<mark_weaver>the build phase of 'nss' took just under 200 seconds on my X60, and the check phase has been going for over 3 hours, and is still in progres <suitsmeveryfine>mark_weaver: when I try to build I get this: guix system: error: could not find bootstrap binary 'guile-2.0.9.tar.xz' for system 'x86_64-linux' <mark_weaver>suitsmeveryfine: it sounds like you didn't run "make" <Jookia>Just normal 'make' - how are you building Guix? <Jookia>Run "./configure --localstatedir=/var" <Jookia>Are you in 'guix environment guix'? <suitsmeveryfine>oops, no I must have exited. I did run all of the above commands within the environment before though <suitsmeveryfine>make[3]: *** Ingen regel för att skapa målet ”../../gnu/packages/weechat.scm”, som behövs av ”guix-packages.pot-update”. Stannar. <suitsmeveryfine>In English: *** No rule to create the target ”../../gnu/packages/weechat.scm” needed for ”guix-packages.pot-update”. Stops. <mark_weaver>suitsmeveryfine: delete po/packages/guix-packages.pot and try again <suitsmeveryfine>mark_weaver: thanks for trying to help me out but I give up for today. <Jookia>suitsmeveryfine: Hey could you paste the output of 'git status' <Jookia>well, you clean run 'git checkout -- po' to reset that, then run bootstrap and configure again then make <Jookia>Is there anything in particular about a guix-packages.pot-update file <Jookia>Could you re-run the bootstrap script and configure with the args then make? <Jookia>Can you git add yor changes, stash then then run 'git clean -dfx', then bootstrap, then configure, then make? <Jookia>shouldn't matter, but from the root of the repo <mark_weaver>suitsmeveryfine: looks like the line containing "gnu/packages/weechat.scm" needs to be removed from po/packages/POTFILES.in <mark_weaver>and these translations need to be updated accordingly <mark_weaver>sneek: later tell civodul: po/packages/POTFILES.in contains "gnu/packages/weechat.scm" which no longer exists, and this caused the following error for someone: *** No rule to create the target ”../../gnu/packages/weechat.scm” needed for ”guix-packages.pot-update”. Stops. <Jookia>who will eat our supply of botsnacks now <mark_weaver>ACTION waits for sneek to come running home upon hearing of botsnacks... <Jookia>I wonder when sneek was last here <Jookia>mark_weaver: So I've heard that the ARM U-Boot support works in QEMU, so I may try to port Guix to that as well as EFI <Jookia>Right now the patch I have supports traditional BIOS and coreboot (well, libreboot+autoboot+hypothetical coreboot) <Jookia>I got the Libreboot and Autoboot devs to agree on using 'coreboot_grub.cfg' (though it still symlinks) and I've heard talk of them pushing the fancy grub menu to coreboot <vimuser>the coreboot grub.cfg is in grub, not coreboot <Jookia>Wow, that's going even higher upstream than I thought <vimuser>I'll send a patch to grub-devel later <vimuser>not exactly huge priority at the moment <Piece_Maker>guys, ive got a slightly nooby question... im trying to add a second user to my guix desktop.scm.. ive basically copied everything from the desktop.scm default from (cons (user-account) down to the (home-directory "dir")) - is this enough, or do i need to add more to that? <mark_weaver>Piece_Maker: it should look like (users (cons* (user-account ...) (user-account ...) %base-user-accounts)) <mark_weaver>we should probably use 'cons*' in our examples, so that it doesn't need to changed when more users are added. <Piece_Maker>hmmm im still getting scheme error... can i paste that section on a pastebin and show you? see if theres anything dumbim doing wrong here... <mark_weaver>okay, but not pastebin.com please, because that site blocks tor users <Piece_Maker>it was an unrelated block that was erroring it, not the users block ;) <Piece_Maker>probably should for the time being add one new config part at a time to make sure im not doing too much at once while i still learn ;D <Piece_Maker>quick query too - say i want both my users to have, eg. GNU icecat package... is it recommended to install it as root, or should i install it seperately on both users? <mark_weaver>I guess it's a matter of preference whether to put it in the per-user profiles or in the 'packages' field of the OS config (which is not the same as installing as root, btw) <mark_weaver>one advantage of per-user profiles is that each user has control over their setup, and is able to roll back if there are problems, etc. <mark_weaver>but maybe if you are trying to support another non-technical user who doesn't want to fiddle with these things, maybe it makes more sense to put in the system profile, dunno! <Jookia>System profile is declarative too <mark_weaver>to be clear, if root runs "guix package -i icecat", that command is only made available in ~root/.guix-profile <Piece_Maker>thats why it goes in the OS config as opposed to a root install <Piece_Maker>i think for 'core' applications ill do it in the OS config as my dad is 'non-techical' but generally knows what he wants... so ill let him do his own thing but put some standard defaults in the config ;D <Piece_Maker>he rarely uses the machine anyway... enough to have his own login but not really enough to know or care what OS its running <mark_weaver>one issue is that for things like web browsers, it's especially important to keep them up-to-date w.r.t. security vulnerabilities. <mark_weaver>so if he's not going to stay on top of that, it might be better in your case to include icecat in the system profile <mark_weaver>and btw, I should mention that "guix pull", which is roughly analogous to "apt-get update", only updates guix for the user who runs it, so each user who runs "guix" needs to run "guix pull" to get the updated package descriptions. <Piece_Maker>so how does one update the software in your system profile? guix pull + reconfigure? <Jookia>If you're going to instruct your dad to update it maybe it'd be best to make it so he can do it as a user <mark_weaver>that means that you need to run it as your normal user so that packages