IRC channel logs


back to list of logs

<doncatnip>magic ... copy pasting the same thing from another vim instance fixed it
<lfam>doncatnip: There must be some mistake in your code. Can you share it?
<lfam>I wonder if your editor actually put the changes on disk
<lfam>I've had issues with Vim that point to something like that. I tell the editor to write the changes, but they don't actually get written to the file.
<doncatnip>aaand its back
<doncatnip>nah its getting written
<lfam>doncatnip: The hash is 1 character too short
<doncatnip>oh man .. did you just count the characters ? ;)
<lfam>The Guile backtrace gave me a hint and then I did `xclip | wc -c` :)
<doncatnip>ERROR: In procedure bytevector-u8-ref: Value out of range: 31
<doncatnip>yep i guess im blind
<lfam>Yup to that error, not that you're blind :)
<doncatnip>figured :)
<lfam>The error message is pretty far from "The hash is the wrong length."
<doncatnip>yea its pretty mean ... such slack .. i should uninstall guix and not bother anymore
<lfam>Also I've seen this one before, since I aliased 'jk' and 'kj' to ESC in Vim. So, there is always a probability that pasting the hashes will not work right :p
<doncatnip>thx btw ;)
<doncatnip>@ build-succeeded \\o/
<jeaye>ACTION uses jk and kj to ESC in vim as well.
<lfam>jeaye: Then you know that sometimes it blows up :)
<jeaye>Haven't had issues with it, aside from when I'm typing something that end with j or k and I want to then quickly exit insert mode.
<lfam>Mode problems
<lfam>You gotta be kidding me! Transmission *again* distributed malware:
<lfam>Can anyone recommend a bittorrent daemon that can serve at least one thousand torrents?
<lfam>I asked their developers more than once to start signing releases, and they ignored my requestes
<doncatnip>tribler, perhaps
<doncatnip>they seem to value security
<doncatnip>although i wouldn't trust the anonymizing features anymore
<lfam>Unfortunately, tribler is not white-listed by the tracker I use
<lfam>Maybe I'll try deluge again. Last time I tried it was a few years ago
<doncatnip>quite a shame transmission-cli used to be my primary choice
<lfam>I didn't upgrade our package since the previous attack. I was waiting for them to start signing releases. Instead, this
<doncatnip>the gnunet guys need to hurry up .. i'd like it to be de-facto standard for file sharing stuff
<lfam>Of course deluge is no better. They give hashes over HTTP
<jeaye>lfam: rtorrent?
<lfam>Yes, maybe rtorrent in screen or tmux
<lfam>Disturbing line to see in a changelog:
<lfam> Build OSX.KeRanger.A ransomware removal into the app
<lfam>Huh? In February transmission announced "bump OpenSSL minimum to v0.9.7", even though 0.9.8 was already end-of-life at the end of 2015
<lfam>Bad choice IMO
<doncatnip>mh one of the reports states it spread via a *signed* client
<lfam>Yes, the OS X signature mechanism
<lfam>So, somebody has control of transmission's OS X key for the special Apple binary signing system.
<lfam>Somebody unauthorized, that is
<lfam>By the way, what is 'mh'?
<lfam>My head?
<doncatnip>hehe if you want it to be ;)
<lfam>As in, "Oh what a headache! My head!"
<doncatnip>i use it in place of hm
<doncatnip>because who doesn't want to be special
<lfam>I certainly don't. And I'm the only one ;)
<doncatnip>not trying to be special is pretty special
<buenouanq>why can't you have both %base-services and %desktop-services? I assume the latter contains them all, but what is actually causing it to break?
<alezost>buenouanq: %desktop-services already contains %base-services, you can't have one service more than once
<civodul>Hello Guix!
<janneke>Hello civodul!
<civodul>hey janneke
<civodul>you reported success about rottlog a while back but i forgot to take notes :-)
<civodul>does it just work out of the box?
<civodul>ooh, good
<civodul>still in my backlog...
<janneke>almost...packdir needs to be set; changing to daily iso weekly is optional
<janneke>...i think, havent tested the weekly only :-)
<brendyn>honestly my locale settings are getting worse and worse the more I try to fix them. Can you believe I'm actually reading the manual now?
