a package's build process can find it?
<pkill9>okay i need to set PYTHONPATH
<pkill9>how do i set an environment variable for the build process?
<civodul>PYTHONPATH should get defined automatically if it's needed
<civodul>another instance of "we don't know what's in containers":
<pkill9>civodul: it doesn't get set by xcb-proto i don't think
<pkill9>if i set PYTHONPATH manually and try to build, it works
<pkill9>i guess a search path just needs to be added
<pkill9>oh actually it might be just because i was using the wrong variable as an input
<pkill9>yes that was the reason >.>
<civodul>cool :-)
<civodul>the LWN article above mentions
<civodul>looks interesting but i'm not sure what to think about it
<civodul>rekado: did we ever hear from uniq10 on guix-devel?
<civodul>i can't even seem to find their email address
<civodul>ACTION is doing a terrible job at GSoC mentoring this year :-/
<rekado>civodul: yes, we have heard from uniq10. Last email to guix-devel was on the 9th of May (“Re: Rebasing guile-daemon branch onto master”)
<rekado>civodul: there was another email just to the mentors on the 16th.
<rekado>civodul: GSoC and Outreachy is a lot to juggle in addition to Guix bugs.
<rekado>civodul: you shouldn’t need to do much for GSoC this time. I think g_bor[m] and I have this covered.
<rekado>we might need to ask for your insights on tricky daemon internals from time to time, though.
***Salt_ is now known as Salt
<civodul>rekado: oh ok, and yes, i'm happy g_bor[m] & you are taking care of things, it's a relief!
<civodul>i was asking because i started merging things from 'guile-daemon'
<rekado>civodul: I just took a look at “/var/run/cuirass/cuirass.db” on and saw that it takes up 12G! Is this expected?
<civodul>not sure
<civodul>i suspect one problem is that we add more entries in the "Builds" table than needed
<civodul>like if two subsequent evals yield the same builds, we add them twice
<rekado>I see
<civodul>i've been suspecting this for a while but i never investigated much
<civodul>but it could be this
<snape>oh... org-contrib depends on qt
<snape>through supercollider
<snape>which doeesn't build
<jonsger>civodul: so guile-sqlite3 will be a new dependency for guix?
<civodul>snape: wat?!
<civodul>jonsger: initially an optional dependency, but at some point it'll be a mandatory dependency i guess
<rekado>supercollider does not depend on qt, but there’s an IDE/GUI using Qt.
<jonsger>civodul: so we should maybe make a 0.0.1 release of guile-sqlite3. That would make packaging in opensuse a little easier
<snape>rekado: to me, having something as input means depending on something
<snape>but maybe I used the wrong words
<rekado>snape: maybe it shouldn’t have qt as an input then.
<snape>it has qtbase
<civodul>jonsger: yeah that's what i wrote in
<civodul>we need someone to step up to take care of guile-sqlite3
<rekado>snape: the words were fine :) It was just a surprise to read that it depends on Qt, when I know that it shouldn’t be required.
<snape>oh yes, sorry for the misunderstanding
<snape>anyway, supercollider doesn't build. I'll report a bug after lunch
<civodul>IWBN to have a command in Emacs-Guix that applies a patch, runs lint & build, and prepares a reply
<pkill9>weird, my system froze after waking up from suspend and then i had to hard reboot
<pkill9>that happened when i ran `echo mem | sudo tee /sys/power/state`
<rekado>works for me.
<civodul>efraim: any progress on the pyqt front? matplotlib & co. depend on it
<efraim>civodul: I'm going to try their dev releases
<efraim>pkill9: how about 'loginctl suspend'?
<adfeno>Hi all! For those who use Org mode's ox-odt: Are you able to export to ODT?
<pkill9>hmm, i did it with both loginctl and `echo mem` and it didn't freeze either time
<pkill9>i forgot to mention, it froze when using bspwm
<efraim>Busy swapping?
<pkill9>instead of xfce4
<pkill9>what do you mean efraim?
<efraim>Sometimes enlightenment freezes up on me, I'm not sure if its related to swapping or not
<efraim>I found it was worse when I had more zram enabled
<jonsger>ACTION wonders how guix-0.14 found its way to openSUSE 15.0 :)
<pkill9>it coudl well be, ii've used this laptop for a long time now, and i got it used
<efraim>Something I've been playing around with in my head is I think debian runs enlightenment with nice -3 so its less than just 'nice' but under general system processes
<civodul>jonsger: are you packaging it?
<jonsger>civodul: together with Sleep_Walker, but only for Tumbleweed. the request to leap did someone else...
<efraim>I think Sleep_Walker did the work, I know I wasn't really involved at all despite good intentions
<jonsger>but its broken :(
<efraim>0.14 was a while ago now
<efraim>One of the problems I ran into trying to bootstrap aarch64 without enabling berlin
<Sleep_Walker>let me check it
<snape>rekado: supercollider is broken since qt update. Thus the dependency isn't superfluous.
