IRC channel logs

***civodul changes topic to 'GNU Guix | http://gnu.org/s/guix/ | http://www.fsf.org/news/fsf-adds-guix-system-distribution-to-list-of-endorsed-distributions | 0.8.2 is in the works! | This channel is logged, see <https://gnunet.org/bot/log/guix>.'

<mark_weaver>hi civodul!

<mark_weaver>hydra fails to evaluate core-updates when subset is "all". I sent you mail about it with the error output.

<mark_weaver>my private armhf build of core-updates is making good progress though

<taylanub>mark_weaver: I'm around now

<mark_weaver>taylanub: and now I must sleep :)

<taylanub>ok, see you later :)

<mark_weaver>I wanted to brainstorm about the qt-5/i686 issue, but it'll have to wait until another time.

<taylanub>ok

<civodul>mark_weaver: ok, i'll check that

<civodul>hydra.gnu.org is surprisingly fast today

<civodul>good ideas to consider: https://github.com/NixOS/nixpkgs/issues/7220

<Sleep_Walker>without CFLAGS?

<civodul>i would use CFLAGS for these things, rather than a spec file or wrapper as the text suggests

<rekado_>there's a problem with latest master: ERROR: No variable named ghc in #<interface (gnu packages haskell) 23cc3f0>

<rekado_>This has been reported to the ML before.

<rekado_> http://lists.gnu.org/archive/html/guix-devel/2015-04/msg00203.html

<rekado_>I get this when trying to install a package into a new user profile.

<civodul>hmm

<civodul>i can't reproduce it

<civodul>did you use "make install"?

<rekado_>yes

<rekado_>I just disabled the ghc profile code and it works fine now.

<civodul>can you check the timestamps of the installed haskell.{scm,go}?

<civodul>and the contents? (is 'ghc' in there?)

<rekado_>ghc is in fact there. haskell.scm is just as it came from master (I did "git checkout -- gnu/packages/haskell.scm" after removing it temporarily.)

<civodul>ok

<civodul>and is there ~/.config/guix/latest ?

<rekado_>I have the problem with multiple user accounts that do not have ~/.config/guix/latest (we don't use "guix pull")

<davexunit>civodul: that hardened nixos issue is interesting

<rekado_>How often is this package list updated? --> http://www.gnu.org/software/guix/package-list.html

<freaj>rekado_: http://hydra.gnu.org/jobset/gnu/master

<rekado_>freaj: I don't understand.

<davexunit>civodul: do we already build with any of these gcc flags that were recommended?

<davexunit>I'm curious what our story is there

<freaj>Sorry rekado_, what I wanted to say is you can see the updates on the link I gave you :P

<davexunit>to everyone: what would you call a data type that represented the type of machine that you were installing an OS onto?

<davexunit>examples: VM, physical host, container

<davexunit>I thought of "infrastructure", but I hate typing it.

<davexunit>I currenly have a <machine> type that encapsulates a name, an <operating-system>, and this as-of-yet unnamed type

<fr33domlover>davexunit, you mean like a host?

<fr33domlover>the thing "uname -a" prints?

<fr33domlover>(or uname -SOMETHING_ELSE)

<fr33domlover>maybe "hardware platform"

<fr33domlover>(takes idea from `man uname`)

<freaj>Yeah I think platform would be cool?

<davexunit>I had thought of platform, but we refer to things like i686-linux-gnu as a platform

<davexunit>so I don't want to cause confusion

<davexunit>I'm writing a deployment tool that abstracts away the details of manipulating any particular "platform". so that you can use the same tool to manage VMs, physical hosts, containers in the future, perhaps things in an OpenStack "cloud"

<freaj>you need a word to designate the kind of "physical hosts/vm/whatever" ?

<davexunit>as a whole, yes.

<davexunit>I think platform fits.

<davexunit>so does infrastructure.

<freaj>wow... that would be a "kind of machine" but I have no idea

<davexunit>hehe

<freaj>we should create a word for that :P

<davexunit>I tried machine-type, but it leads to some ambiguity

<atheia`>davexunit: what about "stratum"? It's a bit abstract, but kind of makes sense, and is short…

<davexunit>atheia`: hmmm maybe

<davexunit>I'm going to try platform for now

<davexunit>and if this project gets anywhere I can revisit

<atheia`>:-)

<davexunit>I'm cloning nixops, more or less.