<civodul>brendyn: maybe you should send the details of what you have on, and we can investigate from there
<brendyn>I don't really know what to ask
<civodul>just report what doesn't work, and details about your setup
<civodul>ideally the manual would help you, but apparently it doesn't ;-), so we need to investigate
<brendyn>There is a 95% change it's 100% my fault
<brendyn>How come when I r `guix environment --ad-hoc bash --pure -- bash' this command does not appear in my history. i.e, when I press up I get the second last command I ran instead
<efraim>sneek: later tell lfam if you're still looking for a torrent client there's aria2, I have a package for an emacs client that needs testing and there's a web controller also
<sneek>Will do.
<efraim>sneek: botsnack
<brendyn>civodul: I think one issue is that libc is not appending the version number /2.23 as per
<ng0__>hi, a short addition to my email: i asked kalithea and issues tracker is on the roadmap but not being worked on atm
<ng0__>i have some further ideas i'll write when i'm on desktop again
***Gottox_ is now known as Gottox
***frafra_ is now known as frafra
<nckx>brendyn on guix environment: for the same reason that running ‘bash; <up arrow>; exit’ won't show ‘bash’ as the last command. Whatever that reason is. The subshell is launched first, *then* the command is added to the history, effectively losing it. It's up to your shell, not guix.
<brendyn>Some files say glp2+, but underneath the gpl notice is a permissive license notice. How am I supposed to indicate them in licenses? Is it just gpl2+?
<efraim>you can have a list of licenses
<efraim>but if you do we like to know what files have which licenses
<brendyn>Yeah but I don't know what license this file is supposed to be when it's under the gpl but contains a permissive license notice that was required to be preserved
<efraim>figuring out the license is sometimes the hard part
<efraim>post a link, i'll help find something similar
<brendyn>I'm pretty sure it's all libre software
<efraim>civodul: I had an obvious idea about my aarch64 boards: if I can install guix armhf on the board and use it then the kernel is good and my patch is bad
<brendyn>stuff like
<brendyn>nested license
<efraim>the first is definately public domain
<efraim> is strange
<efraim>it looks like gpl2+ and lgpl-$(latest)+
<efraim>i take back the part about lgpl
<efraim>it just says to look at it, not that it is released under it
<brendyn>ok so license:public-domain file://num/NUMrandom.cpp or something like that?
<efraim>license:public-domain ; num/NUMrandom.cpp
<efraim>regularExp.h, the second license looks like bsd-3
<brendyn>Maybe I'll take a look to see how debian packages it
<brendyn>It seems they even have manuals and stuff. Maybe I have much more work to do
<brendyn>Hmm, it would be nice if I could simply convert the debian package definition to a Guix one
<efraim>feh, geeqie and sxiv are all image viewers and have their own files, eye of gnome is in gnome, and I have ephoto and viewnior packaged and looking for a home
<efraim>time for an image-viewer equivelant of pdf.scm
<efraim>either that or viewnior goes in viewnior.scm and ephoto goes in enlightenment.scm
<brendyn>Is anyone working on a guix import debian ?
<civodul>efraim: your patch is bad?
<civodul>brendyn: ISTR someone (Alex Vong? Hartmut?) wrote an importer in Python
<civodul>but nothing integrated in 'guix import' currently
<civodul>so your help would be welcome :-)
<kalashkash>hi guys!
<brendyn>civodul: I just found that :) I came to realise that debian packages have already packaged praat much better than I have, so it's a shame to have to repeat everything
<civodul>kalashkash: & gals! :-)
<civodul>i think time has come to add "guile22-" variants of our "guile-" packages
<paroneayea>civodul: ooh
<davexunit>I think you are right
<paroneayea>guile 22 already!
<paroneayea>what a jump :)
<davexunit>yeah would the prefix be "guile22-" or "guile2.2-"
<civodul>i thought "guile22-" but we can probably safely use "guile2.2-"
<civodul>preferences? :-)
<davexunit>slight preference for 2.2 because it's more clear
<dvc>mark_weaver: I tried epiphany, but it doesn't play back html5 video...
<dvc>hopefully bavier will get icecat 45 beta working soon... :)
<wingo>epiphany can do html5 video
<wingo>if it can't do it in guix, that's a guix problem :)
<davexunit>dvc: you were trying to play a webm or something like an mp4?
