IRC channel logs

2015-09-29.log

back to list of logs

<goglosh>hai
<goglosh>what is the package I need to connect wpa?
<goglosh>tried guix package -i wpa_supplicant
<goglosh>ooh... nvm
<goglosh>okay now I do have trouble. installed wpa-supplicant but 'command not found'
<goglosh>ACTION comes in breathing heavily
<goglosh>I managed (don't ask me how) to connect wireless
<goglosh>but this thing won't resolve addresses
<goglosh>oh god I made it. feels good. sorry about asking prematurely, leaving, and then just, figuring it out
<civodul>Hello Guix!
<efraim>hi!
<AndChat404481>Hi
<mongrol>r
<AndChat404481>Hi mongrol are you guix developer ?
<mongrol>no
<rekado>Hi AndChat404481
<Steap>AndChat404481: people usually idle here, so don't ask to ask
<AndChat404481>hi rekado hi steap and other, sorry I havent changed my hostname yet. So is there any guix developer online here now ?
<rekado>AndChat404481: I guess you could call me a Guix developer. Do you have questions about the projects?
<alezost>AndChat404481: just ask, if you have a question
<rekado>(or are you just checking if people are online...? ;))
<AndChat404481>I have many question to ask. Just wait a 5 minute more, I must go to toilet.
<civodul>interesting
<AndChat|404481>So there, Im just a intermediate linux user but Im are big fan of fsf. So I heard there is guixhurd under development, is that true ?
<rekado>AndChat|404481: yes, Guix is being ported to the Hurd.
<AndChat|404481>And will it become guixhurdsd in the future ?
<rekado>eventually, you will be able to run a GuixSD system using the Hurd instead of Linux.
<AndChat|404481>Will guix have a gui in the future or next release ?
***yang_ is now known as yang
<rekado>AndChat|404481: do you mean a GUI for the package manager?
<rekado>we already have an Emacs interface in addition to the command line interface.
<rekado>there also is a minimal web interface, but it provides access to only a tiny subset of all the features of the package manager.
<mkletzan>Yaay! \\o/ I managed to install guix ;-)emacs
<mkletzan>and also type "emacs" in ERC window instead the terminal apparently
<taylanub>:)
<taylanub>ACTION can't wait to use GNU Guix/Hurd with GNU Guile/Emacs running on it :P
<mkletzan>unfortunately it starts into window maker only and I have no idea where the configs are so I can debug why it doesn't start into ratpoison
<rekado>mkletzan: there's an odd bug that changes the session to windowmaker when you mistype the password at the login screen.
<mkletzan>rekado: oh, that might've happened, I forgot to set my passwords the first time
<iyzsong>WindowMaker is used by the 'slim-service' as fallback session, you can override it in the system config or write a ~/.xsession script.
<rekado>iyzsong: can it be removed completely? I find it odd that I have WindowMaker even though I didn't ask for it in my operating-system expression.
<mkletzan>it works now! thanks, that's great
<iyzsong>rekado: well, it's also (mainly) the auto-login command, slim don't choose a default session for auto login, so we have to provide one.
<efraim>gnunet doesn't build because gnurl doesn't build
<efraim>currently gnurl is at 7.37.0 (?), and 7.40.0 fails on the same reason (test 46)
<efraim>gnurl 7.43.0 is missing something so it doesn't compile to begin with
<efraim>gnurl inherits from curl the ability to skip the one test, but I haven't gotten it to work right for me, and disabling all the tests just seems wrong
<efraim>i guess i could leave in the logic and a note for it to be fixed later
<mark_weaver>efraim: gnurl 7.44.0 is needed, I think.
<mark_weaver>after a recent upgrade of something else, curl 7.43.0 failed to build but curl 7.44.0 succeeded. I've forgotten the details.
<efraim>7.43.0 complained about a missing install.sh/install-sh file
<efraim>so that seemed more like a whoops error
<mark_weaver>I have to go now, but I can help with this within a few hours
<mark_weaver>ACTION goes afk
<efraim>thanks mark_weaver
<efraim>here's what I have atm: https://pastee.org/wapwn with lines 42-45 added by me
<mark_weaver>efraim: It should be (zero? (system* ...)) instead of just (system* ...), otherwise errors won't be detected.