<civodul>davexunit: no we don't use _FORTIFY_SOURCE, -fPIE, or anything like that

<davexunit>civodul: thanks

<civodul>i've been meaning to try it out

<civodul>i suspect _FORTIFY_SOURCE can be relatively expensive in that some packages may fail to build

<civodul>and so we'd have to do pretty much upstream work

<davexunit>ah

<davexunit>I see

<civodul>rekado_: the list is updated once a day, except when there are issues

<civodul>davexunit: re your previous question, 'machine-kind'?

<civodul>'platform' sounds good too

<davexunit>civodul: I've found that having a 'machine-' prefix leads to ambiguity

<davexunit>since I have a data type called "machine"

<davexunit>that would have a "machine-kind"

<civodul>yeah i see

<davexunit>I'll stick with platform

<civodul>a common issue with naming ;-)

<civodul>ok

<davexunit>thanks everyone

<davexunit>the <platform> type is essentially just a bunch of hooks to do all the real work of provisioning, installation, etc.

<civodul>makes sense

<civodul>rekado: by any chance do you have new clues regarding that ghc failure? :-)

<civodul>rekado_: in particular is the installed haskell.go older than the installed haskell.scm?

<civodul>oooh, i feel that this could be the reason

<davexunit>hmmm I never have networking on my guixsd system without having to manually bring the interface down, up, and run dhclient

<civodul>using wicd?

<civodul>wicd doesn't work for me, except on wired networks

<davexunit>I've tried wicd

<civodul>never took the time to investigate :-/

<davexunit>I have a wired network

<davexunit>I tried the other service before wicd, too

<davexunit>nothing

<davexunit>and ntp fails because of it

<civodul>dhcp-client-service doesn't work?

<davexunit>so my clock is off by 4 hours

<civodul>it used to work for me

<civodul>oh, but that's not just because of the lack of ntp :-)

<davexunit>yeah

<davexunit>hopefully if I fix networking I will fix this

<davexunit>seems to be a failure of dbus, actually

<davexunit>dbus-system service fails

<davexunit>so networking fails

<davexunit>so ntp fails

<civodul>but dhcp-client-service doesn't depend on dbus IIRC

<davexunit>I guess wicd does

<davexunit>which I tried to use now

<civodul>ah right

<davexunit>I guess I'll switch back to the dhcp service and reconfigure

<davexunit>and hope that works

<davexunit>I'd also like to know why dbus fails :)

<civodul>probably because of a stale PID file no?

<davexunit>oh maybe that's it

<civodul>mark_weaver rightfully complained about it in the past

<davexunit>ugh any my terminal is so messed up

<davexunit>the cursor is always off by one

<civodul>uh

<civodul>what terminal?

<davexunit>I guess it's not the terminal emulators fault, it happens no matter which I use.

<davexunit>might be a bash issue

<davexunit>xterm and xfce4-terminal are what I've tried so far

<civodul>are you using the bash completion thing?

<davexunit>no

<davexunit>every time I press the up arrow to fetch a previous command, the problem occurs

<davexunit>when I press C-a to go to the beginning of the line, the cursor is over the 2nd character

<civodul>would be nice to see if this is reproducible in a VM

<davexunit>but only by appearance, if I edit, the edits happen one character behind where it looks

<civodul>weird

<davexunit>yeah

<davexunit>oh yeah, it might be a $PS1 problem...

<davexunit>maybe?

<davexunit>export PS1="\\u@\\h: \\w\\a$ "

<davexunit>does that look okay?

<mark_weaver>davexunit: try removing the \\a

<mark_weaver>do you really want it to beep at every prompt?

*mark_weaver tries to go back to sleep

*davexunit never heard a beep

<davexunit>don't know how that got in there

<davexunit>that seems to have done the trick

<davexunit>thanks mark_weaver

<davexunit>one problem down

<davexunit>dbus still wrecks everything for me

<davexunit>hmm, the dbus service is running, actually

<davexunit>and the dhcp-client-service

<davexunit>yet I have no network access and gnome-terminal refuses to start

<mark_weaver>davexunit: this is what I'm currently using, although it's not ideal: http://paste.lisp.org/+3549

<mark_weaver>but without it, every time my system crashes it needs a lot of manual cleaning up to get it booted back up cleanly

<davexunit>thanks

<mark_weaver>I'm really amazed that we've gone so long without something like this. I guess that civodul's machine never crashes, or something :)

<mark_weaver>one of our key features is supposed to be that you can (almost) always rely on being able to boot into an older system if the newest one fails, but this problem undermines that, and leaves people with systems that won't boot properly no matter what profile they try to boot into.