<davexunit>webms ought to work fine
<davexunit>mp4s don't work for me in icecat
<davexunit>presumably for free software reasons
<davexunit>so perhaps it's the same for epiphany
<kete>yeah, they don't work in tor browser either.
<brendyn>debian seems to have a standard format for specifying license information with a lot of detail
<brendyn>Do we care much about that?
<brendyn>Decisive response...
<davexunit>packages have a license field
<davexunit>and that is enough
<brendyn>haha, ok
<kete>I think mp4s work on
<kete>if you have enough javascript enabled
<kete>i.e., nonfree software
<davexunit>I enable javascript but my browser still does not play them
<efraim>civodul: I haven't unpacked my arm64 boards yet but I have a workdesk now where I can leave them out
<janneke>anyone has adb running for their cm/android phone?
<dvc>Yes, seems to be an issue with mp4...
<dvc>wingo: Did you see the issue I opened on the elogind project in regard to weston? WDYT?
<wingo>i haven't looked at it
<wingo>no idea really
<htgoebel>What is the best way to find out which license is the one used in some license-file?
<htgoebel>I have one containing only a copyright note and a disclaimer like the original BSD license. But no rules like the BSD license, only the disclaimer.
<htgoebel>How do i find the correct identifier for the `license` entry?
<wingo>i think the answer would be to fix sd_session_get_vt, perhaps it needs to do something else
<dvc>currently it reads the VTNR environment variable
<dvc>Maybe it isn't set because "elogind doesn't manage vt's"
<bavier>dvc: did you get any further on the icecat beta build?
<dvc>bavier: I quit after you said you where working on it and mark suggested an alternative
<bavier>dvc: ok, np
<brendyn>Is there any documentation on substitute* ?
<brendyn>both geiser and don't show anything
<bavier>brendyn: check the source in guix/build/utils.scm
<brendyn>bavier: There is not such file?
***joachifm_ is now known as joachifm
<brendyn>Hmm why isn't that in my cloned repo?
<brendyn>ok so for some reason my git repo only has two directories, build/ and src/
<jlicht>hello guix!
<brendyn>I wonder if I should be messing with build flags?
<davexunit>brendyn: sounds like you've cloned the wrong repo
<brendyn>davexunit: well I deleted everything and started again
<brendyn>Debian adds all these flags like "g -O2 -fstack-protector-strong -Wformat -Werror=format-security" should I be worrying about these?
<brendyn>There was discussion about hardening on the guix mailing list
<davexunit>I don't understand where this is coming from
<davexunit>are you saying that debian's gcc automatically adds these flags?
<davexunit>is this with regards to building the guix daemon?
<brendyn>Sorry I think I was wrong
<brendyn>I mean, the debian version of praat has quite a lot of changes in debian
<brendyn>little things like hardening=+all
<davexunit>we try to do the minimal number of changes required to make the software work
<brendyn>Is it possible to build a good distro that way?
<davexunit>seems like a loaded question
<brendyn>Not for me. I've never build a distro before
<davexunit>would you consider parabola or arch a good distro?
<brendyn>Well, I use parabola
<davexunit>part of their methodology is also to do minimal patching
<davexunit>our feeling is that rather than holding onto a ton of patches, we should reduce the number of patches we need by contributing upstream.
<brendyn>I agree, but it makes me wonder why debian isn't the same
<davexunit>different priorities, I guess.
<brendyn>I would like to make a build twice and then compare the results. Do you have any tips for that ? I ran guix build rounds=2 but only one build was made in /tmp. I'm not sure it even built twice
<kete>some maintainers and users think it's necessary to make the software work together in a system. (stability) I hate it
<brendyn>My personal goal is to build a distro that Can be installed in schools
<davexunit>brendyn: --rounds is the thing, like you've already found
<davexunit>you should be able to use diffoscope on builds that differ
<davexunit>the manual should hopefully have info about how to locate both builds
<davexunit>kete: that's the whole point of a distro, though. :)
<davexunit>we're also trying to make a whole bunch of software work together in a system
<janneke>davexunit: i'll have a stab at building android-tools
<brendyn>Well the output of diffoscope makes no sense to me
<kete>once, I read Apache's ~how to undo what Debian did~ or "Debork"
<jmd>Whilst running guix system reconfigure, I saw:
<brendyn>haha, diffoscope just showed me I'm stupid and my patch didn't work :)
<jmd>rafting '/gnu/store/alpwlmp3hrb2fnn26dysfmavdymidr2d-libcanberra-0.30' -> '/gnu/store/xn2j8qgj05vhdzm4xdxz5ps8269v4963-libcanberra-0.30'...