installed in your user profile are up-to-date, and you also need to run it as root, because root is the user who must run "guix system reconfigure" to update the system-wide profile <Piece_Maker>Jookia: ill be keeping on top of the system updates, im not that evil :P <Piece_Maker>so, (as root) guix pull + guix system reconfigure = updates the stuff in OS config.scm... guix pull + guix -U = updates stuff in my own users /home <petter>Serveraptor (VPS provider) is asking for a qcow2 of a clean GuixSD installation with DHCP disabled <petter>is anyone here interested in producing this? <petter>i can't because my computer crashes when i run QEMU :/ <Piece_Maker>great ;D thanks for all that, ill have some fun messing with this then se if i can get a good working desktop ;D <paron_remote>there's a solution for that now in guix master, assuming you set the right kernel flag <paron_remote> (kernel-arguments '("modprobe.blacklist=kvm_intel")) ; also ",kvm"? <petter>hehe, yeah, i've booted into the 70's a few times too many now <mark_weaver>petter: as for DHCP disabled, just make sure that (dhcp-client-service) isn't there, and use 'static-networking-service' instead. <petter>paron_remote: sure, i'm very interested in this <mark_weaver>np! if you get it working, it would be great if you could send us a message letting us know the details of how you did it, and maybe an example config :) <Jookia>WIthout KVM it takes around 15 minutes to build a base system VM <Jookia>I wonder why QEMU is used rather than a 'real' disk builder system <suitsmeveryfine>I removed evdev from gnu/services/xorg.scm and reconfigured but it didn't solve the touchpad issue. Synaptics still doesn't want to load. <paron_remote>because everyone else has kvm other than us x200 libreboot users? :) <mark_weaver>suitsmeveryfine: there are two places where it needed to be removed: from %default-xorg-modules and also the line: ModulePath \\"" xf86-input-evdev "/lib/xorg/modules/input\\" <Jookia>paron_remote: Well, I think there are other ways to build a system especially if we're jsut copying files. Mounting the drive, using libguestfs, etc. Maybe it was just cooler this way? :) <paron_remote>ACTION assumes everything suits suitsmeveryfine very fine <mark_weaver>suitsmeveryfine: well, this is not an area of expertise for me, but at this point it would be easy for you to try various other things. running "guix system reconfigure" after merely changing gnu/services/xorg.scm will be very fast. <mark_weaver>if you see in there which drivers are being loaded, and which one is trying to take responsibility the trackpad, maybe try removing that one? dunno, my knowledge here is weak. <mark_weaver>wingo made some suggestions above, maybe worth looking over again. ***koz__ is now known as koz_
***koz__ is now known as koz_
<rain1>would this be possible to include in guix? <Jookia>mark_weaver: Coreboot patches sent to mailing list as an RFC <Jookia>rain1: Could be useful, try packaging it <Jookia>rain1: Noto is a trademark, I wonder what that means <rain1>where tofu is the boxes you get when you dont have fonts for unicode symbols <rain1>i don't know, maybe best to avoid - working on symbola now ***Satyr1con is now known as Satyricon
<SusWombat>I think i asked that in the past. But i forgot what i got as answer :/ Does guix plan in the long term to go hurd? or does it stay on linux? <efraim>I think the plan is to support both <Jookia>Coreboot patches on mailing list woo <jmd>Any chance of a Hurd based GuixSD soon? <Jookia>jmd: There's a hurd-wip branch, you could try that :) <civodul>jmd: not soon, but that's a likely GSoC project ***civodul changes topic to 'GNU Guix | http://gnu.org/s/guix/ | videos at https://gnu.org/s/guix/help/#talks | 0.9.1 is in the works | https://libreplanet.org/wiki/Group:Guix/GSoC-2016 | avoid pastebin.com, it blocks Tor | channel logged: <https://gnunet.org/bot/log/guix>.'
<Jookia>It's a shame Wget2 isn't a GNU project <NiAsterisk>what's up. this powwow software is fun in licenses: public domain, gpl2, gpl2+ and gplv3 in headers. I searched for a newer version, found one, it now includes gplv3.. what do you do in this case? name every file :D? <Jookia>NiAsterisk: public domain or not licensed? <NiAsterisk>also, our "lynx" webbrowser is build without support for HTTPS URLs. why? <NiAsterisk>wait. it might be that cc0 is gone, i have to check this for every file. so if it's a combination of gpl2,gpl2+,gpl3, I do (license: gpl2 gpl2+ gpl3 and name all the indivual files? can't be right <Jookia>Hmm, what other projects are multilicensed <NiAsterisk>wild. can they do that? this is strange... the 1998 version of catrw.c was public domain, now it is gpl2 <NiAsterisk>changelog says "update to gpl2+" with no specific file, so I assume it's gpl2+ now and some newer ones are gpl3 <iyzsong>wget2 look cool, it list socks5 proxy as its #1 issue (github), that's the one I always want. <NiAsterisk>Jookia: I see what they did. the previous pd files have now a gpls2 with the addition of (created by: name email date) before the gpl2 header starts. <Jookia>So it's 10PM and I'm working on getting LVM+LUKS to play nice <NiAsterisk>i think what I should do is list it as gpl2+ and gpl3+ and name the ones which are gpl3+ :) <Jookia>I figure once I get some hacks set up and a dev environment I can clean this up <civodul>they could have chosen a different name <Jookia>civodul: I think it's the same developers as wget <civodul>dunno, i haven't seen this name befoer <Jookia>civodul: Have any thoughts on handling mapped devices that depend on mapped devices? I'm unsure if adding a 'dependencies' field is the right way to do this <NiAsterisk>hm.. if I have gpl2+ and then there's gpl3+, gpl2+ is somewhat inherited, covered by, included into, transfers to (sorry I lack the word to describe what I want to say) gpl3+ ... so I can just simply state the licenses and name no files <civodul>Jookia: we do have a 'dependencies' field in 'file-system' for similar reasons, but i agree it's not super elegant <Jookia>Perhaps mapped devices could be a special type of file system? <Jookia>Automatically building a dependency graph would be interesting but I don't know if that'd be a good idea <civodul>but maybe dependencies just cannot be determined automatically? <Jookia>Perhaps, though the code already attempts to do this somewhat <NiAsterisk>what I don't get is install-sh ... it is # Copyright (C) 1994 X Consortium and goes on with FSF at some point. licenses can be weird if you just know basic copylefts.. <civodul>Jookia: rriiiight :-) but only between file systems and mapped devices <Jookia>civodul: Yeah, so it breaks down when I have a LUKS->LVM->Btrfs setup. Or (LUKS,LUKS)->(LVM RAID)->Btrfs setup <efraim>NiAsterisk: re lynx: these things happen sometimes, our version of irssi also used to be built without support for https <Jookia>I think my temporary solution is to just map all block devices at boot ;) <Jookia>At least until I figure this out some more <NiAsterisk>3 files with gpl3+ , rest gpl2+ . problem solved I'll submit the patch any minute. <NiAsterisk>is it okay if I use "MUD" in the description or is this something completly unknow these days? <Jookia>Maybe Guix packages should have tags? ;) <rekado>NiAsterisk: if it's an abbreviation we normally would do something like this "foo is a @dfn{Multi-User Dungeon} (MUD) ..." <Jookia>LVM devices found, but not enabled! :) One more hack should do it <iyzsong>I thought MUD and IF are cool, but haven't get change to enjoy them (my English, time, etc.) <roelj>I'm trying to build a package, which fails with: /gnu/store/5s9lvpwiszryhv3zvqm35fayyvmrr2px-python-3.4.3: bad interpreter: Permission denied. And I don't know what this means in Guix.. I thought it could either not access python at all, when it's not part of the inputs, or access it normally when it is part of the inputs. <rekado>NiAsterisk: in the synopsis we don't do this. <NiAsterisk>okay. I used MUD in the synposis (and telnet etc) and explained MUD in the description <rekado>roelj: is this using autoconf? If so you can take a look at the exact test in config.log <Jookia>civodul: LVM can also create MULTIPLE targets that file systems might use, so we can't autodetect dependencies anymore :( <roelj>rekado: Thanks. I was messing up, my Scheme code points to a directory, not to the interpreter. <roelj>rekado: A completely different thing. Do you think the ant-build-system and maven-build-system should install jars in a maven repository layout/structure? <rekado>and I'm not leaning towards either side. Completely undecided. <wingo>does libreoffice work in git master? <rekado>the ant-build-system probably should not install them in a maven repo layout. <rekado>(because: how could it know where to put it?) <rekado>roelj: do you know what other distributions do? <roelj>rekado: No, I don't. But I could certainly investigate :) <wingo>ACTION attempts to understand hydra's web interface <civodul>wingo: LO is blocked by a build failure of Vigra, a dependency <civodul>Jookia: re LVM and multiple targets yeah, that was my understanding <civodul>so yeah, that part probably needs to be changed <Jookia>civodul: I'll try and figure out what to do about it <Jookia>Actually I think I can fix the 'address@hidden' <Jookia>gnu lists delete anything beginning with #$@ <civodul>yeah but gmane should have the exact text <civodul>or you can download the raw mbox file from lists.gnu.org somewhere <NiAsterisk>I can take a look, I use gmane for guix interaction <NiAsterisk>Date: Wed, 15 Apr 2015 07:07:56 +0200 by From: Tomáš Čech ? <NiAsterisk>i couldn#t give the html link to gmane as I use it with Gnus, but good that you did find it <Jookia>civodul: If we go crazy could we start thinking about this as functions that take an input + expected output of a file or directory? (luks /dev/sda hdd -> /dev/mapper/hdd), (lvm /dev/mapper/hdd matrix -> /dev/mapper/matrix-root,/dev/mapper/matrix-swap,/dev/mapper/matrix-scratch), (btrfs /dev/mapper/matrix-root -> /) <Jookia>civodul: This would also fix the issue of dependencies since you could just match outputs to inputs <civodul>at this point you may have a clearer idea about it than me ;-) <civodul>so if you come up with ideas on how to model this on the Scheme side, don't hesitate to share them on guix-devel <Jookia>I certainly will. Any attempt to cut down on complexity is good news to me :P <civodul>(i'm a bit lagging behind at the moment, but the message won't get lost) <Jookia>I think I almost have basic LVM patches done but I think we might be missing the 'dm-mod' kernel module <Jookia>Wow, that's a lot of stuff to read and review <civodul>looks like no single person can read and process all of it now <NiAsterisk>yes... partly why I switched from nnml to newsgroups <Jookia>I feel like I'm the only person that gets somewhat excited about reading email <Jookia>I suppose it's because it's not my responsibility to reply or review :P <NiAsterisk>I had ~60 lists, and many low frequented, but it was still something around 300 messages/day, and if gmane can do the filtering, good.. now I'm down to maybe 5-15 per day <Jookia>Oh fantastic I made the same typo multiple times in a terminal then managed to typo it in to the code that runs my Guix init <Jookia>civodul: I know this is possibly out of the question but at the guile prompt can I get an 'ls' command? <Jookia>Well aren't I the genius, the mount command kept saying 'no such device' but it turns out I formatted root as ext4 instead of btrfs. <rekado>Jookia: there's the bournish language <rekado>it should give you an ls command <rekado>Jookia: see also guix/build/bournish.scm <Jookia>When init fails it doesn't seem to jump to the bournish language even though the code indicates it would <Jookia>Maybe that was just an import, not sure <rekado>Should openblas be built with openmpi support? <rekado>Users here see weird errors when running numpy on the cluster in non-interactive mode. <rekado>the error is OpenBLAS trying to spawn more than 1 million threads, which