<rekado>snape: we should pass -DSC_QT=off then.
<Sleep_Walker>jonsger: feel free to report me bug, I'm a bit busy now but I'm interested in Guix on openSUSE
<efraim>I also have fixing x265 on armhf and aarch64 on my todo list
<civodul>efraim: re QTest, i found this:
<civodul>there's a one-liner for pyqt
<civodul>too beautiful to be true?
<efraim>I should spend more time with the gentoo guys, they must see every random bug
<jonsger>Sleep_Walker: on Leap 15.0 I get this I think we had this issue on opensuse already...
<efraim>Wow, that is a nice patch, definately worth trying
<efraim>I looked and looked, didn't see pyqt's or any other repo on their site
<Sleep_Walker>jonsger: could you please check if your Guile packages were built using the same Guile version?
<rekado>snape: we should make a separate package for the supercollider IDE and GUI.
<rekado>snape: I just built it without Qt.
<snape>rekado: nice
<jonsger>Sleep_Walker: seems to be guile 2.0.14
<Sleep_Walker>jonsger: I can reproduce it in docker, but I'm afraid that it is far beyond my guile-fu
<Sleep_Walker>jonsger: I'm afraid that it might be consequence of using guile 2.0 instead of guile 2.2 (which I use in my project home:sleep_walker:guile)
<Sleep_Walker>jonsger: if you have spare cycles, please ask on #guile or ML
<jonsger>Sleep_Walker: that's what I guessing, there goes something wrong
<Sleep_Walker>jonsger: quick googling lead to
<jonsger>Sleep_Walker: guix in tumbleweed is also failing with the same error
<Sleep_Walker>jonsger: there is 2.0 as well
<Sleep_Walker>jonsger: osc bco devel:languages:misc guix; wget -O gnutls-magic-fix.patch; $EDITOR guix.spec; osc vc -m 'fix gnutls failure'; osc ci; osc sr -m 'I have fixed that!'
<snape>VLC doesn't build either...
***Digitteknohippie is now known as Digit
<cbaines>I've just seen that Xonotic has hit master! :D
<cbaines>This might not be great for my already low productivity concerning Guix patches...
<civodul>heya cbaines!
<cbaines>Hi civodul :)
<civodul>speaking of patches, the RabbitMQ patches look almost ready, no?
<cbaines>There is a locale issue with the service, which it logs about on startup
<jonsger>Sleep_Walker: pushed it to
<cbaines>Also, apart from getting the package building, I haven't poked around inside much to check if there's anything lurking in there
<jonsger>home:jbrielmaier:branches:devel:languages:misc/guix but OBS seems to a bit slow today...
<Sleep_Walker>jonsger: heh, I thought you haven't seen that
<cbaines>but yes, I want to make some time to look at it
<Sleep_Walker>I just prepared mine
<jonsger>Sleep_Walker: I thought it was a instruction for me :)
<Sleep_Walker>jonsger: btw, please, report such issues to me if you meet it
<Sleep_Walker>jonsger: well, yeah, but without response :b
<Sleep_Walker>I hardly ever get to clean installation
<Sleep_Walker>so I can miss issues like this
<jonsger>Sleep_Walker: me too, but get a "new" workstation and installed leap 15.0 and then I saw that guix is availabe for it
<Sleep_Walker>jonsger: I tried to fix guile 2.0 problems so it got updated in Factory once again and someone sent it to Leap...
<Sleep_Walker>jonsger: was 15.0 released already?
<jonsger>Sleep_Walker: on friday in prag at the oSCon :)
<Sleep_Walker>I missed oSC this time :/
<Sleep_Walker>I added Leap 15.0 as a built target so you can verify the fix directly from devel:languages:misc
<jonsger>for the next guix release we should propably package guile-sqlite3
<g_bor[m]>Hello guix!
<pkill9>why does the mpd service use port 6666 instead of the default mpd port 6600?
<pkill9>oh nevermind
<pkill9>it is 6600 by default
<pkill9>i'm losign track of what i'm doing lol
<jonsger>Sleep_Walker: should't we call "guix archive --authorizy >" in the %post section of guix?
<rekado>“<” instead of “>”, and also for
<jonsger>Sleep_Walker: the patch fixes the problem on tumbleweed. I will test it on leap 15 when its ready
<jonsger>rekado: yeah sure
<chewzerita>Hi there, im having a problem when trying to launch 'pioneer' from guixsd, it says "GLXBadFBConfig", thinkpad x200, details:
<Sleep_Walker>jonsger: excellent, I'll push it to Tumbleweed
<rekado>hey, the “etc/profile” file has two behaviours dependent on whether the GUIX_PROFILE variable is set.
<rekado>why is this useful?