<jmd>** (inkscape:4): WARNING **: Unable to create profile directory (Permission denied) (13)
<jmd>** Message: Cannot create profile directory /homeless-shelter/.config/inkscape.
<jmd>** Message: Inkscape will run with default settings, and new settings will not be saved.
<jmd>** (inkscape:4): WARNING **: Could not create extension error log file '/homeless-shelter/.config/inkscape/extension-errors.log'
<jmd>Background RRGGBBAA: ffffff00
<jmd>Area 0:0:1024:768 exported to 1024 x 768 pixels (90 dpi)
<jmd>Bitmap saved as: /gnu/store/fxwhblb2ql9q75m1vy6x7r2fphrd18dl-grub-image.png
<jmd>Oh dear. and later it bombs out altogether.
***detrout is now known as detrout_
<jmd>Anyone have any ideas:
<doncatnip>ohman didn't realize the lua dynamic lib support issue has been going on for a year now ... discouraging
<brendyn>How can I test out a (substitute* ...) in a repl?
<davexunit>by evaluating it?
<davexunit>I guess I don't understand the question
<brendyn>but substitute* doesn't seem to exist even after I run ,use(guix)
<davexunit>that's because the (guix) module doesn't export it
<davexunit>it's in (guix build utils)
<brendyn>Can you see what is wrong with the substitute* here ?
<davexunit>brendyn: ("")
<davexunit>the parens mean "call a procedure"
<davexunit>and the "procedure" is the empty string
<doncatnip>i was about to write that .. nice, progress, by tomorrow i'm guile expert
<brendyn>I'm currently an expert in nothing.
<davexunit>you'll pick it up
<doncatnip>neither am i
<doncatnip>yea its not too bad after the initial scare
<davexunit>there's really only a few basic rules
<davexunit>if you can get the hang of those, you can learn everything else.
<brendyn>It's hard to ask help because I'll be stuck on something for house and people are like "RTFM"
<doncatnip>welp im coming from python so those parentheses are quite intemidating
<brendyn>I think the idea is to pretend the parens don't exist
<davexunit>generally people help out a lot here
<davexunit>I don't think there's much RTFMing going on
<brendyn>jmd: inscape works for me
<davexunit>I do think there's lots of questions asked that people could answer themselves, and I'd like to focus on teaching to fish than rather than just give them a fish.
<doncatnip>still can't really make sense out of the use of ` and #:constructs
<doncatnip>but i didnt read nothing yet
<davexunit>#:foo is just syntax for a keyword
<davexunit>in this case, the keyword named "foo"
<davexunit>similar to how 'foo is the symbol named "foo"
<davexunit>` is the "quasiquote" operator. to understand it, you should first be familar with the ' operator called "quote"
<davexunit>are you familar with quote?
<doncatnip>like filling kwargs ?
<brendyn>Well the fix also didn't work.
<brendyn>Let me see...
<davexunit>doncatnip: is this about the keyword syntax?
<davexunit>keywords are used for procedures with keyword arguments, yes.
<davexunit>but they are a first-class data type that you can use for whatever reason you'd like
<doncatnip>alright easy
<davexunit>the 'define*' and 'lambda*' macros allow you to make procedures that have optional or keyword arguments
<brendyn>Maybe I shouldn't have this in the source part
<brendyn>cd-hit has a similar thing, but it's in modify-phases
<davexunit>brendyn: snippets are used for things like removing non-free parts of source releases
<davexunit>so that the source code that the user sees is only free software
<davexunit>this feature was implemented in order to meet the FSF's free distribution guidelines
<brendyn>Man this is annoying. I'm still getting the date printed in the output
<brendyn>and it's about 10 minutes to test it and rebuild. guile promised me a repl :(
<davexunit>you have a repl
<brendyn>But I can't use it
<davexunit>why not?