<mark_weaver>honestly, I think we should just apply the patch I just cited for now, if no one has the time to make a better one.

<davexunit>I support that

<mark_weaver>but now I need to try to get more sleep...

<davexunit>gets us over this issue for the time being

<davexunit>I can't come up with the magical incantation needed to start dbus properly

<davexunit>I deleted /var/run/dbus and rebooted

<mark_weaver>davexunit: run "deco status dmd" and tell me what is stopped?

<davexunit>nothing is stopped

<davexunit>I can actually see the dbus daemon running

<davexunit>but it seems that nothing that actually requires dbus works

<davexunit>despite all the services running

<davexunit>and I can't start gnome-terminal, which complains about a dbus issue

<mark_weaver>it might be an issue with the session dbus, not the system-wide dbus

<davexunit>hmmm

<davexunit>and ntp doesn't work at all. the process is defunct when I check for it with ps -ef | grep ntp

<mark_weaver>when logged in to XFCE, you should have two dbus processes, one with uid 'messagebus', and one with your uid.

<davexunit>ah so xfce starts it automatically?

<mark_weaver>yes

<davexunit>I'm using ratpoison, so that explains some things

<mark_weaver>ah

<davexunit>I'll turn my focus back to why wicd-service nor dhcp-client-service gets me a DHCP lease

<mark_weaver>davexunit: don't run both at the same time.

<davexunit>I'm not

<mark_weaver>ah

<davexunit>I've tried both separately

<davexunit>to no avail

<iyzsong>davexunit: you can try 'exec dbus-launch $1' in your ~/.xsession. If you install both dbus and gnome-terminal into profile.

<mark_weaver>I think you should apply the patch I cited, reconfigure and reboot, for starters

<davexunit>iyzsong: thanks

<davexunit>mark_weaver: yeah... I guess so.

<davexunit>that will have to wait until later. running out of time before work.

<mark_weaver>I've found that sometimes things are just subtly broken

<davexunit>I see a dhclient process running for my ethernet interface

<davexunit>I don't know why it doesn't actually work

<mark_weaver>look for diagnostic output on tty1

<mark_weaver>(first virtual terminal)

<mark_weaver>maybe try 'deco restart networking' and then look there.

<mark_weaver>I've found dhclient to be somewhat unreliable, and often needs manual restarting

<mark_weaver>but since using wicd-service, that seems to be less of a problem for me. maybe wicd takes care of restarting it automatically or something, dunno.

<davexunit>well, 'deco restart networking' spawned a new dhclient process without stopping the other one.

<mark_weaver>so, the main downside to the patch I cited is that it assumes that /var and /tmp are on the root partition. but then we are still applying new patches that make the same assumption, such as http://git.savannah.gnu.org/cgit/guix.git/commit/?h=core-updates&id=7ce597ff9e7232f91016d5e4945cd24ec691223c which create things in /var at activation time, before any other filesystems are mounted

<mark_weaver>davexunit: was the other one started manually by you? or maybe by wicd?

<davexunit>no

<davexunit>I haven't run wicd

<mark_weaver>which networking service are you using now?

<davexunit>dhcp-client-service

<mark_weaver>okay, I don't have much experience with that one.

<mark_weaver>I recommend wicd-service for now

<davexunit>I can try it again

<davexunit>but it did nothing last time

<mark_weaver>notes about setting it up: (1) you also need 'wicd' in the list passed to 'dbus-service', (2) 'wicd' should probably be in your system-wide profile, and (3) add 'netdev' to your user's supplementary-groups

<mark_weaver>with those things, you should be able to run 'wicd-gtk' from within your X session to configure networking.

<davexunit>oh it won't do a wired connection automatically?

<mark_weaver>(or 'wicd-curses' if you prefer)

<mark_weaver>you can configure it that way; I forget what it does by default there.

<mark_weaver>how it deals with wired connections can be configured from wicd-gtk though (and maybe also wicd-curses)

<mark_weaver>and the configuration is persistent. once you set things up, wicd just works automatically from then on.

<mark_weaver>it stores configuration in /etc/wicd

<mark_weaver>(which is stateful, not managed by guix)

<davexunit>okay, launching wicd-gtk did the trick

<davexunit>I will stick with that for now

<mark_weaver>I am going to be so fried today...

<mark_weaver>cool!