<rekado>e.g. when I want to enable a separate profile, why do I have to set GUIX_PROFILE to that new profile’s location before doing “source /path/to/etc/profile”?
<g_bor[m]>rekado: do you know something about rukako? I might have missed some communication, but the 25th has passed.
<rekado>g_bor[m]: Rukako did appear here before. Here’s the relevant communication:
<civodul>rekado: re etc/profile: the reason the "symbolic" profile name doesn't appear in there is because otherwise profile derivations would depend on site-specific details
<civodul>if you see what i mean
<civodul>that is, you'd have etc/profile referring to /home/rekado
<civodul>and i'd have /home/ludo
<rekado>ah, true.
<civodul>even though it's the same profile
<rekado>I think it would be lovely if we had a simpler command for “enabling” a separate profile.
<rekado>currently this involves eval and “guix package -p … --search-path…”
<civodul>the changes to 'guix environment' that davexunit proposed a while back would cover this, i think
<g_bor[m]>Thanks, I have to go now. Will be back soon.
<happy_gnu[m]>I have a thinkpad x60 with libreboot
<happy_gnu[m]>anyone knows how does it work with guix?
<happy_gnu[m]>It gets a little hot so I am worried about having to compile
<rekado>happy_gnu[m]: it’s using an i686 CPU, isn’t it?
<rekado>we do support i686, but the build farm is often a little bit behind in building binaries for this architecture.
<pkill9>does the mpd service set XDG_RUNTIME_DIR?
<rekado>it also doesn’t have as many users as x86_64, which is why you may run into bugs that haven’t been detected by other people.
<pkill9>it can't connect to my user's pulseaudio daemon session
<pkill9>but it does when i run mpd as my user
<rekado>pulseaudio runs as your user, but the mpd service runs as a separate user.
<nckx>chewzerita: Pioneer seems to be ‘maintained’ by bavier. If they're not around, the best way to keep your bug report from getting lost in the noise is to send it to
<chewzerita>nckx: thank you
<pkill9>rekado: i specified my user in the configuration of the mpd service
<lyr3>wow...10GB wont do. 20GB must do as GuixSD relies a lot on /tmp
<pkill9>i put /tmp in my /home partition and then specify a bind-mount in my system's configuration
<lyr3>Interesting, still, too hackish for me. I like to follow standard pattern of SysOps...meh
<rekado>you only need a lot of space in /tmp if you’re compiling a lot of big things from source.
<rekado>you can also simply set TMPDIR in the daemon’s environment to something that has more storage.
<rekado>(no need for bind mounts)
<pkill9>oh TIL
<lyr3>Oh, living and learning
<pkill9>i'll just leave it as it is, since it means all other software will use it
<pkill9>rekado: how do you set set an environment variable in the daemon's environment?
<lyr3>what is the ip alternative for ifconfig xxx up?
<rekado>ip link set xxx up, I think
<pkill9>what is with all the different internet adapter management programs, e.g. ip, ifconfig ,etc
<lyr3>ifconfig relies in earlier implementations of kernel linux.
<lyr3>most output of it are lying hahaha
<lyr3>net-tools are what everyone are using...
<lyr3>wow...-enable-kvm makes everything so fast and smooth
<lyr3>my bad, here a comparison:
<rekado>“guix environment -l foo.scm --root=$PWD/.guix-profile” is neat, but it’s also dangerous, because the resulting thing at $PWD/.guix-profile is not a real profile.
<rekado>you can’t upgrade its contents, for example.
<rekado>is there a better way to get an environment profile that behaves like something that was created with “guix package -p …”?
<civodul>rekado: you can't upgrade because it lacks a level of indirection?
<rekado>I forgot why; I just remember that this was a problem we had when developing pigx.
<jonsger>Sleep_Walker: I packaged guile-sqlite3 here
<rekado>let me try that
<rekado>another unrelated thing that is mildly annoying: why do we need to add an equal sign between long option and value, e.g. in “guix package --manifest=file”?
<rekado>civodul: huh, never mind. Updating the profile *does* work after all.
<rekado>sorry for the noise
<lyr3>verbose arguments are quite common on kernel tools...some (s/guys/people) like it.
<lyr3>alias them if you can keep with their verbose
<rekado>that’s not the point
<rekado>we accept only “--manifest=file”, but not “--manifest file”
<rekado>(of course we also have short options)
<lyr3>usually, some arguments are like" lsmod -append=xx or lsmod -a mod...
<lyr3>I never see and lsmod -append xx
<rekado>Python’s argparse, for example, does support “--foo bar” as well as “--foo=bar”.
<lyr3>I knew haha...python boys wont like Unix traditons!
<rekado>ACTION doesn’t like Unix traditions either
<rekado>that’s a bit of an odd statement to make
<lyr3>well...that is a pattern throughout most system admin tools
<rekado>lyr3: coreutils (e.g. ls and cut) also support a space between a long option and its value.