<brendyn>I don't know how
<brendyn>I don't even know what the working directory is or if guile even has a working directory
<davexunit>I told you how to import the module that has 'substitute*' at the repl
<davexunit>from there you can test your substitution on a temporary file
<davexunit>run (getcwd) to get the current working directory
<davexunit>from your guix git checkout, run './pre-inst-env guile' to start a REPL
<davexunit>a REPL that has all of the guix modules on its load path
<davexunit>brendyn: did you import (guix build utils)?
<davexunit>I can evaluate that code at my repl once I import that module
<brendyn>ok I ran ,use(guix build utils)
<brendyn>And I get the same error
<efraim>I just realized we don't have cowsay
<brendyn>Maybe prepending / doesn't make it start from /
<davexunit>brendyn: is it *really* the same error? it shouldn't be a syntax error
<davexunit>I get file not found, because of course I don't have such a file on my system
<davexunit>which is what I expected
<davexunit>but the macro itself expanded just fine
<brendyn>Maybe it's because I pasted the definitions from utils.scm into the repl and they errored
<brendyn>so I've overridden everything with borked code
<jmd>brendyn: If you're in tas, isn't it rather early in the morning for you?
<brendyn>let me start a new repl
<brendyn>jmd: 3 am
<jmd>time to feed the echidnas I think.
<brendyn>I ate them
<brendyn>I had a kangaroo burger yesterday actually
<jmd>roos are over the whole continent. I thought there you ate the tassy devils.
<brendyn>Now what I suspect is that the string of text I've got is somehow incorrect
<brendyn>tassy devils are all dying of cancer so there are efforts to preserve them
<brendyn>probably they will eat you first anyway
<jmd>like the tassies tiger?
<brendyn>No we shot all the tassie tigers, and all the aboriginals
<brendyn>I'm not sure there are many other places in the world to actually complete a genocide
<brendyn>Ok so the string I've provided in the regex doesn't actually match the line in the comment
<brendyn>Can you see why? I can see how I've missed a space or anything
<brendyn>I tested it with just "TIME" and it works
<brendyn>_ aren't special characters? no missing space? I can't see it
<brendyn>Ok, I think ( need to be proceeded by \\\\
<brendyn>But why?
<jmd>That's the way it is. They have special meaning unless escaped.
<brendyn>Ah yes, ofcourse.
<jmd>Somewhere, I think there is a version of substitute which interprets all characters literally.
<jmd>Unfortunately I cannot remember what it is called.
<brendyn>It's a wonder we aren't using Scheme regexp
<brendyn>Amazing, It was actually built reproducibly! Guix should congratulate the user when this happens
<davexunit>brendyn: congrats!
<davexunit>we should have 'guix build' play a little jingle or open up a gif in your web browser
<davexunit>like this one
<brendyn>davexunit: I sent the updated patch to the mailing list :)
<brendyn>davexunit: Can you tell me if the patch is malformed again?
<davexunit>sorry I can't right now
<paroneayea>looks like it's time to get out another CVE fix, or we'll have *no* secure GUI browsers
<paroneayea>mark_weaver: ^
<paroneayea>remote code execution in webkit
<paroneayea>ugh, I'm a moron
<paroneayea>we're fine, sorry :P
<paroneayea>ACTION sits in corner with dunce cap on
<janneke>ACTION sent a package/hack to build adb to -devel
<cbaines>I'm using Notmuch in Emacs on GuixSD, and getting some issues with font rendering as I'm seeing the unicode replacement character in the tree view, does anyone have any advice?
<paroneayea> shows how to use DNS rebinding and cross-protocol scripting to use a developer's browser to take or modify their local data (basically: rebind to be, and that "runs on localhost only" unauthenticated HTTP API you have can now be snarfed)
<paroneayea>could be also interesting if you know someone is using %facebook-host-aliases :)
<paroneayea>in guix
<paroneayea>davexunit: you may enjoy the above :)
<davexunit>and scary
<doncatnip><3 guix system vm
<doncatnip>such beauty
<efraim>i just realized I'd been sitting on a patch to package python-requests-mock, and all I had to do was upgrade the package in openstack.scm
***malthe_ is now known as malthe
<kete>cbaines, have you installed a font? I haven't done much since installing GuixSD, and I think I need to install a font.