<mark_weaver>also, since 'outputs' is not used in the new check phase, you can change "lambda* (#:key ...)" to just "lambda _"
<efraim>thanks
<mark_weaver>efraim: but also, we really need 7.43.0, because 7.40.0 has security flaws
<mark_weaver>(iirc)
<efraim>so i guess i could take a diff of curl 7.43.0 to 7.44.0 and check the gnurl 7.43.0 diff provided and try to meld them
<mark_weaver>efraim: there's a new tarball for gnurl 7.43.0, isn't there?
<efraim>yeah
<efraim>but it fails to build
<mark_weaver>okay. taking the curl-7.44.0 diff and trying to apply it to gnurl-7.43.0 sounds good to me
<mark_weaver>ACTION goes afk again
<civodul>GNUnet doesn't build i'll talk to grothoff since he's around :-)
***francis7 is now known as francis7-fsf30
***francis7-fsf30 is now known as francis7
<paroneayea>wtf we don't have nethack packaged???
<paroneayea>I must rectify this, I can't give a demo without showing nethack
<paroneayea>what kind of distro would we be :)
<daviid>:)
<goglosh>I'm going to run `guix system reconfigure` but I badly need wpa-supplicant to come in there, how should I declare it in the system file?
<lfam>I see the dmd manual was last updated 2014-07-08. Have there been changes to dmd since then? I'm planning to install GuixSD soon and I'd like to read the docs in advance.
<davexunit>lfam: nothing major AFAIK
<lfam>I know that GuixSD is "alpha" but how much pain am I really in for?
<lfam>on a scale of Ubuntu to LFS
<davexunit>only one way to find out.
<lfam>Fair enough!
<davexunit>I use it daily.
<davexunit>I wish we had GNOME, but things work reliably.
<davexunit>I'd put it more in the "Parabola" category.
<lfam>That would be good for getting users but I won't miss it (never used it)
<taylanub>given how much pain I've had with the Ubuntu installation on a family laptop, your scale is pretty narrow :P
<lfam>Haha, fair enough. You know what I meant ;)
<lfam>On the right hardware everything works out of the box
***Gonbe is now known as _`_
<civodul>hey, remi`bd!
<civodul>remi`bd: what are you up to?
<efraim>i see that graphviz can be updated from 2.28.0 to 2.38.0 but a lot of stuff depends on it, including guix
<civodul>efraim: yes but it has to be updated someday anyway :-)
<civodul>so it's fine
<civodul>110 packages depend on it apparently
<civodul>it would be great if at the same time you prepared a second patch that moves its doc to a separate "doc" output
<civodul>since the doc represents more than half of its size
<efraim>a bit more work than "find stuff I have installed so I can update it" but sure
<efraim>:)
<civodul>yeah well, that's bonus :-)
<efraim>hmm, just finished building, it looks like /bin is 60MB and /share is 6.1MB
<civodul>oh?
<civodul>i had noted 10MB for all of Graphviz
<civodul>according to "du -msc"
<efraim>20M on /bin/cluster, 20M on /bin/edgepaint
<efraim>and 20 on /bin/gvmap
<efraim>maybe they're new?
<civodul>edgepaint is new but not the others
<civodul>all of bin/ is less than 2M her
<civodul>*here
<efraim>its almost 11pm here, i'll take another look at it tomorrow
<civodul>sure!
<alezost>almost 11pm here as well :-)
<civodul>heh
<paroneayea>okay, I'm finally feeling excited about my talk
<paroneayea>I think I also included the most boiled down intro to functional programing in here, since I know many people in the audience aren't familiar, and use that to guide towards "so how to we apply this to packaging?"
<davexunit>paroneayea: are we going to get a sneak preview? :)
<paroneayea>davexunit: hm, sure! It's still not done
<paroneayea>also, I decided to throw out "proper presentation" and I'm just going to unfold an orgmode document as I go :)
<davexunit>:)
<paroneayea>I was beating my head against trying to do a nice presentation and I decided, screw it, this is a moderately technical talk, people can survive me unfolding an orgmode tree as a presentation format :)
<civodul>sounds reasonable :-)
<paroneayea> http://dustycloud.org/tmp/guix_talk.org http://dustycloud.org/tmp/guix_talk.html
<paroneayea>there's totally incomplete sections still
<paroneayea>but I finally am in the right groove of writing this thing out
<civodul>yeah nice!