<lyr3>coreutils aint unix ...haha
<rekado>well, yeah, neither is Guix
<lyr3>either way, you can always push yet another argument to guix package dev
<rekado>I don’t understand this sentence.
<rekado>what is “guix package dev”?
<lyr3>haha. you can add that option to guix package
<rekado>I think we’re miscommunicating.
<lyr3>yep, anyway:
<chewzerita>For anyone reading the logs wanting to get pioneer working on guixsd: add "RendererName=Opengl 2.1" to ~/.pioneer/config.ini
<rekado>ACTION goes to tend to the pumpkins.
<pkill9>i can't open the manpage for feh ._.
<civodul>lyr3: please avoid elusive sentences, "haha"s, and all
<civodul>people are genuinely trying to help and the passive-aggressive style doesn't help us communicate better
<ArneBab_work>I got an error when updating my guix emacs package, potentially because I was too impatient and aborted the checking: Symbol’s chain of function indirections contains a loop: guix-newest-available-packages [4 times]
<cbaines>I'm having some issues with the new ABI checking in guix records, if a version of Guix without these changes is used to compile the Guile code, then the abi-cookie's are missing
<cbaines>It would be good to release a new Guix snapshot with these changes, at least then I think I could avoid the errors by using an updated version of Guix
<rekado>someone updated our emacs package, but emacs-minimal no longer builds now.
<rekado>fails with “configure: error: The following required libraries were not found: gnutls”
<g_bor[m]>Oops... I hope I'm lucky with that... Guix pull is running now...
<rekado>oh, and now it has been reverted
<rekado>with a comment that says it should probably be done on core-updates.
<rekado>but why?
<rekado>it only causes about 383 packages to be rebuilt, but many of them are small emacs packages.
***abbe_ is now known as abbe
<jonsger>ACTION thinks it's maybe time for proper CI builds _before_ pushing to master
<nckx>chewzerita-in-the-logs: Glad to hear that you got it to work!
<adfeno>Hi all: rekado: I second that request to push the patch now
<adfeno>I know I'm not a developer, so please ignore the last message :D
<adfeno>I just typed that because I'm somewhat on a deadline, and would like to test ox-odt from new Emacs.
<nckx>adfeno: Why not apply it locally?
<g_bor[m]>rekado: also CI builds is a great idea, but I have a feeling, that it is not intuitive for emacs to break emacs minimal. It would be much more acceptable the other way around...
<nckx>g_bor[m]: I think intuitive is subjective in this case (I'd find an ‘emacs’ that extends an ‘emacs-minimal’ that itself is ‘regular emacs’ with bits removed to be... unintuitive at best). I wonder, though, wether ‘guix refresh -l’ should consider inherited packages to be dependents and how this could be done.
<nckx>ACTION has a very ugly WIP somewhere that (ab)uses the properties field to do so.
<g_bor[m]>Most probably part of then confusion arises from the fact, that package relations are not easily mapped to iheritance... It would be nice to have a tool to examine inherited packages.
<nckx>Right! My reference to guix refresh was simply because ‘guix build `guix refresh -l $foo`’ is 99% of my ‘CI’ solution, but wouldn't have caught this case, and I hacked it to do so.
<nckx>ACTION should not be rambling on IRC when completely distracted, sorry.
<atw>I'm confused about why gnutls wasn't found during my emacs 26 build. The package definition includes gnutls as an input:
<g_bor[m]>atw: that is interesting
<soundtoxin>I'm getting these errors when trying to reconfigure:
<rekado>atw: this is about emacs-minimal, isn’t it?
<rekado>soundtoxin: do you get the same error when replacing the deprecated “title” fields?
<soundtoxin>rekado: I don't know how to do that, and it looks like the area effected is related to file systems, so I'm nervous to screw it up.
<soundtoxin>looks like it's all the lines that start with (file-system or (file-systems
<soundtoxin>is there just a new similar term I should put there?
<atw>rekado: I don't think so. I was trying install emacs produced by the linked package definition. Not at all urgent or anything, just curious to me
<soundtoxin>rekado: here's a scrot of the relevant section in case you can advise me on what to change
<nckx>soundtoxin: Unless you have good reason to suspect you won't be able to just roll back to a working generation, don't be too nervous.
<g_bor[m]>rekado: What should be done with the title fields? I can't find this information...
<soundtoxin>that's a good point, I guess, but avoiding problems in the first place sounds even better to me
<soundtoxin>oh wait
<soundtoxin>there are actual lines with (title 'label)
<soundtoxin>should I delete these lines?
<soundtoxin>maybe I can get away with commenting them out
<rekado>soundtoxin: have you looked at the latest version of the manual yet?
<nckx>Just replace (device "my-label") (title 'label) → (device (file-system-label "my-label"))
<nckx>2 lines become 1.
