IRC channel logs


back to list of logs

<taylanub>Steap: might I ask how you managed to do that? I'll be looking for a new job soon and want to do something as ethical as possible, with programming being probably my only good choice.
<taylanub>I'm in Germany, by the way. hm, maybe Suse or so...
<mark_weaver>Steap: yes, the employers dominate the working class, and yes, we all need to eat, but it does not follow that you would starve if you didn't use free software. that last suggestion is what I have an issue with.
<mark_weaver>don't get me wrong, almost everyone I know uses non-free software, and I'm not in the habit of getting in their faces about it.
<mark_weaver>what I have an issue with is you propagating this idea that we have to use non-free software because otherwise we would starve.
<f4lk0r_>Does anyone have any idea why when using the installation media, cryptsetup seems to hang whenever trying to do anything other than luksDump?
<f4lk0r_>luksOpen just sits there waiting way longer than it would normally
<Steap>mark_weaver: I'm just saying that we have to make sure we can survive
<Steap>and I can't blame someone for writing/using non-free software in order to do so
<Steap>Sure it's best if they can avoid it, but it is not always possible
<mark_weaver>and I'm just saying that I don't believe that your survival is at stake there. I think you should be honest that the real issue is that you aren't willing to make significant sacrifices in order to avoid non-free software.
<Steap>for instance I use Google Hangout on a regular basis at work; but 100% of my code is pushed upstream, so all in all, it's a good deal
<Steap>I could indeed leave my job to stop using Google Hangout, but that would not be the smartest thing to do, neither for me or Free Software in general
<Steap>taylanub: I joined a small company that was working on OpenStack and committed to do "open source"; later it was bought by Red Hat
<mark_weaver>okay, all I'm saying is that I don't believe you would starve, and I don't think you should pretend that you would starve.
<Steap>mark_weaver: Being jobless is not exactly easy, even in my country :)
<Steap>By "not easy", I mean that I have to pay for my own glasses, and they are terribly expensive, so I need money to do something as simple as seeing the floor
<mark_weaver>are you suggesting that there is no job you could get that doesn't entail using non-free software?
<Steap>mark_weaver: there is no "perfect job" that would not cause me to be "flexible" about my ethics
<mark_weaver>btw, I'm not sure you're wrong that the net outcome of your situation is beneficial to free software.
<Steap>so it is beneficial ? :p
<mark_weaver>it's just that I don't like BS arguments being propagated
<mark_weaver>I don't know, it's a hard question and I don't know the details.
<Steap>It's a more general point I'm trying to make
<Steap>Capitalism makes sure that some people are dominating some others, and that is the issue to adress
<mark_weaver>well, yes
<mark_weaver>but for one thing, there are plenty of jobs that don't involve using computers at all
<Steap>because as long as you have to feed yourself and your family, it's hard to think about ethics
<mark_weaver>maybe not jobs you would want...
<Steap>mark_weaver: yes, I could work in a weapon factory
<mark_weaver>and if one is willing to live on a lot less, then that opens up a lot of options for the resourceful person.
<Steap>I could work with GMOs, Monsanto is hiring !
<Steap>(Almost) Any job you take requires
<Steap>=> Almost any job you take requires that you are "flexible" about your ethics/beliefs
<mark_weaver>the "almost" is the key word there.
<Steap>yeah, I wouldn't leave my job for a better-paying job that would involve writing non-free software
<mark_weaver>and there are also ways to make some money without having a boss too
<Steap>mark_weaver: by "almost", I mean "there are not enough for everybody"
<Steap>mark_weaver: yes, but that does not solve the issue of capitalism either
<mark_weaver>I agree that writing non-free software is worse than using it
<Steap>+ finding a solution for yourself is not good enough: society should be fixed.
<mark_weaver>well, I agree, but I don't see how your choices are helping with that either
<Steap>Well I'm pretty sure we're fucked.