<davexunit>thanks for the help

<davexunit>we need some more robust daemons

<mark_weaver>np! thanks for your persistence :)

<mark_weaver>since applying the /var/run and /tmp cleanup, and using wicd-service, my system boots up reliably now

<mark_weaver>occasionally I have to "deco restart nscd" or "deco restart tor" (if routing web through tor, as I do)

<davexunit>going to switch IRC over to the guixsd machine, brb

<civodul>mark_weaver: interestingly, i only have stale /tmp/.X0-lock ~once every 20 boots, but no other such issues

<mark_weaver>civodul: the /tmp/X.

<civodul>perhaps #:grace-delay should be increased as well

<civodul>maybe it's sorta ok for fast machines, and too low for slower machines

<mark_weaver>it's also relatively rare for me that X failed to start up, but dbus failing to start was extremely common. almost every time after the crash.

<mark_weaver>(and I seem to recall it happened sometimes even when shutting down normally)

<mark_weaver>avahi doesn't come up properly if networking is not working yet. for wireless users, this is often the case.

<civodul>mark_weaver: but you should apply the patch that cleans up /tmp

<civodul>i'll work up something for the /var/run issue that you mentioned

<civodul>but the /tmp cleanup would already be an improvement

<mark_weaver>civodul: which patch are you referring to? http://paste.lisp.org/+3549 ?

<mark_weaver>that's the only one I recall coming up with for this issue

<mark_weaver>I had started to think about another approach, but it had other problems.

<mark_weaver>but as long as we are adding new services that create directories in /var/lib at activation time, I don't see a reason not to apply http://paste.lisp.org/+3549 or something close to it.

<mark_weaver>(e.g. 7ce597ff9e7232f91016d5e4945cd24ec691223c)

<mark_weaver>btw, the reason I'm so paranoid with those 'false-if-exception' thingies is because I'm worried that things will fail because of incorrectly-encoded filenames.

<mark_weaver>but surely it could be done better..

<civodul>mark_weaver: i was thinking of a 'cleanup' essential service, as you initially suggested in your message

<civodul>except that it would only handle /tmp

<mark_weaver> /tmp is a problem in maybe 1/10 crashes, or less.

<mark_weaver> /var/run is a problem on *every* crash.

<civodul>oh, ok

<mark_weaver>I'm not very motivated to find a solution for /tmp only

<civodul>yeah you could commit http://paste.lisp.org/+3549

<civodul>however, doesn't it remove /tmp itself?

<civodul>oh it recreates it after, nvm

<mark_weaver>civodul: yes, and then it recreates it.

<civodul>so yeah, let's take that route for now

<mark_weaver>civodul: will do, thanks!

<civodul>thank you!

<civodul>have you been using it?

<mark_weaver>you're welcome. I apologize for being grumpy about it

<mark_weaver>yes, I've been using it for at least a couple of weeks. it has made a huge difference for me.

<civodul>i would be as grumpy if i suffered more from this problem ;-)

<civodul>ok

<civodul>good

<mark_weaver>it actually completely solved the issues for me.

<mark_weaver>pushed

<civodul>thanks!

<davexunit>yay!

<davexunit>I'll reconfigure and reboot later and see how it goes

<davexunit>civodul: which guix image did you use when you made stickers?

<civodul>the former logo, with "The GNU System"

<civodul>but i think it went through Scribus or something

<civodul>someone helped me

<civodul>i have the PDF here but i don't remember how it was obtained

<civodul>yeah the PDF properties say Scribus

<civodul>it has color management features

<davexunit>ah okay

<civodul>so you can somehow use it to make a CMYK thing

<davexunit>I wanted to try to make a small batch of either Guix or GuixSD stickers

<civodul>cool

<civodul>i could try to mail you a bunch of those Guix stickers that i have :-)

<davexunit>it would probably be less expensive to just get new ones that are shipped domestically

<davexunit>but thanks :)

<mark_weaver>damn, one of libtool's tests fails consistently on armhf

<mark_weaver>blah

<mark_weaver>170: Run tests with low max_cmd_len

<mark_weaver>it seems that almost no one tests their software on anything other than intel these days

<mark_weaver>scary stuff, given how locked down modern intel systems are becoming...