<soundtoxin>no, rekado
<rekado>“device” also supports uuids with (device (uuid "4dab5feb-d176-45de-b287-9b0a6e4c01cb"))
<nckx>For anyone interested in the diff: 9ceeca0880248eee8c2796f5e6ebc172b95277f7
<nckx>Right, but best to change one thing at a time.
<lyr3>emacs 26.1 is out
<nckx>soundtoxin: Mad propz for posting a PNG screenshot of your configuration file. Unfortunately Icecat won't load it.
<soundtoxin>I'm guessing that's sarcasm
<rekado>nckx: eww loads it ;)
<soundtoxin>I've heard people don't like screenshots of text, but it's never bothered me
<nckx>soundtoxin: Not really.
<rekado>the font is a bit too small when viewed on my laptop display, makes it hard for me to read it.
<rekado>atw: emacs 26.1 builds fine for me.
<g_bor[m]>I am confused... how do I access the latest manual...
<rekado>it’s just emacs-minimal that fails unless you pass “--with-gnutls=no” as a configure flag.
<rekado>g_bor[m]: using an info reader.
<nckx>rekado: I'm still trying to not be disappointed by mu4e. Trying eww will have to wait.
<rekado>g_bor[m]: e.g. with Emacs: C-u C-h ~/dev/guix/doc/ RET
<soundtoxin> here was my attempt to make the changes, and here is the error output now:
<rekado>hmm, the error messages don’t look much better :(
<soundtoxin>yeah, just lost the title-related bits at the top
<soundtoxin>but it's still talking about fstab at the bottom
<rekado>soundtoxin: is this Guix from a git checkout? Or are you using “guix pull”?
<soundtoxin>guix pull
<rekado>hmm, too bad.
<soundtoxin>I do sudo guix pull && sudo guix system reconfigure /etc/config.scm
<g_bor[m]>I just tried info guix after a guix pull...
<rekado>g_bor[m]: “guix pull” currently only installs the latest Guile modules. I don’t think it updates the manual (or the daemon). That’s a know defect.
<g_bor[m]>Oh,fine my bad... I remember reading something about that. Thanks anyway...
<rekado>it’s very unfortunate, because we also don’t publish the latest version of the manual online.
<atw>ah, ok, the elisp packages in my manifest were pulling in emacs-minimal, whose build fails on guix@bb82c3d42. Thanks rekado!
<soundtoxin>huh? this sounds like a somewhat serious and confusing issue
<soundtoxin>where do you get the latest manual then? git?
<nckx>I do...
<soundtoxin>I'll admit I rarely look at the manual anyway, but I'd feel a little safer if I had a modern version of it
<rekado>nckx: you do get the latest version of the manual when running “guix pull”?
<nckx>rekado: I meant: ‘where do you get the latest manual then? git?’ → ‘I do’.
<rekado>soundtoxin: I usually get it from git, yes.
<rekado>nckx: ah, I see.
<rekado>soundtoxin: but that’s just because I use Guix from git anyway.
<soundtoxin>ah okay
<rekado>it’s not ideal.
<soundtoxin>so since my reconfigure fails right now, a reboot would just boot the last-working configuration automatically, right?
<nckx>soundtoxin: Yes.
<soundtoxin>okay. I won't worry too much then
<nckx>Things get risky when the error happens later in the process, but at the syntax-barfing level you're usually safe.
<chewzerita>What's the current status with Emacs 26.1?
<rekado>we had it a few commits ago, then it got reverted.
<rekado>should be ready soonish.
<chewzerita>rekado: i cant wait for soonish
***snape` is now known as snape
<snape>do we have a place that references ideas of easy (or not so easy) contributions?
<snape>for people who want to contribute but don't know what to do
<nckx>ACTION can't get the new file-system syntax to work either...
<nckx>Is there a(n easy) way to tell Guix/Guile not to ellipsise error messages?
<nckx>In unknown file: 1 (find #<procedure 7f96e1c34368 at gnu/system/linux-ini…> …) 0 (string-prefix? "ext" () #<undefined> #<undefined> #<u…> …) is not helpful.
<g_bor[m]>nckx: I'm also interested in that... It would make debugging easier...
<g_bor[m]>rekado: I remember something about this manual not updated thing, can you tell me why is this so?
<nckx>ACTION will take a non-easy way too, but can't imagine the current default ever being useful.
<atw>snape: I'd be interested. I have at least one thing that's potentially low-hanging fruit
<roptat>is there a way to "grep" a file in guile?
<roptat>in a package I need to find every java file that contains "@Named"
<g_bor[m]>snape: I believe we have some bugs in the tracker marked as easy. However this flag is not used consistently.
<snape>g_bor[m]: yes, but I was thinking about things that are not bugs.
<roptat>and then I need to write their names in a file
<snape>i.e. Update VLC to 3.0.3
<pkill9>is 60 degrees CPU temperature bad?