<mark_weaver>haha, me too
<Steap>And well, you know, I'd still like to live a bit
<mark_weaver>and that's why the capitalists win. we're addicted to money
<Steap>because hanging myself is not exactly what I'm willing to do
<Steap>in the meantime, writing Free Software is nice
<mark_weaver>so most people will compromise their values in order to get money.
<Steap>plus it's really funny to keep using the words "Free SOftware" when talking to bunisess people
<mark_weaver>and obviously we need to eat, but most people will compromise to make a lot more money than they need.
<Steap>I cant blame someone for the things they ahve to do to support their families
<mark_weaver>RMS is an example of someone who made very significant sacrifices in order to live in accordance with his ideals.
<Steap>Now, screwing the whole working class to get a new private jet is a different story.
<mark_weaver>do you have children?
<Steap>mark_weaver: Yeah, and to think he is a good example is terrifying
<Steap>mark_weaver: nope, but what about people who do ?
<mark_weaver>well, I chose not to have children for that reason, and so did RMS>
<mark_weaver>why is it terrifying?
<f4lk0r_>you chose not to have children for the sake of free software?
<mark_weaver>people don't _have_ to have children. it's not as if the population isn't going up way too fast
<Steap>Well, he's not really good at t alking to the public,right ? :p
<Steap>so having him as an example of how to fight capitalism... :D
<Steap>f4lk0r_: or just no to have them living as slaves to the capitalists
<mark_weaver>not specifically for free software, but more generally to preserve my freedom to follow my conscience and not become a wage slave.
<Steap>mark_weaver: friends asked me what I was spending my income on recently
<Steap>I could not really give them an answer apart from "well, rent, food, a laptop every 5 years, the occasional beer..."
<Steap>that felt pretty good :D
<Steap>"damn, I'm a slave, but not so much"
<mark_weaver>btw, I'm not saying that you're a bad person
<Steap>(which is still sad)
<Steap>mark_weaver: well I hope not.
<mark_weaver>I appreciate that you are working on free software, and contributing to guix, etc.
<mark_weaver>but when I hear bogus arguments that are hurting us, I feel the need to challenge them.
<mark_weaver>yes, refusing to use non-free software entails sacrifices. not everyone is willing to suffer those.
<mark_weaver>but don't pretend that you wouldn't be able to survive without it.
<mark_weaver>gotta go afk for a bit.
<Steap>yeah, I get your point, and I'm glad we can disagree without hating one another
<mark_weaver>we tend to delude ourselves into believing that we have no choice, to make ourselves more comfortable with a decision we've already made..
<Steap>I dont think there is a simple answer
<f4lk0r_>why is the cryptsetup package in the repos 1.6.1 when the latest version upstream is 1.6.8?
<f4lk0r_>maybe that's why I can't get cryptsetup to format or open anything?
<mark_weaver>f4lk0r_: sorry, I haven't played with cryptsetup yet.
<mark_weaver>the reason it is old is no doubt because no one has thought to update it yet.
<mark_weaver>we have over 1500 packages and a few dozen contributors, so it's hard to keep on top of all the updates
<mark_weaver>(we need better automation to detect upstream releases)
<f4lk0r_>so I'm new to guix
<f4lk0r_>trying to install it right now
<f4lk0r_>used arch for years
<f4lk0r_>how do you install just one package?
<mark_weaver>guix package -i <package-name>
<f4lk0r_>I did guix package --install cryptsetup
<f4lk0r_>and its pulling in all sorts of random stuff
<f4lk0r_>like font config stuff, and I think it's doing a full system update?
<f4lk0r_>there's no way that fontconfig is a dependency of cryptsetup
<mark_weaver>sorry, gotta go afk for a couple of hours
<{}grant>Offloading is really cool, wowzers!
<davexunit>f4lk0r_: it's downloading all of things needed for cryptsetup to run
<davexunit>all the dependencies
<davexunit>only cryptsetup (and any dependencies that are "propagated", which should few or none) end up in your user profile.
*{}grant has been looking into git-annex again, since Haskell building seems a-go ... man, there's over 30 Haskell package dependiences. :^P
<davexunit>I periodically chip away at big packages like that
<davexunit>being able to run mediagoblin would be cool, for example.