doesn't seem right. <rekado>I wonder if we need to build openblas with openmpi support to make sure it can be run in parallel. <Jookia>Guess who's booted from LUKS->LVM->ext4 (gave up on btrfs for now) <rekado>Jookia: I didn't quite understand the problems wrt "targets" before. What does this mean? <suitsmeveryfine>Hi! I just reconfigured the system last night without any hacks and now the macbook2,1 touchpad works very well in GNOME 3. <suitsmeveryfine>But I've rebooted three times and it worked every time. Now I have two-finger scrolling and three finger tapping for middle click <Jookia>rekado: Basically mapped-devices can't depend on each other and it's problematic for things like LUKS and LVM. The problem is that mapped-devices don't really have a 'target' that works well, and in some cases may require multiple inputs or multiple targets <Jookia>rekado: I wrote a quick idea on how to unify filesystems and mapped-devices in to a system that allows this and has dependency resolution for free <wingo>suitsmeveryfine: "without any hacks" what does that mean? :) <Jookia>rekado: My current hack to work around all this is to remove dependency resolution and just load all mapped devices then filesystems, with an LVM mapped filesystem thing added <suitsmeveryfine>wingo: it means I didn't have to remove any drivers from xorg.scm which I first attempted earlier yesterday evening. <Jookia>Is there a way to copy a store to another partition, or use it to download things from <suitsmeveryfine>I believe that libinput is now being used instead of synaptics, because synclient commands do not work. <pizzaiolo>so compiling guixSD from source did it for you? <suitsmeveryfine>pizzaiolo: I don't think that did it actually since it worked without any hacks on my part. <suitsmeveryfine>including all of the features that I need, like two-finger scrolling and middle clock <suitsmeveryfine>pizzaiolo: It would be good if you try as well. Simply add 'gnome' to your OS definition and reconfigure. Then, inside GNOME, take a look at the mouse settings <wingo>ACTION should try gnome too, i haven't done it yet <NiAsterisk>i think I do 2 or 3 more packages and then I'll take a look at awesome-wm etc on my list.. <Jookia>I need to rewrite my firewall rules in Guile since I don't understand them (I barely did in the first place) and as a result I now have a permanent proxifier for my second Ethernet port ***mv is now known as marxistvegan
<Jookia>rsyncing 1.3TB of files at 2MB/s is sure going to be a fun week <roelj>civodul: Thanks for the article :) <roelj>Cool. I'm not completely sure how grafting works, but I can definitely see its power. <Jookia>roelj: It's a bit like copy-on-write <rekado>Jookia: "cp -a" is faster than rsync. <rekado>bleh, yet another small bioinfo tool that bundles so many slightly modified versions of upstream libraries :-( <rekado>there's no way I can package that <rekado>bundling is bad enough but forking libs (and never applying patches any more) is so much worse. <Jookia>rekado: You'd love anything done by Google then! <rekado>yeah, same thing. I won't touch Chrome/Chromium. <Jookia>Why would you ever contribute upstream? :P <roelj>rekado: It seems to be a bioinformatics thing.. Which program is it? <rekado>already talked to upstream about this, but it's unlikely they'll change. <roelj>rekado: I see. That's a pity <rekado>paron_remote: argh, the comments! <rekado>"GNU doesn't like dynamically linking to libraries, instead preferring to statically link all the code." <rain1>the people on slash dot do not seem very intelligent <Jookia>Never read the comments on anything ever <paron_remote>and mostly go to the comments to see "what entertaining junk comments will there be on slashdot today" <rain1>there is a site called soylent news which is supposed to be better but I don't think is <rekado>"The GNU folks took an existing package manager, Nix [nixos.org] and modified it to use Guile Scheme" <rain1>might be! it's a bunch of people that quit slashdot because the changed the UI or something <Jookia>paron_remote: Oh I use to do that but with reddit, but then it just got me angry since people read reddit <Jookia>I wish we could invent some kind of CAPTCHA but for testing whether people have actually read the link they're commenting on <rain1>it makes me a little disappointed that p.eople still use wordpress <rain1>it's just so clear nthat a huge group of people will never stop doing the things that are harmful <civodul>the /. comments are terrible, indeed <paron_remote>console yourself in that the /. comments are always terrible, at least :) <civodul>the wget2 page about GNU was not much better <civodul>it would take a whole marketing team to improve GNU's perception <paron_remote>and it would also take a whole brainwashing team to improve slashdot commenters :) <rain1>what I've often wondered is, where is any interesting/good discussion online? <rain1>since it's not /. or reddit.. <rain1>there was lambda the ultimate which is a bit fun, could maybe share some guile news there <rain1>its not as lively as it used to be <amz3`>not sure it's the good audience <rekado>maybe we should make a fundraiser to pay for a marketing team to improve GNU's perception then...? <Jookia>Why do people not like GNU again? <civodul>the wget2 folks don't like it because of the ugly web pages and licensing discussions (!) <civodul>the /. commenters because of static linking (?!) <rekado>some people claim GNU suffers from NIH syndrome and has to has its own thing of every successful thing but with "GNU" stamped on it. <Jookia>civodul: I think they don't know what functional package management is and thought that everything is statically linked <rain1>one reason people don't like GNU is some of the code is really long <rain1>(if you look at old gnu programs) <rekado>oh yeah, there are people who hate "ls". <civodul>all in all, there's a lot of miscommunication :-) <Jookia>[insert comparison to BSD vs GNU 'true'] <rain1>gnu cat vs plan9 cat is a good one <Jookia>Still, GNU