*mark_weaver goes afk for a while

<civodul>mark_weaver: yes we have that libtool failure elsewhere as well

<civodul>i'll try to investigate later today

<civodul>ouch, libtool-2.4.6.tar.gz includes MacOS X meta-data files

<civodul>that's silly

*civodul files a bug/rant report

<davexunit>oof

<civodul>it was actually discussed before: https://www.mail-archive.com/libtool@gnu.org/msg13604.html

<civodul>i find it crazy that the maintainer sees no issue with using MacOS

<civodul>anyways

<Steap>hehe

<Steap>civodul: we've got a lot of Free Software hackers running Mac OS because GNU/Linux does not work

<davexunit>umm

<Steap>funny how they were all running GNU/Linux when it was hell (back in the 90s)

<Steap>and now that it works out of the box, it seems too hard

<davexunit>lol

<davexunit>OS X *still* ships with GNU software from 2007!

<Steap>hahah

<davexunit>madness!

<davexunit>they refuse to include GPLv3 licensed software.

<Steap>the funniest thing is that they all run VMs or Docker-on-Mac to do development

<davexunit>yup

<davexunit>and their production machines are GNU/Linux

<Steap>of course.

<taylanub>I have Emacs 22.1.1 here, yay! (using OS X at work, mandatorily for iOS develoment)

<davexunit>it's all due to proprietary userspace stuff that everyone relies upon

<davexunit>Skype and co.

<davexunit>I've run Skype on work computer, but never will run it on one of my own.

<Steap>davexunit: I heard they make cool containers for that

<Steap>so you can run Skype easily

<Steap>if you want to do that for reasons that have yet to be determined

<davexunit>haha I don't currently

<davexunit>I work at the FSF so I don't ever have to use skype

<Steap>I guess that's the only place where you will not have to install non-free software

<Steap>do you ever do audio/video chat ?

<davexunit>personally or at the FSF?

<davexunit>we have SIP phones

<Steap>oh, ok

<Steap>I'm asking because at work, we use either Google Hangout or Bluejeans

<Steap>and I'm not too found of that

<Steap>though I must admit they work really well

<davexunit>webrtc looks promising to us

<Steap>yes, but I haven't found anything that works really well yet and is Free Software

<davexunit>yeah

<civodul>i guess there are more and more corporate free software developers

<civodul>but those are not "hacking for freedom"

<civodul>it's this kind of details that illustrate this trend

<davexunit>yeah

<davexunit>I have mixed feelings about it

<civodul>yeah, because on one hand it's definitely a good thing

<davexunit>I would prefer people to hack for freedom's sake

<civodul>but on the other, there's some "dilution" of the ideals

<civodul>yeah

<Steap>davexunit: yeah, well, I'm weak and I need to eat

<davexunit>but being able to produce free software "on the clock"

<davexunit>is great

<civodul>right

<davexunit>of course, compromise of ideals comes with that most times.

<civodul>which reminds me i'd like to watch Mako's talk on user empowerment

<davexunit>civodul: that's on the way

<davexunit>soon!

<civodul>cool :-)

<davexunit>civodul: remember the "rock your emacs" talk that got messed up? well, the screencast and audio was still fully recorded: http://www.sturm.com.au/2015/talks/rock-your-emacs-libreplanet/

<davexunit>I want to mirror the video on media.libreplanet.org for some more exposure

<civodul>neat, i should watch that one as well

<davexunit>it's introductory material

<davexunit>for folks who've been afraid to take the plunge into Emacs Lisp

<davexunit>but it's entertaining.

<davexunit>Ben builds up some elisp functions to help automatically generate emails to send his parents

<civodul>heh

<bavier>babies at libreplanet, how exciting!

<davexunit>heh

<davexunit>there's *always* a baby in the room for every talk somehow

<davexunit>it always happens

<bavier>:)

<civodul>oh the "rock your emacs" slides even mention Guix :-P

<davexunit>civodul: haha really?

<davexunit>which slide?

<davexunit>oh I found it

<davexunit>nice!

<davexunit>Ben was excited for guile-emacs

<davexunit>some day..

<civodul>yeah

<civodul>it's one of these projects everyone hopes for but nobody works on ;-)

<civodul>(me included ofc)

<civodul>libtool should be fixed now

<civodul>later!

<mark_weaver>Steap: are you suggesting that you would be unable to feed yourself if you didn't use proprietary software?

<mark_weaver>it may be that you would have to make some sacrifices that you'd rather not make, but please don't pull out the "I need to eat" excuse.

<mark_weaver>the "I'm weak" part is the crux of it, I'd say

<joehillen>I really hate those kinds of discussions. The idea that things like Mac and Skype exist and don't support freedom are somehow free software's fault is stupid.