<cbaines>kete, I have a few fonts installed, and haven't configured emacs (to my knowledge) to use a specific one
<kete>do you need to ask #emacs?
<davexunit>bummer, a piece of software I packaged has changed from GPLv2 to a proprietary license.
<davexunit>I don't even know if they have the legal authority to do it (maybe there's a CLA, dunno), but they did.
<sneek>Welcome back lfam, you have 1 message.
<sneek>lfam, efraim says: if you're still looking for a torrent client there's aria2, I have a package for an emacs client that needs testing and there's a web controller also
<lfam>davexunit: Which one, out of curiosity?
<rekado>davexunit: fork it?
<davexunit>lfam: aseprite
<davexunit>rekado: a fork has been made, it seems.
<davexunit>sad article
<davexunit>"GPL is about “giving everything to the user” and asking for nothing."
<davexunit>"That kind of mentality might contribute to the destruction of the software"
<lfam>I don't know if they always had a CLA, but they do now:
<davexunit>lfam: looks like they did
<davexunit>I didn't realize that until now
<davexunit>oh well, I just use the gimp for pixel art.
<davexunit>aseprite had some neat features, but lacked some things I took for granted with gimp
<lfam>LOL. They blame Heartbleed on free licenses
<lfam>Great scare tactic
<davexunit>it's a shame.
<lfam>I don't understand that part of their argument. But, I'm prejudiced against their conclusion, so I can't really give a fair reading. And, as they say, English is not their first language.
<lfam>Oh well
<davexunit>they seem to have gotten frustrated with distros packaging aseprite and circumventing the paywall for binaries
<lfam>Our package is still a few versions behind ;)
<davexunit>I packaged it awhile ago, yeah.
<davexunit>I have some qualms about all the bundled software aseprite uses, too.
<davexunit>I unbundled a good amount, but they bundled and forked allegro 4
<davexunit>and I couldn't fix that
<lfam>Maybe it's time to donate to GIMP what Aseprite wants to charge ;)
<davexunit>I would like it if I could fund specific features of software
<lfam>My turn, too. I've used GIMP at work a few times recently
<rekado>I wonder if it would be a good idea or a bad idea if distributions provided a way for users to financially support software authors.
<davexunit>like I would be happy to fund someone that wanted to add more tools for pixel artists to gimp
<lfam>rekado: Hm, tricky to implement but, in general, I think it's a good idea
<davexunit>rekado: someone brought that up here recently.
<davexunit>I think it might be nice, if presented in the right way.
<lfam>But, I have a big bias in favor of commercial support for free software. I think it's essential in the long run.
<lfam>Not that I have much evidence or experience to back up this bias ;)
<rekado>lfam: I find it puzzling that there aren’t more companies that offer support for selected free software applications.
<davexunit>for me, rather than just donate money into the "whatever" fund, I'm more interested in projects that present certain goals or features and ask for money to get them done.
<lfam>davexunit: Seems like a good way to rustle up support
<rekado>davexunit: I find this really difficult. It doesn’t always work this way.
<rekado>ardour, for example, has something like that.
<lfam>One could even hire a developer, with the blessing of the project maintainers, to work on a feature
<rekado>people would pledge to donate money if MIDI support was added.
<davexunit>like with guix I'm happy to give money for the purpose of growing the build farm
<rekado>this took several years and the total amount was tiny.
<davexunit>I don't know what the right way to do it is exactly
<lfam>rekado: So, it was not that successful?
<rekado>the problem with funding is that it’s usually not enough to let a developer rely on it.
<rekado>lfam: no, not really.
<lfam>Speaking of our build farm, what is the status of the new machine?
<davexunit>rekado: yes, that's true.
<rekado>Paul usually struggles to get enough donations / subscription money, with little spikes after releases.
<davexunit>which is why I don't free software full time
<davexunit>don't do*
<davexunit>I've seen a "subscription" model work OK
<rekado>I do free software full time, but I admit it’s an unusual position.
<rekado>I once had high hopes for flattr.