Guix and GNU Mediagoblin are two projects off the top of my head that feel 'modern' <Jookia>The GNU Social network's had some tough times recently, though it would be nice to package it as a service for Guix <rain1>it's too bad they're PHP but they're attacking really important problems <Jookia>A lot of 'bloat' in GNU tends to be stuff learned from decades of use <Jookia>Often you can see projects that reinvent things (cmake for one) and create more problems than they solve <rain1>one thing that still baffles me is ./configure <rain1>the scripts can come out to sizes like 10k lines, and they often check very strange things like whether i have 'strcat' function <rain1>it seems like there's baggage in them that could be removed <rain1>most of the info they detect isn't used and could be cached anyway <Jookia>I've heard a lot is copy pasted but I'm unsure <CompanionCube>ACTION thinks that autoconf/autotools in general is a good idea....but the implementation likely deserves to be thrown in the fire <CompanionCube>the huge ./configure and such scripts generated by M4 macros just feels like it could be done better <rain1>that's what I feel too, but then I don't think cmake improved things - just produced more docs to read to get things done <rain1>so it's a hard problem. guix makes things extremely smooth and nice though <bavier>in my experience the autotools are more hackable, in part because one can always fall back to plain sh or make. <paron_remote>bavier: "falling back" on plain sh usually means "falling back" on m4sh though, right? <paron_remote>> You may be wondering, Why was Autoconf originally written? How did it get into its present form? (Why does it look like gorilla spit?) <bavier>paron_remote: yes, depending on the level of portability desired <rain1>how would I use functions from ./guix/hash.scm in my own program? <rain1>(for now I'll just copy that bit of code over) <paron_remote>don't think we have many "production" postgres users yet but <paron_remote>> To cope with these kinds of requirements, configuration-management systems tend to invent a domain-specific configuration language that (eventually) becomes Turing complete, starting from the desire to perform computation on the data in the configuration (e.g., to adjust the amount of memory to give a server as a function of the number of shards in the service). The result is the kind of inscrutable "configuration is code" that peopl <paron_remote> were trying to avoid by eliminating hard-coded parameters in the application's source code. It doesn't reduce operational complexity or make the configurations easier to debug or change; it just moves the computations from a real programming language to a domain-specific one, which typically has weaker development tools (e.g., debuggers, unit test frameworks, etc). <paron_remote>> We believe the most effective approach is to accept this need, embrace the inevitability of programmatic configuration, and maintain a clean separation between computation and data. The language to represent the data should be a simple, data-only format such as JSON or YAML, and programmatic modification of this data should be done in a real programming language, where there are well-understood semantics, as well as good <paron_remote> tooling. Interestingly, this same separation of computation and data can be seen in the disparate field of front-end development with frameworks such as Angular that maintain a crisp separation between the worlds of markup (data) and JavaScript (computation). <rain1>yeah, people slowly start to realize <jmd>When I run guix --install I just get : <jmd>repeating ad infinitum <paron_remote>heya jmd, sorry to hear it.. I've had it hang there before sometimes <jmd>Well I get a different error :) <rain1>what about that other mirror? <jmd>Perhaps I need to initialise the acl <NiAsterisk>"rain1 what I've often wondered is, where is any interesting/good discussion online?" mostly in the places where few people go to. <NiAsterisk>"Jookia Why do people not like GNU again?" the most comments I get are "stallmans / gnu project visions are detached from reality of market etc etc.." up to the point where I had to criticize an author on the lack of research on (among other things) free software in general "people are afraid of open sourcecode" "all technology in the past 25 years was build for surveilance / etc etc". <NiAsterisk>at some point I will translate my initial review of the book to english. it's so sad you won't even laugh about the lack of competence to search "what is free software -> gnu.org -> philosophy" etc <amz3>I have a possible GSoC idea, I think I already talked about last year but I don't remember what was the result of the talk <amz3>I throw the idea (again) here, to have feedback and maybe help me write a summary of the idea if it good enough <amz3>the idea is to create a computing grid of guix boxes <amz3>I am not very familiar or not familiar at all with existing solutions <amz3>so I'm not sure how this could be an improvement over existing solution <amz3>also I don't know what guarantees we can provide over the correctness of the result (has results been tempered somehow?) <efraim>build in parallel across the boxes and compare the results? <amz3>yes, do several time the same job you mean? <efraim>we already have --rounds and offloading, why not offload the rounds <amz3>why I think about this and guix is because with guix, the job provider doesn't have to create a specific program to be run on some cluster with a custom programming semantic instead he just have to use his everyday programs and programming paradigm, split the dataset if any, and provide the job to a guix cluster <amz3>I'm not talking about building programs, but running programs on guix boxes with guix packages <amz3>and retrieve the result of the program <amz3>the idea behind that is that guix org could then ask guixsd user to subscribe to the program to help research, we could maybe even give money to people helping the project but i don't beleive