<davexunit>so I'll take care of a python lib or two sometimes
<davexunit>haven't done that in awhile though
<davexunit>I packaged 30 of them and got burnt out :)
<{}grant>davexunit: Yeah, I really want Mediagoblin and Gitlab support at somepoint. Gitlab is "the worst" at this, from what I've seen so far venturing into dependency hell.
<davexunit>ugh, good luck with gitlab.
<davexunit>that's definitely dependency hell
<davexunit>we haven't even gotten off the ground with ruby gems yet
<davexunit>a couple of blocking issues
<{}grant>After 0.8.2 lands, I plan on contributing a lot more generally again hopefully... I've had a long break, since doing anything in-terms of Guix.
<f4lk0r_>Well I don't know what to do
<f4lk0r_>without cryptsetup I can't set up my laptop how I'd like
<davexunit>f4lk0r_: just install cryptsetup
<davexunit>did you 'guix pull' before installing it or something?
<davexunit>that could trigger downloading a bunch of packages
<davexunit>due to rebuilds
<f4lk0r_>No, I didn't, I just did guix package -i cryptsetup
<f4lk0r_>which was fine I guess despite pulling in some dependencies
<f4lk0r_>the real problem is cryptsetup hangs
<f4lk0r_>luksDump works but luksOpen and/or luksFormat just sits there
<f4lk0r_>and sits there
<f4lk0r_>and sits
<davexunit>f4lk0r_: hmm
<davexunit>I'm sorry I can't really help further. I haven't used cryptsetup myself.
<davexunit>f4lk0r_: if you could send an email about this to, with the relevant error messages, that would be good.
<davexunit>that way, someone, perhaps our maintainer, can give you more help
<f4lk0r_>well, I would if there was any usable debugging error to give
<f4lk0r_>it just hangs there
<f4lk0r_>I could try to recompile with debugging flags or something
<davexunit>okay, then just say it hangs, and see if someone can reproduce it
<f4lk0r_>alright I'll send an email
<davexunit>mention if this is from GuixSD or another distro
<davexunit>perhaps run the program with strace
<davexunit>and see if there's any useful information that could be passed along
<davexunit>thanks for being patient and eating some dog food.
<f4lk0r_>well, the last thing in the strace is a semop() which is semiphores so the problem is probably a race condition
<f4lk0r_>are the package definitions stored on the hard drive somewhere?
<f4lk0r_>^ is that file on the hard drive of a guix installation
<{}grant>Hm, I'm trying to do a bootstrap eintall and I get a "no code for module (guix build-system haskell).
<{}grant>Weird. I was able to do it under a root user, without an issue -- but not with sudo.
<{}grant>Ha! After all that, the kernel paniced on reboot ... which my laptop, you can't hard-poweroff. X^D
<f4lk0r_> /close
<rekado>sneek: later tell davexunit I still think it's a good idea to substitute all literal strings passed to "require" with full store paths to avoid having to propagate *all* inputs. This won't work for dynamic requires, but I suppose that's not quite as common as regular requires with string literals.
<sneek>Got it.
<rekado>re systemd: I haven't really met any haters yet. (Maybe I'm not following the flamewars closely enough.)
<rekado>as a sysadmin I don't like it much, because it means I have to learn something new for little gain.
<rekado>as a user I don't like it much because I find it incredibly complicated and don't want/need all these features in an init system.
<rekado>as a programmer I don't like it much because of the intra-module dependencies (but maybe that's just projects that assume everyone uses the same subset of systemd).
<rekado>as a GNOMEr I don't like that its dependency on logind keeps me from using GNOME on Guix.
<mark_weaver>we'll get GNOME in Guix without systemd
<mark_weaver>wingo looked into it, and it seems that logind shouldn't be too hard to extract from systemd
<rekado>other than that I guess systemd is fine for me. (i.e. I don't care.)
<mark_weaver>although it's annoying that we will have to essentially create a fork, similar to eudev.