<jlicht>happy emacs release day, all :D
<chewzerita>jlicht: yipee!
<nckx>pkill9: Depends entirely on the chip? It's high but not dangerous on your average x86-alike.
<nckx>pkill9: Running ‘sensors’ (from the lm-sensors package) should give you a ‘high’ and ‘critical’ reference temperature.
<nckx>Usually a lot higher than 60.
<pkill9>ah cool thanks
<rekado>g_bor[m]: it’s just because of the way “guix pull” is implemented.
<pkill9>the critical temperatur is a lot higher than 60 degrees
<rekado>g_bor[m]: it currently only builds some Guile modules and makes them available at ~/.config/guix/latest.
<rekado>g_bor[m]: we would need a different “guix pull” to include the manual.
<rekado>g_bor[m]: civodul and I are aware of this problem, and I’m positive there’s a bug report about an overhaul of “guix pull” to address this.
<rekado>civodul: how do you find the relevant bugs so quickly?
<rekado>I find it difficult to find bug reports with M-x debbugs-gnu
<jlicht>+1 for expert tips on using debbugs effectively ;-)
<civodul>rekado: it's just that i knew what keywords to look for :-)
<rekado>I should resume work on … what was it called… mumi?
<nckx>What's wrong with this picture? (no, not the %base-file-systems indentation mistake :-)
<pkill9>anyone know why libcurl isn't reading certs from /etc/ssl/certs even though it finds the file there? I got that issue with weechat and now mpd
<civodul>rekado: definitely :-)
<civodul>i'm sure you could quite easily recruit people to work on it
<civodul>nckx: could it be that you did not recompile everything that uses (gnu system file-systems)?
<jlicht>how do people build their recently pulled, but potentially very much unstable guix git repo's?
<vagrantc>ACTION has wondered this too
<jlicht>I currently have a 'normal' guix that I use to build my git repo which I update in lockstep with the actual repo, which while functional seems a bit clunky
<vagrantc>i've often pondered if the latest symlink shouldn't be versioned much like the user-profiles and system profiles links
<civodul>vagrantc: yeah, that's the plan :-)
<janneke>jlicht: i'm using multiple git worktrees
<vagrantc>civodul: cool
<civodul>jlicht: same as janneke, and i just run "make"
<civodul>("unstable"? what do you mean? :-))
<vagrantc>i've occasionally had an update go badly enough and pretty much had to pick a guix at random and hope.
<janneke>i only update ~/src/guix after ~/src/guix-master built correctly
<janneke>yeah, that rarely happens, but when it does...such a waste of time
<civodul>ACTION just fearlessly pulls
<nckx>civodul: That's possible. I'll try recompiling. Thanks.
<jlicht>civodul: ever the Fearless Leader ;)
<janneke>civodul: that's the spirit!
<vagrantc>ACTION still slightly nervously but mostly boldly pulls
<jlicht>I will try to work it out with the worktrees, although it is getting a bit out of hand (guix-system, guix, guix-core-updates, ...)
<vagrantc>ACTION has a few patches that need following up still
<janneke>ACTION has 2 worktrees for wip-bootstrap
<cbaines>I also just pull, but I don't use multiple worktrees... it does mean that when I change branch, I do lots of Guile recompiling, maybe I should try using worktrees
<civodul>i've come to use quite a few worktrees
<civodul>it's just too wasteful otherwise
<janneke>guile just needs to compile faster!
<nckx>Yeah, Guile!
<jlicht>when is the next core-updates -> master merge?
<nckx>Seriously: once you try worktrees, it's very unlikely you'll look back. It's very likely you'll wish you'd switched much earlier.
<janneke>come to think of it; why is it that: git clean -fdx; make; would take considerable time?
<janneke>things could be cached?
<vagrantc>what's this worktrees thing?
<janneke>vagrantc: git worktree --help
<janneke>worktrees are parallel checkouts from the same .git in separate directories
<vagrantc>oh, *nice*
<janneke>quite similar to multiple git clone's, but you don't have to pull patches from one .git to another
<snape>so rekado can I push my first patch that removes IDE support to supercollider? That would at least fix the build.
<pkill9>is there something i could use to test libcurl/
<pkill9>i need to debug it
<snape>(I should add that I don't have time to do another patch for the IDE :p)
<rekado>snape: I don’t object, but I don’t use supercollider.
<rekado>I don’t know if the IDE is important.
<vagrantc>janneke: and it's a little late, but was thrilled to see your report on mes 0.14. so excited to see it progressing!
<rekado>it would be nice if we could build it separately, but maybe that’s for later.
<janneke>vagrantc: *thank you*
<snape>okay. Thanks
<janneke>ACTION feels happy and wants to note that all support is so helfpul!
<rekado>to make us not feel too bad about removing this feature, would you mind sending an email to about this, just so that we remember adding it back at some point?