in this <amz3>anyway, another advantage I can think is that since the grid use conventinonal programs the speed of programs would be better... <amz3>you may say, why other distros didn't do it... you would pin a good point, but with guix reproducibility in mind... maybe it makes more sens for guix to do that <efraim>with BOINC programs they're already shipping you a pre-compiled binary to run on your machine <amz3>this doesn't look like really open-science <amz3>that's another argument for a guix grid, programs would be free software and reviewed <amz3>I was thinking about thing like BOINC indeed <amz3>this also could be later paired with guixops <amz3>this also could later be paired with guixops <amz3>paron_remote: what do you think about a "BOINC for guix" or "BOINC on guix" kind of software? <paron_remote>amz3: I think distributed computation of substitutes is a great idea and would pair well with the gnunet work, if that's also what you're talking of <amz3>yes, at some point gnunet would be nice to have too <jmd>How do I list package names? <rain1>or the emacs mode command guix-all-available-packages is even nicer <amz3>let's continue here you prefer, we can debug this and create the package in one go <rain1>guix environment guix --ad-hoc gettext autoconf automake <rain1>oh better idea: guix environment guile-cairo <rain1>I thought about making a guile library wrapping zenity <rain1>may be nice and easy for beginners <rain1>guix environment guile-cairo --ad-hoc gettext autoconf automake libtool texinfo <rain1>that gives a good env to build it <amz3>and what about g-wrap you did forgot g-wrap, no? <amz3>there is a package for g-wrap AFAIK <jmd>What package provides the crt1.o necessary to compile/link things ? <bavier>jmd: if you're working from your profile, you'll want to install gcc-toolchain <pizzaiolo>suitsmeveryfine: I won't be able to try gnome on guixSD tonight, only a few days from now <janneke>i tried it several times in a vm without luck <janneke>as it turns out, due to a silly gnome bug/thingy with video acceleration <lfam>What has happened to our sneek-y bot? <lfam>janneke: I had that same issue <janneke>guix system init on bare iron works fine <janneke>lfam: I posted to guix-devel with request for info... <lfam>janneke: Any insight into making it work? I might try digging around in the QEMU manual section on video <lfam>This channel is decreased without sneek :( <rain1>what needs to happen for sneek back? <janneke>lfam: yes, it seems that you need to start gnome session using <janneke>`gnome session --disable-acceleration-check' <lfam>Oh, and that works in a virtualized environment? <janneke>but I failed to patch the right file that actually executes gnome-session <lfam>Is that .xsession or alike? <janneke>lfam: from what I've read all over the internet (launchpad/stackoverflow) <janneke>about gnome-session and vnc, qemu etc; that should do it <lfam>Okay, I'll try that later today <lfam>It would really make it easier to review patches related to GNOME stuff if I had a GNOME system :p <lfam>efraim: Do I remember correctly that you tried packaging connman? Did that go anywhere? <efraim>i got distracted and put it down for a bit <janneke>on debian it is /usr/share/xsessions/gnome.desktop I think <efraim>I got it building nicely and wrote a service for it but never got around to testing it <janneke>ACTION really wonders why the GNOME folks don't care about vnc/qemu <lfam>efraim: Would you be willing to share it? I'm going to try converting one of my machines to GuixSD soon and I like connman over the alternatives <lfam>It's also helpful for me to read patches of services so I can get a better understanding of how to write them. This is great :) <pizzaiolo>idea: make a wiki so that anyone can contribute to Guix(SD) documentation <lfam>pizzaiolo: Anybody can contribute to the manual already :) We even take non-texinfo patches and clean them up! <rain1>i have a wiki on gitlab people can add info to <pizzaiolo>lfam: hmm patching is not quite the same, though <lfam>By "non-texinfo" I also meant "non-patch-formatted" <lfam>Anyways, we've had this discussion at least once before. I think if you search the guix-devel archives you can find people's arguments one way or the other. I would read that and then bring it back up if you think there is a new angle on the subject, or if you just really want a wiki. <lfam>My preference is to put it in the manual. I think that wikis are a second-best alternative to a comprehensive and up-to-date manual <lfam>I just hate digging around the web for information that should be included in a manual that comes with the software in question <lfam>paron_remote: white knuckles! <rain1>pizzaiolo, is it of any interest? <paron_remote>in any other system I'd wait for someone else to do it :) <rain1>its just for notes and stuff <lfam>paron_remote: We can always rollback but I really hate putting broken commits in the history! <lfam>Ideally a user could check out any commit and have a working system <koz_>I *loathe* only-only documentation in any form. <NiAsterisk>wow.. watching a video on the company RealDoll and with their research into AI implementation, it is moving to a scary, interesting direction almost Android like you see in the game "RememberMe"... weird future. <lfam>NiAsterisk: There is a documentary about those dolls and their owners. Fascinating! *VERY* bad date movie <NiAsterisk>lfam: i'm watching the episode of real future (i think that's vice.com) <NiAsterisk>i really wonder if this is a first preview on androids for the mass market... you never know. strange it is. <rain1>someone better get started on a free software eversion <NiAsterisk>and when it is, will it be another puzzle in the ever closing proprietary software future with internet of things which is largely ignored with its dangers by the general populus? <NiAsterisk>even if you talk to people and try to point out security risks they introduce, they are in denial. I tried it again last weekend... the app which controls the