<rekado>is there any chance to "fix" GNOME, a GNU project, rather than forking logind?
*rekado goes afk
<mark_weaver>although I haven't looked into it, I suspect that logind itself is probably a good idea
<mark_weaver>it's not a drag that it (and so many other things) have become part of systemd
<civodul>Hello Guix!
<civodul>iyzsong: thanks for fixing these RUNPATH issues!
<iyzsong>sure, there still have some remains :-)
<civodul>yep :-)
<civodul>have you looked at nspr yet?
<iyzsong>I'm on iptables
<civodul>ok, i take nspr then
<iyzsong>also, I think we can make icecat link with mesa and gst-plugins-base to gain webgl and h264 video support.
<civodul>oh sure
<civodul>and somehow icecat doesn't seem to use PulseAudio, i don't know why
<civodul>but this causes problems when PA is already running
<iyzsong>maybe it's dlopen too (libGL and GStreamer is dlopened).
<iyzsong>I figure out by set LD_LIBRARY_PATH=$HOME/.guix-profile/lib, and then open and
<civodul>you mean libpulse is dlopened?
*civodul now in elfutils
<iyzsong>yes, ldd doesn't show
<civodul>we need to make sure it actually tries to dlopen it
<civodul>elfutils is tricky, leaving it for later
<davexunit>rekado: got your message from sneek in #guile. that method sounds good, and I've experimented with it, but I'm not excited about relying on a regexp match to do the right thing.
***boegel|afk is now known as boegel
<effa>Hi, I'm a bit confused about 'quasiquote.
<effa>I have a procedure that generates
<effa>(quasiquote (("name" ("cabal-install")) ... (unquote-splicing (if (flag debug) (quasiquote (("build-depends" ("directory >= 1"))))))))
<effa>then another procedure that takes the above, then defines locally the function (flag name) and I would like to evaluate the above expression with that flag defined.
<effa>However, I just get the identical unexpanded expression.
<effa>If I copy the expression in the REPL and replace (flag degub) with #t and press ENTER, then it does get evaluated.
<effa>... confused. How do I evaluate it in a function?
<civodul>hi, effa
<effa>Hi civodul
<effa>I'm giving a go at restructuring the hackage importer :-)
<civodul>(let ((flag 'foo)) `((foo bar baz ,(if flag 'x 'y))))
<civodul>the unquote means that unquote code is in the scope where it appears
<civodul>effa: ooh, good :-)
<civodul>the above is equivalent to:
<civodul>(let ((flag 'foo)) (let ((thing (if flag 'x 'y))) `((foo bar baz ,thing))))
<civodul>(comma is unquote and backquote is quasiquote)
<effa>yes, that is clear.
<civodul>elfutils → ✓
<effa>But I'm generating a function whose result is `(foo ,@(if (flag x) 'yes 'no))
<effa>and I would like to define a function (flag x) in another function which takes the above as parameter.
<effa>Maybe, should I use delay and force instead?
<civodul>not sure what you mean
<civodul>you can't "generate a function"
<civodul>i mean there's 'eval', but we'll do as if it didn't exist
<civodul>what you can do is "return a function"
<effa>what I means is that:
<effa>(define (eval-cabal sexp env) (define (flag x)) sexp)
<effa>takes the quasiquoted expression above as the 1st argument.
<effa>Is that wrong?
<effa>Ooops, the above quasiquote is anyway wrong.
<effa>`(foo ,@(if (flag x) '(yes) '(no)))
<civodul>no, but you should think of the sexp as an AST, a data structure that you need to iterate on to perform the actual interpretation
<civodul>as in
<civodul>(tell me if i'm just stating the obvious, not sure what your background is)
<effa>I'm an EE engineer. So I have a 1 year on programming, but it was with MODULA-2 back in the days :-)
<civodul>ok :-)
<civodul>well you seem pretty well versed in all this anyway ;-)
<effa>I think I understand what you mean. But, I thought that I could evaluate the AST by evaluating a quasiquoted sexp in an environment where the conditionals are functions defined locally.