<janneke>vagrantc: just built my first m4-boot using tcc-boot -- it runs but buggy
<rekado>pkill9: libcurl on its own doesn’t read anything.
<lyr3>pkill9: my feh man page is working fine
<rekado>pkill9: users of libcurl are expected to configure it to read the cert bundle from some location.
<janneke>vagrantc: working up to building gcc, first build all its dependencies
<rekado>pkill9: exactly how that is done depends on the application.
<pkill9>oh ok
<cbaines>Ooh, Guix must have got a new version of magit recently, as restarting Emacs has given me the % key to do worktree stuff!
<rekado>pkill9: so you’re better off investigating how the application itself sets the bundle location.
<vagrantc>janneke: getting that far ... wouldn't have even thought it'd be something to seriously talk about a couple years ago :)
<janneke>vagrantc: i've been hacking hard for 2y straight and got a lot of unexpected help
<vagrantc>janneke: sorry that i don't have much skill to contribute, just enthusiasm :)
<janneke>enthusiasm and support, spreading the word/idea, is so important -- please don't under estimate that
<lyr3>did you guys got pdf-tools to work?...still dont know which packages are to be installed
<rekado>lyr3: s/guys/folks/
<rekado>lyr3: install emacs-pdf-tools
<vagrantc>janneke: yeah, i can do that much :)
<lyr3>my bad. Yep, I prefer to build it through (use-package)
<jlicht>lyr3: use-package should work with guix-managed emacs packages
<lyr3>jlicht: it does work. well, time to find each package pdf-tools requires to build
<pkill9>hmm i can't seem to read manpages of any package in guix environment
<pkill9>$MANPATH doesn't seem to get set
<pkill9>or something is conflicting
<pkill9>yeah it doesn't get set
<civodul>pkill9: for MANPATH to be set in 'guix environment', you need to add man-db to the environment
<pkill9>ohh ok
<civodul>otherwise guix doesn't know about MANPATH
<jonsger>janneke: did I told you how amazing lilypond is?
<rekado>lyr3: I strongly recommend using Emacs packages from Guix, especially for more complicated packages like pdf-tools.
<lyr3>rekado: I agree. it aint compiling correctly..
<janneke>jonsger: not sure, you just did anyways -- thanks!!!
<janneke>jonsger: i had an amazing life for ~15y working on that together with my dear friend hanwen
<jonsger>It's a shame that I discovered the Mutopiaproject just recently...
<janneke>oh yeah i thought that might grow to be more important than lily itself
<rekado>lilypond is the foundation of my composition workflow.
<rekado>it’s really great.
<rekado>Scheme engravers allow me to generate custom Chapman Stick “stafftab” notation.
<janneke>ACTION cries -- thank you
<lyr3>I think I will create some guix custom packages...
<janneke>when this mes/bootstrapping stuff is done, i'd like to create a gui for lilypond that enables mouse-shovers to use lily and inspire them to use their keyboard to enter music
<rekado>janneke: my partner uses frescobaldi as a GUI for lilypond. How would the GUI differ from that offered by frescobaldi?
<snape>rekado: any source and example of that Chapman Stick?
<rekado>snape: erm… you mean… a recording? Or a notation example?
<snape>I meant a pdf with the output and the source :-)
<snape>otherwise nevermind
<snape>I've done some Lilypond a few years ago, but I didn't dig into the Scheme code
<snape>it's definitely great
<rekado>snape: here’s one:
<pkill9>anyone here use weechat?
<rekado>it’s not the most beautiful pdf
<rekado>the stick parts are in stafftab notation. The shapes of the note heads encode the finger.
<rekado>every note also has a string marker that indicates which of the twelve strings the finger should be placed on.
<rekado>the note lines double as string indicators.
<rekado>the numbers above the notes show the frets.
<rekado>the string markers are shifted according to the shape of the fretting hand
<snape>oh, nice!
<rekado> is the source for just the stick part.
<rekado>I have a somewhat complicated makefile to produce full scores like that pdf but also per-instrument scores.
<snape>yes I can imagine. I thought you had written Scheme code, but I misunderstood
<rekado>it’s great for rehearsals (in theory, i.e. if people came prepared ;))
<rekado>no, I did write Scheme code.
<snape>do you play any of those?
<rekado>that’s the code of the custom engraver
<rekado>any of the instruments?
<rekado>I play all of them, but for the recording I’ll just play the stick + sing.
<snape>impressive. You must have spent so much time doing that
<snape>oh you submitted the engraver, of course
<rekado>most of the time is lost on rewrites. I easily get bored by the songs, so this happened very often :-/
<rekado>I’ve got all of the songs for this album as lilypond scores in a git repo.
<snape>and what about the recordings? :-D
<rekado>hah, well…
<rekado>the drummer is preparing to move to the US, but he really really wants to finish laying down the drum tracks before then.