<df_>mark_weaver: my attitude is that if you succeed in living up to all your ideals, you don't have enough ideals

<joehillen>df_: ++

<df_>so it is ok to compromise occasionally as long as you admit to yourself why

<df_>and the answer may well be fear or laziness

<joehillen>bingo, it's been hard, but I'm happy with the number of ideals I've been able to live by

<df_>yeah same, and I'm working towards improving it

<mark_weaver>df_: I agree with you

<df_>but beating yourself up for being 'weak' is not productive

<mark_weaver>it's just that I'm allergic to BS, and when someone claims that they need to use proprietary software because they "need to eat", I think that's BS.

<df_>to some extent, but everyone has priorities

<mark_weaver>sure, but that's not what he said

<df_>ok

<df_>I was lucky enough to be paid to spend a year working on a free software (well, they called it open source) project

<mark_weaver>I'm not able to live up to all of my ideals either. I wouldn't claim such.

<df_>but my colleagues had no problem using macbooks

<joehillen>I've never been able to understand why Macbooks are so popular among FOSS developers. The ones that I ask say "It just works" which I haven't found to be true either.

<mark_weaver>yeah

<df_>it's horrible

<davexunit>I used OS X for 2 years at a past job

<davexunit>it sure as hell didn't "just work" for doing software development

<df_>I've used it, I spent all my time in a full screen terminal with a screen session

<mark_weaver>my guess is that they don't want to sacrifice things like the nice hardware and multitouch trackpad

<df_>you don't have to run os x on them!

<davexunit>the *only* thing that makes me jealous of current apple hardware are the high DPI displays

<mark_weaver>well, true, but there are a lot of downsides to choosing apple hardware to run free software.

<df_>my current work machine is a thinkpad x1, which allows you to run mostly-free software

<df_>a compromise

<mark_weaver>the one button mouse, the keyboard layout, not to mention all the blobs you need.

<df_>my home machines are an x60 and an x200, both running libreboot and free software from the ground up

<mark_weaver>df_: I'm glad to hear that :)

<df_>only one of them running guix so far :)

<davexunit>the BIOS is that all that remains to be freed on most of my machines

<mark_weaver>heh :)

<davexunit>sans the Novena

<joehillen>I wanted to buy a librem (even though it isn't 100% free because of intel), but it wasn't going to be ready in time for my newest job https://www.crowdsupply.com/purism/librem-laptop

<mark_weaver>heh, ask about the librem on #libreboot :-/

<df_>if you wish to start a flamefest

<df_>it doesn't have a free bios, it just potentially could at some point in the future

<joehillen>oh yeah? what's the story? I don't follow BIOS related stuff

<mark_weaver>where "potentially" == "not going to happen"

<df_>well, anything could happen

<df_>apple could release all their source code, firmware included, tomorrow

<df_>but some things are more likely than others

<mark_weaver>as in, all of the molecules of air in this room could suddenly go into one corner, theoretically

<joehillen>haha

<df_>that's slightly more likely than the apple thing I suspect

<mark_weaver>:)

<joehillen>oic, " Intel includes an FSP binary into the BIOS (Purism is working to have that binary freed). "

<joehillen>yeah, good luck with that

<joehillen>would it have been better if they went with ARM?

<mark_weaver>there seems to be a lot more hope in the ARM world, if you choose components carefully.

<df_>and avoid raspberry pi like the plague

<mark_weaver>anyway, I have to go afk for several hours. ttyl!

<davexunit>bye mark_weaver

<df_>olimex seem to be genuinely committed to getting to a situation where everything that runs on their hardware is free though (imo)

<fchmmr><mark_weaver> heh, ask about the librem on #libreboot :-/

<fchmmr>yes, eventually I'll run out of things to say about how criminally incompetent and sucky Todd Weaver is

<joehillen>fchmmr: that's not fair, are you building any hardware that promotes freedom?

<fchmmr>Ivi Inc.

<fchmmr>ivi.tv <-- look on web.archive.org at this site

<fchmmr>Todd's old company.

<fchmmr>proprietary software (video streaming)

<fchmmr>company dissolved in 2007, until then it was still serving people

<df_>joehillen: well yeah, he kind of is

<fchmmr>2015-2007 = 8 years

<fchmmr>Todd says he's been a free software supporter for 15 years.

<fchmmr>That's bullshit, just for that fact alone.

<fchmmr>Management Engine <-- impossible to replace

<fchmmr>FSP <-- will take years to replace

<fchmmr>Video BIOS <-- ditto

<fchmmr>Microcode updates for CPU <-- impossible to replace

<fchmmr>All of these 4 blobs above are present on the librem.