<civodul>that would be possible if we were to use things like 'eval', but there are good reasons not to do it
<civodul>"eval is evil" is what people say ;-)
<effa>Ah, OK, that way would require eval...
<effa>would it be reasonable to use delay and force?
<effa>instead of quasiquote?
<civodul>that wouldn't help
<civodul>because scoping is static ("lexical")
<civodul>so the reference to the 'flags' variable is resolved in the lexical environment
<effa>I guess I need to change strategy... thanks!
<civodul>really the 'eval-cabal' procedure should resemble that in the SICP example above
<civodul>simpler, probably
*civodul has to go
<mark_weaver>hi wingo!
<wingo>heya mark_weaver :)
<wingo>i wrote something while my network was down
<wingo><wingo> mark_weaver: did you see that i made some progress on logind?
<wingo><wingo> git clone
<wingo><wingo> anyways, had to stop that for now, there are other hacks to hack :)
<mark_weaver>oooh, nice!
*mark_weaver clones it
<wingo>it's systemd's git, with a set of slash-and-burn patches on top
<wingo>but, it does configure and it just builds logind and pam_elogind (which was called pam_systemd)
<wingo>i tried being more incremental in the beginning but it was slow going so i hacked away
<mark_weaver>heh :)
<wingo>there's a bunch of dead code there...
<wingo>anyway :)
<wingo>building and running without linker errors is a start :)
<mark_weaver>hopefully we'll have core-updates merged within a couple of days, so then we can make some more progress on GNOME in Guix
<wingo>that would be nice
<wingo>i built most of a chicken coop today and guix feels very DIY in the same way ;)
<davexunit>oooh chicken coop
<davexunit>had one as a kid.
<davexunit>I miss it.
<davexunit>was cool to have chickens roaming around the yard
<wingo>i never had one and am looking forward to more ridiculousness in my life
<wingo>funny chickens in the year will be nice :)
<mark_weaver>obviously we aim for Guix to have a nice DE out of the box, but obviously we have a ways to go
<davexunit>the day we have a usable GNOME 3 desktop will be a fantastic day
<davexunit>GuixSD will feel more normal
<mark_weaver>I'm really looking forward to having gnome-shell again.
<wingo>mark_weaver: it seems not terribly far away though. logind is really the key piece afaict
<mark_weaver>yay :)
<wingo>lennart has always said "people should just rip this out and maintain a fork of it" and i don't know why people haven't done it yet
<mark_weaver>thank you for spearheading this
<wingo>heh ;)
<wingo>am hoping to back off in terms of intensity
<mark_weaver>I expect that there are many other people who would be willing to pick this up. all the people who either don't like or cannot use systemd.
<wingo>but would be nice to get some logind release working
<mark_weaver>if lennart is cool with having it separated, that's even better.
<ijp>wingo: elogind pronounced "elegant"?
<wingo>i don't think systemd people will want to use an external logind
<mark_weaver>oh, okay. I read too much into that quote, I guess.
<wingo>but i don't see why gnome-session, gnome-settings-daemon etc wouldn't be ok with alternate implementations of the same interface
<wingo>ijp: lol :)
<mark_weaver>they shouldn't even be able to tell the difference, if done properly, right?
<wingo>i just chose that random name because eudev, and the name wasn't used
<mark_weaver>(assuming that they don't go out of their way to detect it)
<wingo>mark_weaver: pretty much but right now it's all bundled into systemd
<wingo>so knowing that you have a recent logind means PKG_CHECK_MODULES(SYSTEMD, systemd > 209)
<mark_weaver>for some reason my 'git clone' is hung. hmm.
<wingo>which implies a lot more functionality
<wingo>mark_weaver: i think maybe not hung but slow because http
<wingo>do a du -sm elogind
<wingo>and see if there are things there
<wingo>maybe i needed to gc the repo or something
<wingo>for example gnome when it has systemd will log to the journal
<wingo>which is something that elogind wouldn't have
<wingo>so some users of systemd would need to complicate their version checks to support elogind
<wingo>and add more conditionals
<wingo>which is always a negative ;)
<wingo>so, a little bit of a sell. but i think e.g. gnome would be ok with it.