<rekado>don’t know if that will happen, but hard deadlines like this are a blessing.
<rekado>(without them we would never ever get anything done)
<snape>yeah, so hopefully you'll get it done and recorded
<rekado>maybe by the end of this year if all goes well.
<pkill9>i wish i knew why libcurl says no CAfile or certs when strace says it reads it
<snape>got it. I'll check from time to time then :p
<rekado>I’ll certainly announce it there to the two or three occasional readers.
<rekado>pkill9: do you have a test case that I could try to reproduce?
<rekado>pkill9: I assume you have set all the common SSL environment variables?
<pkill9>yeah i have
<janneke>rekado: last time i checked, in frescobaldi you use the keyboard to enter .ly, not the mouse like in commercial programs
<lyr3>cant find cc under guixSD anyone knows which package is to be installed
<janneke>rekado: installing frescobaldi to check, but sadly it's building QT
<pkill9>rekado: i'm writing a command that you can use to see if you can reproduce the issue
<chewzerita>lyr3: clang? gcc?
<rekado>lyr3: there is no “cc” executable, but many makefiles accept CC=gcc
<rekado>lyr3: to use the GCC compiler toolchain you only need the gcc-toolchain package.
<lyr3>rekado: that is it! haha thanks
<rekado>janneke: yes, in frescobaldi you can use either type of keyboard: a computer keyboard or a MIDI keyboard.
<pkill9>rekado: this will make a temporary directory, run weechat with that directory as the session directory, and run '/script' when it runs weechat which on my system gives an error that it cannot find CAfile: mkdir /tmp/weechattest && guix environment --ad-hoc weechat -- weechat --dir /tmp/weechattest --run-command /script
<rekado>pkill9: thanks. Give me a moment and I’ll try it.
<snape>oh I remember you talked about it months ago
<rekado>what is “/script”?
<pkill9>it's for downloading plugin scripts from the weechat website
<rekado>ah, okay
<pkill9>it tries to download a list of plugins when you run it
<rekado>okay, I can reproduce it.
<pkill9>i ran it with strace and it reads the global CAfile successfully
<pkill9>someone here was helping me debug it
<pkill9>some days ago
<rekado>yes, I see in strace that “gnutls_ca_file” is set to a valid file.
<rekado>but that’s for gnutls; it’s not necessarily used by curl.
<rekado>I also don’t see any code in weechat that would configure libcurl to set the CAfile.
<rekado>actually, there is:
<rekado>that’s done #if LIBCURL_VERSION_NUM >= 0x070402 /* 7.4.2 */
<rekado>it seems to me that this is missing:
<rekado>we may need to patch the weechat code in src/core/wee-url.c to make sure that CURLOPT_CAINFO is given a value, preferably the return value of getenv("SSL_CERT_FILE") or similar.
<pkill9>do all programs that use libcurl need to set CURLOPT_CAINFO? could curl specify the default certificates store as /etc/ssl/certs like gnutls does?
<rekado>pkill9: yes, that’s possible, but there are drawbacks to doing this.
<rekado>it also makes it impossible for unprivileged users to change the set of certificates.
<janneke>rekado: ah, midi keyboard -- i managed to run frescobaldi
<pkill9>ah ok
<janneke>the guix i envision would make it possible to click on a staff with your mouse, and enter notes like that
<janneke>i think it's *really* clumsy, but that's how most people create music with a computer
<rekado>ACTION also always writes GUIx when trying to write “GUI” ;)
<rekado>ah, I see.
<janneke>using midi keyboard is probably the best, second best is keyboard; most choose for mouse-shoving
<rekado>I used the mouse long ago before I knew anything about notation.
<janneke>rekado: that's what i think and hope: awareness, awareness, awareness
<rekado>most important was the fact that it played the note as I dragged across the staff.
<janneke>people do and use what they see or are indoctrinated to do
<rekado>I let go of the mouse as soon as it sounded right.
<rekado>but a MIDI keyboard would have been far superior.
<pkill9>rekado: why would setting a default path to CA bundle prevent users settings their own?
<rekado>pkill9: it wouldn’t necessarily do that. But libcurl will just pick the default if none other is specified.
<janneke>not everyone owns one -- i bought one once, pretty expensive but great piano, and it turned out its midi was not compatible with linux
<pkill9>i found a configure option for curl --with-ca-bundle, i'll try inheriting curl and adding that config flag and using that as input for weeechat
<pkill9>i didn't think of that because i was under the impression the CA bundle stuff was being handled by gnutls
<rekado>janneke: yeah, I bought a Doepfer with 88 hammer keys from my music teacher later. It was really lovely, but much too heavy :)
<rekado>now I use a very cheap USB MIDI keyboard.
<pkill9>thank you for helping debug rekado
<rekado>pkill9: you’re welcome!
<rekado>pkill9: I think we should do something about the libcurl defaults.