<fchmmr>When Todd says he "talked with Intel", he actually spoke with a sales person.

<fchmmr>He doesn't show up at any free software conferences, because he's not stupid enough to do that.

<joehillen>I don't dispote those, but my point is that it's easy to critize, it's infinitely hard to make shit happen

<fchmmr>Instead, he appeals to all the news/media outlets which are ignorant and do not care about free software.

<fchmmr><joehillen> I don't dispote those, but my point is that it's easy to critize, it's infinitely hard to make shit happen

<joehillen>*harder

<fchmmr>Todd isn't "making shit happen", he's taking your money and laughing while he does it.

<fchmmr>He's just a small-time petty con artist.

<fchmmr>That's all there is.

<fchmmr>Everything he's doing with puri.sm is fraud. He could even go to prison for what he's done.

<fchmmr>ah

<fchmmr>it's not ivi.tv

<fchmmr>ivi.me I think. let me find it

<fchmmr>yep. ivi.me

<fchmmr>ok, not that

<fchmmr>fuck, let me find it on wikipedia

<fchmmr>actuallyp, it is ivi.tv

<joehillen>well, then look at it this way, he's demonstrated that there is at least some market for a consumer product that promotes freedom. now all somebody needs to do is actually create something that delivers on its promise

<fchmmr> https://web.archive.org/web/20110306140609/http://www.ivi.tv/

<fchmmr>there you go

<fchmmr>ahaha

<fchmmr>it's not 2007

<fchmmr>it dissolved in like 2012

<fchmmr>He sold it, now it's not run by him anymore.

<fchmmr><joehillen> well, then look at it this way, he's demonstrated that there is at least some market for a consumer product that promotes freedom. now all somebody needs to do is actually create something that delivers on its promise

<fchmmr>Well, sure

<fchmmr>Actually, no.

<fchmmr>lemote showed that there is a market for it

<fchmmr>then gluglug did

<fchmmr>what puri.sm did was just exploit people who were desperate for something faster that was also libre, and then conned people out of their money

<fchmmr>using all kinds of claims

<fchmmr>Most people don't understand what to look for to know if a system is free or not.

<fchmmr>Anything Intel beyond 2008 will most likely not be libre any time soon.

<fchmmr>The name "librem" itself is misleading.

<mark_weaver>yeah, it's the false claims that I find the most objectionable

<mark_weaver>I don't think the librem is promoting freedom. instead, it's tempting people to abandon their software freedom, by choosing librem instead of machines like the Libreboot X200 that really can be run without several megabytes of blobs and backdoors.

<df_>it's pretty much cashing in on people's desire for free hardware

<df_>see also: raspberry pi

<mark_weaver>and part of the method Todd is using is by deceiving people into thinking that the librem might later be fully free, when in fact that's extremely unlikely

<df_>(I have a bee in my bonnet about that project)

<mark_weaver>(at least before the machine becomes obsolete)

<mark_weaver>anyway, gotta go afk again...

<joehillen>to change the subject, what is the "policy" for adding less than free packages to guix? would there be an objection to adding virtualbox for examples? or steam?

<mark_weaver>guix is committed to following the GNU FSDG

<mark_weaver>virtualbox requires a non-free compiler (Open Watcom) to build the BIOS, so we cannot add it because it steers people toward non-free software

<mark_weaver>steam is definitely out

<joehillen>personally I don't see a problem since the code doesn't live in the Guix repo, but the packaging/build instructions do. is that incompatible with FSDG? preventing me from installing proprietary software?

<mark_weaver>however, guix is designed to be hackable, so it's easy to add your own stuff in your private branch

<joehillen>but I would have to fork guix

<mark_weaver>I run guix out of my private branch, which has some modifications that are not yet ready to upstream.

<mark_weaver>I periodically rebase my private branch onto upstream.

<mark_weaver>it's easy

<davexunit>joehillen: we support an environment variable called GUIX_PACKAGE_PATH

<mark_weaver> https://www.gnu.org/distros/free-system-distribution-guidelines.html

<davexunit>that you can use to point to your own package modules

<mark_weaver>sure, that's another way to do it

<mark_weaver>maybe some people prefer that way

<davexunit>so you do not have to fork guix.

*mark_weaver goes afk

<fchmmr>mark_weaver, Todd actually claimed once that he had been in contact with the libreboot project.

<fchmmr>This is bullshit.

<joehillen>alright, that could work

<fchmmr>I know that for a fact.