<lfam>Even if the new machine is still not online, I think we should make a public update soon. I'd hate for donors to be wondering "What happened to my money?"
<davexunit>there's a pool of people that continually sustain the software
<davexunit>lfam: yeah, probably a good idea.
<davexunit>if for nothing else than transparency.
<rekado>lfam: agreed. I don’t know what the current status is, but I haven’t yet see the end of the long list of unread email…
<lfam>"Where are my millions?" ;)
<davexunit>gotta run
<davexunit>happy hacking everyone
<doncatnip>can anyone confirm gdk-pixbuf failing with rounds=2 in current master ?
<lfam>doncatnip: If it fails rounds=2 for you, it almost certainly fails rounds=2 for everybody
<lfam>For now, reproducible packages are a very good thing, but not required. We don't have the infrastructure to track the reproducibility state of packages yet
<doncatnip>right .. well, i didn't break it ;) but i guess you should know
<lfam>I guess we'd need to make the build farm build everything twice, and add some fields to the database to track the results
<lfam>And add another "fail" state that is different from a regular build failure
<lfam>And, we would rather move to another build farm system entirely, rather than add feature to Hydra. Looking forward to Cuirass:
<lfam>Oh, notabug is offline. Save that link :)
<doncatnip>wut hydra is such a cool name though
<lfam>Yes, but it seems that none of us want to work on it
<lfam>Wrong language ;)
<doncatnip>cuirass is scheme then ?
<lfam>Yes. It was a Google Summer of Code project this year. The source repo may be offline, but there is some good discussion on the guix-devel archives, if you are interested.
<doncatnip>interesting indeed
<doncatnip>so are many things
<doncatnip>i need 48 hour days
<lfam>Me too
<efraim>24*7 = 28 * 6 = 20 * 8, go ahead and choose how long you want to sleep each "day"
<efraim>wait, 21*8
<doncatnip>i was thinking about trying polyphasic sleep
<doncatnip>but its pretty incompatible to the rest of the world
<lfam>I think you really can't have a job and do the crazy sleep systems
<civodul>ACTION just discovered
<civodul>funny name, but the thing is super well written!
<efraim>3 libidn CVEs, not sure if we're affected
<efraim>master has 1.32, 1.33 is safe according to debian
<civodul>'guix lint -c cve libidn' has nothing to say :-/
<efraim> doesn't have anything to say either
<efraim>i haven't checked mitre but i assume they're not saying anything yet either
<civodul>right, still marked as reserved:
<efraim>i'm guessing this is one of them
<efraim>from the 20th
<efraim> this might be another
<efraim>might just need to keep an eye on it for a few days and see where we stand
<mark_weaver>dvc: to support mp4 files, you need to install gst-libav
<mark_weaver>epiphany absolutely supports playing videos, including mp4 videos. I do it regularly.
<mark_weaver>dvc: you'll also need to export GST_PLUGIN_PATH=$HOME/.guix-profile/lib/gstreamer-1.0
<civodul>we should modify the GNU maintainer's guide to suggests maintainers email us about security issues...
<mark_weaver>(although that should be handled automatically on GuixSD)
<bavier>efraim: gsl 2.2.1 was released, if you update again, could you check the build for i686?
<mark_weaver>ah yes, that would be great
<bavier>I noticed the 2.2 version ftbfs for i686, which prevents system reconfiguration
<bavier>since inkscape needs it
<civodul>we really need to avoid inkscape there
<bavier>that would be nice, yes
<bavier>are there other options?
<civodul>rsvg-convert, from librsvg
<civodul>this was discussed a while back with mark_weaver
<mark_weaver>I tried to do that, but it didn't work
<civodul>because of the checkerboard pattern, right?
<civodul>but then we got a new SVG that should be convertable, IIRC
<mark_weaver>yeah, although it made a SVG quite large, and it just didn't seem like a great solution.
<mark_weaver>but if you want to switch to the new SVG that he hacked up, I'm okay with that. I don't feel strongly.
<civodul>yeah, it might be a tradeoff we have to make
<mark_weaver>why do we have to make it, though?
<mark_weaver>I would prefer to avoid inkscape, but I don't feel compelled to do so.