intrusion detection and smoke detectors in a house of someone I know who works for a larger company. <a_e>Guix Europe, here it comes: <lfam>a_e: I get a warning about your TLS certificate <a_e>lfam: Yes. So far, I did not bother to think about "let's encrypt". <a_e>Anyway, it is only public information. <rain1>Gégégé, get the gerogerigegege to open <rain1>i got the same warning but i think the cert is good <lfam>"And you will be the first in line when there is additional work to distribute!" Haha <fhmgufs>It's a pity that guix.org isn't available ... <NiAsterisk>a_e: the associazione guix europe, could this be used for for example getting all current copylefts signed over / dualcopyrighted to it? I have no idea why I ask this, there was a question which faded.. something with copyrights and like it is handled in gentoo with the gentoo foundation being the only name mentioned in the headers. <a_e>NiAsterisk: No, this is not a goal. If there were any copyright assignments to make, it should go to the FSF. <NiAsterisk>fhmgufs: guix.is would be... with my registrar it would be twice the price including optional privacy whois etc, but the normal price is just ~30 euro. <wingo>all hail the guix politburo! :-) <a_e>But our policy has been not to bother and to accept contributions without copyright assignments. It lowers the barrier for contribution. And everything is GPL anyway. <wingo>ah i didn't realize a_e was andreas. hello a_e :) <a_e>wingo: Hehe, so it was a more efficient anonymiser than I had expected :-) <wingo>graph coloring between irc nicks and real people; one node just coalesced :) <efraim>hmm, I might need to test and push a patch to core-utils adding vdpau to mesa <lfam>a_e: I only realized a couple weeks ago ;) <a_e>Time to change my nick... <NiAsterisk>w/should I have the enormous amount of money 10 years would cost with orangewebsite.com for an .is (compared to nic.is), I would opt to donate 10 years of guix.is to transfered ownership to guix. but that's like 689 € vs 300 € normal price. <a_e>NiAsterisk: Why should we use .is? I briefly considered guix.sd, but it is not a very common top level domain and not sold everywhere... <a_e>Look up the country! <NiAsterisk>out of all country tld's, I personally have a favor for the position of iceland vs usa and others. that's basically everything. <a_e>There is no country .ix. <lfam>Oh Sudan. Probably not the best choice for reliability <NiAsterisk>whoever owns that. and it would be longer than the name <efraim>the only fun .co.il that I can think of is tesla.co.il <civodul>NiAsterisk: +1 on avoiding .io, glad i'm not the only one on that boat :-) <NiAsterisk>i did not know about it until I went domain searching a couple of years ago <fhmgufs>.org is good. Just thought that guix is better than guixsd. <fhmgufs>I didn't know that there will be such a discussion. :) <NiAsterisk>domain names are always political attached in this current system, that's what .io shows <lfam>I'd rather not put our lot in with a specific nation-state <lfam>I don't think letsencrypt has anything to do with this <NiAsterisk>i think this can be discussed and voted on when there's need for a domain outside of gnu.org <NiAsterisk>personally i think some tlds should be avoided due to political reasons or endorsement, and some for technical reasons (.xxx and .sexy makes no sense for example) <suitsmeveryfine>I think if you use Let's encrypt for the TLS certificates you're not limited to Gandi's list of available top domains. That was my only point. <NiAsterisk>but gandi certs and gandi registrar are two pair of shoes? <lfam>Huh, I don't see how that would make a difference <lfam>I've never heard that metaphor before <a_e>wingo and others: I packaged a recent snapshot of vigra in their release candidate branch. Right now I am trying to build libreoffice with it. If it all works, I will send it tomorrow. <efraim>don't forget to update libreoffice to 5.1.1 for the 2 CVEs <a_e>On February 12, they mentioned there would be a release in two weeks. As time passes, I think it will be better to have a temporary package. <a_e>efraim: Well, that is a differrent topic. I might let you do it :-) <lfam>I don't use libreoffice but I think it's important to have it in the distribution. It's the sort of thing that drives away novice users who might be looking to switch from a proprietary tool <NiAsterisk>lfam: german metaphor "das sind zwei paar stiefel/schuhe" when two things are not comparable to each other <a_e>Let us hope it will take just a version and hash change. Otherwise, it could mean some work. <suitsmeveryfine>The .SD domain name rules state that "Domain Names must not Contain Words, Phrases or Abbreviations that are obscene, scandalous, indecent, or Contrary to Sudan Law & Religions." <lfam>suitsmeveryfine: Exactly. All these national TLDs have absurd rules <lfam>And you can't trust the security of the registrar itself in most cases <a_e>efraim: It would be nice if we could build libreoffice in parallel. Maybe we should give it another try. <NiAsterisk>not endorsing it, but from what I read iceland is okay though.. and .pw <lfam>NiAsterisk: It's amazing how close that is to English <lfam>NiAsterisk: The only non-cognate is "zwei" <lfam>suitsmeveryfine: Is that possible? <NiAsterisk>from what I read, my registrar is run by someone who favors freespeech etc and its reflected in their policies. <suitsmeveryfine>Yes I think it's possible to register top domains now. I think it's really expensive though. <a_e>suitsmeveryfine: 185000$ just for applying at ICANN. <lfam>NiAsterisk: Also stiefel, but if schuhe is an alternative, that's close to shoe <lfam>a_e: I think we raised about that much in our fundraiser, yes? ;) <a_e>For everyone applying for the same TLD, that is. <NiAsterisk>suitsmeveryfine: it can be /really/ hard to get this through <a_e>These generic TLD are an absurd money waster, or maybe money making machine. <a_e>My opinion is that we are best served with .org. <a_e>It translates exactly what we are and