<fchmmr>Because I *run* the libreboot project

<davexunit>guix is hackable, do what you'd like with it, but we cannot accept patches to add proprietary software.

<davexunit>it would compromise our ethical and technical goals.

<joehillen>well, be clear here, you aren't adding proprietary software, you are refusing build instructions for proprietary software

<fchmmr><joehillen> but I would have to fork guix

<davexunit>since proprietary software cannot be built with our toolchain, it can never be reproducible, and users would have to trust a third party.

<fchmmr>joehillen, yes, that is what you have to do.

<fchmmr>If you want proprietary software, you'll have to fork Guix. Guix upstream will never accept proprietary software, or software that encourages proprietary addons, or scripts (metadata) that download and build proprietary software, etc.

<davexunit>one of the technical goals of guix is to ensure that no one has to trust a single third party for binaries

<fchmmr>is it me, or does #guix have more people in it than #trisquel?

<davexunit>heh

<fchmmr>I can see some potential improvements to the website

<fchmmr> https://www.gnu.org/software/guix/

<fchmmr>It doesn't very clearly distinguish between guix the package manager and guix the GNU/LInux distro

<davexunit>well, they are one in the same.

<fchmmr>The documentation looks quite involved: https://www.gnu.org/software/guix/manual/guix.html

<fchmmr>How do I download it in git?

<fchmmr>How do I send patches?

<fchmmr>ok.... it's lower down on the homepage

<davexunit>see https://gnu.org/software/guix/#contribute

<davexunit>yeah

<davexunit>there's also some stuff about it in the manual

<davexunit>and we have a HACKING file in the source tree

***boegel|afk is now known as boegel

<fchmmr>how do I submit changes to the documentation?

<davexunit>fchmmr: edit the texinfo source and send a patch to guix-devel@gnu.org

<fchmmr>where do I find those?

<davexunit>in the guix git repository

<davexunit>in the doc/ directory

<fchmmr>ok, cloning now

<davexunit>cool

<fchmmr>(I actually am genuinely interested in submitting patches to guix at some point in the future, especially when I start to use it)

<fchmmr>(but what I'm doing here is gauging how easy it is for someone to download and install guix - and how obvious it is how to contribute, for people who want to do that)

<fchmmr>doc/guix.texi?

<davexunit>emacs 24.5 is out!

<davexunit> I had no ida1

<davexunit>idea*

<davexunit>wooooo!

<mark_weaver>at present, guix still has a lot of rough edges, and is mainly useful for developers who are willing to fiddle a lot to get it working nicely

<davexunit>fchmmr: yes, that should be it.

<fchmmr>wow

<fchmmr>this has the same problem libreboot used to have

<fchmmr>everything is 1 file, 1 page ;)

<fchmmr>(as far as documentation goes)

<mark_weaver>but on the plus side, once you get it working, you don't have to worry about breakage the way you do with other bleeding edge distros because of the rollback features

<mark_weaver>yes, we need better docs

<davexunit>I think the manual could stand to be split into a few source files at least.

<fchmmr>I'm watching the fosdem15 video

<davexunit>it's my favorite guix talk yet.

<rekado>joehillen: at work some people use a few applications that impose non-commercial restrictions; GUIX_PACKAGE_PATH is sufficient for these cases. The Guile module containing those recipes is completely separate from the Guix repo.

<rekado>I should say, though, that some developers are even willing to consider relicensing. (Already two agreed to a license change, but haven't yet made their decision public/quotable.)

<davexunit>rekado: oh wow, that's good to hear!

<rekado>Actually, one of the cases was a GPL violation and has already been made public; the other is code released under Artistic 1.0 and the author has agreed to relicense under Clarified Artistic.

<rekado>but anyway, I made it a point to ask.

<rekado>It seems to me that many people don't really licensing serious, so it is possible that their license choice was simply uninformed.

<rekado>*take ... serious

<mark_weaver>rekado: s/serious/seriously/

<mark_weaver>rekado: thanks very much for your work on that! it's very encouraging :)

<rekado>mark_weaver: thanks for fixing the adverb ;)

<mark_weaver>np!

<civodul>indeed, excellent work rekado!

*civodul looks at the netpbm issue

***boegel is now known as boegel|afk

<Steap>mark_weaver: I'm saying that the employers, who have the money, are completely dominating the working class, who actually needs to eat, yeah

<Steap>mark_weaver: FWIW, I managed to make a living out of writing Free Software, so I got that going for me

IRC channel logs

2015-04-10.log