<civodul>ah, sure
<civodul>it's only a problem when binaries are missing, or in similar situations
<mark_weaver>if inkscape doesn't build on any of our platforms, that's a problem that should be addressed, independent of its use generating the grub background.
<dvc>mark_weaver: Thanks, I'll try that.
<dvc>mark_weaver: Is js disabled by default on epiphany guix?
<mark_weaver>there's been a beta release of icecat 45, but there are no digital signatures on either the source or binary releases :-(
<mark_weaver>given this, I'm not comfortable even testing it
<bavier>I tried building the icecat beta, but hit a configuration error regarding the branding
<bavier>I haven't looked into it more
<dvc>So somethings wrong. I'm getting typeerror blah is undefined in the developer console.
<mark_weaver>such a vague problem report is not something I can do anything with
<bavier>I didn't mean it as a bug report, just a status update
<mark_weaver>sorry, that was directed at dvc
<bavier>ah, I see
<mark_weaver>bavier: thanks for working on it. I myself am not willing to touch it until either I have a signed source tarball or the updated icecat script is pushed to the savannah gnuzilla repo. at this point, all we have is a huge unsigned source tarball.
<dvc>for example youtube does that work for you? (after following your steps above)
<bavier>mark_weaver: understood
<civodul>bah, sad state of affairs
<mark_weaver>dvc: yes, youtube works for me
<mark_weaver>in epiphany
<dvc>TypeError: this.b.load is not a function. (In 'this.b.load()', 'this.b.load' is undefined)
<dvc>that's what I get...
<mark_weaver>dvc: earlier, you mentioned that html5 video didn't work. so I guess that indicates that epiphany mostly works? or mostly worked in the past?
<mark_weaver>at what point does that error occur?
<mark_weaver>and what, if anything, works properly with epiphany for you?
<dvc>well I haven't used it that much yet.
<dvc>it shows my gmail email and wikipedia/stackoverflow...
<dvc>I get the error whenever I press the play button on youtube
<mark_weaver>does anyone else see this problem? it works for me.
<bavier>I haven't seen it
<mark_weaver>dvc: is this on GuixSD, or Guix running on top of another distro?
<dvc>on nixos
<dvc>so that might be the problem - I've had glitches before...
<mark_weaver>oh, well, that might well be the issue
<mark_weaver>epiphany, like most gnome apps, depends upon various daemons and services. in your case, those daemons and services are from NixOS
<mark_weaver>(I guess)
<mark_weaver>maybe NixOS is on a different major version of GNOME than we are. the version mismatch might be bad.
<dvc>ok, thanks
<dvc>so what's our release cycle like? when is core-updates getting merged? :)
<civodul>bavier, mark_weaver: any ideas regarding argv[0] in ?
<mark_weaver>dvc: there's a nasty blocker in core-updates: newer versions of gcc (even 4.9.4) are failing to bootstrap on armhf.
<mark_weaver>the stage 2 and stage 3 binaries differ
<dvc>ah yes you told me about it before... or did you get further on the issue?
<mark_weaver>not yet
<bavier>civodul: I saw the -real behavior with a zsh script I packaged recently
<mark_weaver>civodul: hmm, I'm ignorant on that issue!
<civodul>i'm discovering i'm more ignorant than i thought actually ;-)
<civodul>i thought the only options to find the program name were argv[0] and /proc
<civodul>but it seems python uses something else
<mark_weaver>civodul: maybe /proc/self/exe ?
<mark_weaver>which is a symlink
<civodul>yeah, but it doesn't use /proc at all AFAICS
<mark_weaver>strace would show it
<efraim>gsl-2.2.1 failed to build for me for i686
<mark_weaver>efraim: thanks for trying. bah :-(
<bavier>efraim: for me it was a numerical inaccuracy issue in the cholesky test suite
<efraim>i sent it through a second time, don't remember the issue mine had
<bavier>I didn't have time to look into how the tests compare expected vs actual results
<mark_weaver>I looked, and the differences were quite significant.
<bavier>mark_weaver: on my machine, the differences where very small, less than 10e-8
<bavier>mark_weaver: oh cool, at least our "wrong" answers are the same ;)
<efraim>oh, mine failed on core-updates, not sure why I tried it there
<mark_weaver>one of those numbers are off by about 1e-6
<efraim>ACTION heads off to bed