<mark_weaver>sure, and integration means fewer interfaces to keep stable, fewer backward-compatibility hacks, etc.
<wingo>mark_weaver: right from systemd's perspective; but i was referring to gnome-settings-daemon's perspective
<mark_weaver>I guess we'll just have to reimplement the systemd interfaces that GNOME depends on.
<wingo>mark_weaver: pretty sure the only ones are the logind ones, and some incidental journal usage. the other thing that is coming is a new protocol for dbus, though.
<wingo>not sure what to do about that.
<wingo>but it hasn't really landed yet so nothing to do as of yet.
<mark_weaver>sure, the set of important interfaces may grow with time.
<mark_weaver>I'm not too worried about it. they'll have to keep those interfaces relatively stable, since software higher in the stack depends on it.
<mark_weaver>probably these are good interfaces to have anyway.
<mark_weaver>new protocol for dbus? is dbus going to be integrated with systemd also?
<mark_weaver>can you cite something for me to read about it?
<df_>only when the linux kernel is eventually integrated into systemd...
<mark_weaver>wingo: not sure why, but this clone is just not happening.
<mark_weaver>oh, actually, the contents of .git are slowly getting larger.
<mark_weaver>very slowly
<mark_weaver>after 12 minutes, it's at 4.6M. so about 6K per second
<mark_weaver>reminds me of my old dialup days :)
<mark_weaver>except with no download progress bar
<wingo>yeah the git push was pretty weird like that too
<wingo>my web site seems fine over http and icecat
<wingo>so i wonder if it is something with the version of git in guix
<wingo>which i also used for the push
<mark_weaver>I've cloned a lot of different repos and pushed to savannah with git from guix, and never had this problem before
<mark_weaver>it seems more likely something to do with your VM
<wingo>could be
<mark_weaver>up to 11M
<mark_weaver>dum de dum...
<wingo>yeah, better go knit a sweater
<mark_weaver>leave it running overnight :)
<mark_weaver>wingo: how big is the .git directory in your elogin checkout, as measured by 'du -sh' ?
*mark_weaver would like a rough estimate of how long this will take
<mark_weaver>obviously we will have to host this git repo somewhere else, unless the problem can be fixed. my guess is that your VM provider is throttling the connection.
<wingo>mark_weaver: 38 mb
<wingo>i don't think it's the vm provider, probably just a misconfiguration
<mark_weaver>okay, that's doable. I'm half way there.
<mark_weaver>so the total time to clone should be about 84 minutes
<wingo>that's ridiculous
<wingo>so that's just a repo initialized with git repo init --bare
<wingo>and having enabled the usual post-update hook for http
<wingo>maybe linode has disk problems with that thing
<mark_weaver>yeah, it takes a *lot* less time for me to clone the linux repo, or gecko-dev
<mark_weaver>my ancient sparc server would do better than this
<wingo>well those use special git-knowing protocols i think
<wingo>this is just a dumb directory
<wingo>i see requests to files that don't exist, in the http error log
<davexunit>wingo: check out git-http-backend
<davexunit>comes with git, simple enough to use
<mark_weaver>wingo: I cloned gecko-dev over https with git from guix
<mark_weaver>it's true that I used the git: protocol for linux though
<wingo>davexunit: that's a lot more work to set up than a simple directory
<wingo>given that the web server exists already
<davexunit>I understand, but it's only a few lines of config to add to apache or nginx or whatever you use
<wingo>so i get errors looking for http-alternates...
<wingo>i wonder if there are files that are not being generated
*wingo using the "dumb" protocol apparently :P
<mark_weaver>okay, it's done cloning
<wingo>bah, can't be bothered to understand the security implications of the smart protocol today :)
<mark_weaver>so it was actually only about 50 minutes
<mark_weaver>the kids are back, so I have to go afk again.. will look at elogind later. thanks for working on it!