do. <NiAsterisk>ICANN is a money machine, and I'm happy when .gnu will start to be around. <lfam>Yes, this is an epic bikeshed <a_e>A bit of good-natured bike shedding can be a fun activity. <lfam>Indeed! I like some of these suggestions. guix.system *is* pretty nice <paron_remote>I forgot a_e was andreas too, though I went through that last year :) <paron_remote>says the person with the least recognizable (and now corrupted) username in the channel <a_e>paron_remote: I also wondered if you were you :-) But right now, your user name is more pronounceable at least. <lfam>Is there a story behind your name paron? <a_e>I tend to read it as "paranoia" otherwise :-) <paron_remote>and so I used a barely-used MUD username I had from highschool <NiAsterisk>ng was taken here. otherwise my name would be much shorter <paron_remote>which of course had a badly written "dark and tragic" broody backstory <paron_remote>but that's what you get for being a teenager playing MUDs <rekado>what's wrong with ".io" domains? I thought they make *anything* cool. <paron_remote>anyway, in leiu of cwebber, paroneayea. and in lieu of paroneayea, I'm paron_remote :) <a_e>rekado: Read up on the politics in Wikipedia. <a_e>Jookia: How come your four patches arrive now with dates starting in the middle of February? Very confusing! <lfam>If you rebase or amend old commits, the dates don't change <amz3>so, i finaly read the purpose of rebase/amend synergy <NiAsterisk>rekado: all money goes to UK directly, it was promised to go back to the .io location, but nothing happened so far <a_e>I have four days of delay on guix-devel. Now I am close to reddition in front of 400 messages! <a_e>lfam: The date in the e-mail could change. <lfam>a_e: I see. That is weird! <efraim>the connman patches I sent lfam have a date of dec 31st <NiAsterisk>lfam: stiefel is boots (roughly, there are many words for shoes in any language :) ) and schuhe shoes <rekado>oh, I thought it's for aesthetic reasons. <NiAsterisk>i think you could just say shoes as i like to interchange some words which have the same functionality. there are books/wikis out there who deal with the origin and explanation of these sentence constructions <paron_remote>suspicious ownership or permission on `/gnu/store/xmpk0cj27i8bb54b4b3biniqgmwpyck9-fdm-1.9'; rejecting this build output <NiAsterisk>i use kamm bürste (comb, brush) interchangeable. doesn't make much sense, same with a selection of other things I don't care much about in language i use daily.. it's just not that important for my personal dictionary, where as somebody in the business which deals with hair would think it's very important. <lfam>paron_remote: IDK never seen that before. Maybe the build process trieds to chmod the output dir? <rekado>"I understand our practice is not in line with other packages but we have had enough bad <rekado>experience with all sort of installation and dependency problems and chose <rekado>to take the matter in our hands." <rekado>and not updating third-party libs, not applying security fixes. <lfam>Do you have any examples on unpatched vulnerabilities in the code they bundle? That may persuade them... <rekado>shouldn't be too hard. They bundle an old modified version of sqlite which I think has had security fixes since then <rekado>dunno, I kinda lost motivation to work on this. <lfam>That's fair. Some things aren't worth distributing ***paron_remote is now known as paroneayea
<lfam>a_e: I want to update python-pyicu and create a python-3 version. I see you added it in 2013 with a note about problems with a python-3 version. Do you remember any details? <lfam>I updated it and it builds fine with python-3 <a_e>lfam: No idea. If it works out of the box now, all the better, go for it! I needed just the python 2 version for another package (probably calibre). <lfam>It also has a test suite now! <a_e>rekado: Do you remember what the language on the website of unfonts is? I have installed the out, kr and cn outputs of font-adobe-source-han-sans, which is still not enough. <efraim>is there a second chinese option? <a_e>I could try them all, but that would somewhat defeat the purpose of multiple outputs. I could try Japanese, there seems to be a Japanese character in the mix. <a_e>But there is not enough hiragana for a typical Japanese text, I would say. <lfam>I am unable to reach git://git.sv.gnu.org/guix.git <NiAsterisk>it's on the dev-list, but I am waiting for response of the domain admin to fix mime <a_e>Yes, I am just reading the e-mail exchange with Ricardo, but the characters do not show up in my mail reader, nor in icecat. <a_e>So I wonder which font to install. Or is the web page only readable with unfonts? That would be a fun situation! <NiAsterisk>www.arphic.com.tw is the source of the fontset (arphic) which gets selected for the fonts loaded in the webbrowser when I open the unfonts page <NiAsterisk>and the domain admin is behind a privacywhois, but my email might get through and maybe they change the mime type. <a_e>Now I added all of font-adobe-source-han-sans, without success. <a_e>Well, that is not a real problem, since I understand none of these languages anyway :-) <a_e>But thanks for your help nevertheless! <a_e>And good luck with the mime-type. <NiAsterisk>no, i mean that's a fontset, arphic. let me search the package in gentoo <NiAsterisk>as we don't have any apparent "arphic" here, I might package that <NiAsterisk>try.. in some way. it all looks manual at this point when there's no Makefile for the font. <NiAsterisk>if the mime type is not changed, would it be bad for guix to change the downloader? <NiAsterisk>can I do something which will be just for this one package? <a_e>Is there any mirror which works? <NiAsterisk>for people not following the list that much: an oportunity for me to maybe add another hydra slave, or whatever guix needs, maybe even a third mirror, came up. i send an email to the devel list about this. <NiAsterisk>but, it's OVH/kimsufi, so if you share my doubts about them, that's a thing I would respect :)