<civodul>i guess you can probably demo the command-line stuff
<civodul>or M-x guix if you have the appropriate audience ;-)
<paroneayea>civodul: I will probably do a demo of M-x guix-all-available-packages and M-x guix-generations
<paroneayea>but I will probably not do M-x guix because I just tried it today and I don't get how to use it yet
<paroneayea>so many options!
<civodul>right :-)
<civodul>i was thinking more of what you suggests actually
<civodul>did you know you can diff generations in M-x guix-generations?
<paroneayea>civodul: yes, and I'm totally demo'ing that
<paroneayea>one of my fave features
<civodul>good :-)
<paroneayea>I wonder if the "git-like" part of my talk is a bit too much detail
<paroneayea>I'm going to give a test demonstration of the talk to my brother tonight so we'\\ll see
<civodul>heh
<paroneayea>I think it's an easy idea for many audience members to catch on to though
<paroneayea>most devs know git now, and it makes the hashes seem familiar
<paroneayea>plus rolling back in git is an easy thing to understand as "whoa cool"
<civodul>yeah
<civodul>though the store is not content-addressed
<paroneayea>that's true
<paroneayea>it's a hash of inputs
<civodul>right
<paroneayea>I'll remove the "content addressed" mention
<civodul>often i talk of the "Corresponding Source" as the GPL defines it
<civodul>just to say that here, we really have it all :-)
<paroneayea> - We refer to commits by their hashes... we'll see more hashes in
<paroneayea> Guix for similar-ish reasons...
<paroneayea>maybe a bit closer to making the hashes familiar without explaining why
<civodul>yes
<davexunit>civodul: hmm, I always considered the store to be content addressed
<paroneayea>I just realized:
<davexunit>where that "content" was the full dependency graph of the derivation
<paroneayea>the store is all memoization :)
<civodul>exactly
<davexunit>yup
<civodul>davexunit: it's not content-addressed in the sense that the hashes are not hashes of the contents
<davexunit>yes, that's true.
<paroneayea>so, kinda close
<paroneayea>in a sense, neither are git hashes though
<civodul>if you look at Eelco's thesis, he did try content-addressing too
<civodul>only as a PoC though
<paroneayea>they aren't hashes of the actual files being revisioned (though they are of the "objects")
<paroneayea>so, similarly vaguely "nearly but not really CAS" :)
<davexunit>paroneayea: "I've fallen and I can't get out!" hahahahaha
<paroneayea>FAIAS? (Function And Input Addressed Storage)
<davexunit>LifeAlert + Turing tarpit humor.
<davexunit>Memoized Storage
<paroneayea>memoized storage, yeah!
<civodul>aka. "cache" :-)
<davexunit>;)
<davexunit>I distinguish memoization from the traditional notion of a cache.
<civodul>yeah
<davexunit>but often caches can be invalidated later
<davexunit>what was true then may not be true now
<davexunit>not so for a pure function!
<davexunit>s/but/because/
<civodul>good point!
<davexunit>and I see that paroneayea mentions this point in his draft
<davexunit>"Memoization means caching with no cache invalidation challenges"
<civodul>mark_weaver: do you have a patch for gnurl?
<civodul>since you discussed it this morning
<mark_weaver>civodul: not yet
<civodul>ok
<mark_weaver>I can look into it later tonight
<mark_weaver>I'd been hoping that gnurl-7.44.0 would be released, but I guess we'll have to figure something else out in the meantime.
<mark_weaver>lfam: btw, civodul recently added the ability to clear entries from the failed builds cache.
<civodul>mark_weaver: i'll see if we can get gnurl released ;-)
<mark_weaver>that would be helpful :)
<civodul>yup!
<mark_weaver>lfam: if you run "guix pull", then hopefully "guix gc --list-failures" and "guix gc --clear-failures" subcommands will be available
<civodul>alezost: C-c . b is awesome, as usual!
<civodul>exactly what i wanted :-)
<alezost>here I also use "C-c . s" to perform "guix-download" on the current package (to see its hash), I'll send a patch for it soon
<alezost>I mean the hash of the package source
<civodul>excellent, i was about to propose it :-)
<civodul>C-c . l for lint also ;-)