<civodul>(i've already paid the Tobin-patch tax, so i can afford bikeshedding) <mark_weaver>we should apply that suggestion more widely in bikeshed debates :) <jxself>Some people owe patches already I see. <jxself>I think Stallman can be excepted though. :) <civodul>looks like everything is in place for the release <civodul>make sure to upvote and whatnot tomorrow! ;-) <jgrant>Why would "checking if the c compiler works" fail in a ./configure? <jgrant>Yeah, is that supposed to be at the top-level of the directory? Couldn't find it. *jgrant is starting to get fuzzy-brained, he'll look into it tomorrow or-so. <jgrant>Probably the dumb way of doing this anyways, compliling all locally from source instead of using Nix to import the package for stumpwm. <mark_weaver>this is how GNU GSD will grow. people are motivated to add the packages they use. *jgrant still needs to either figure out Clisp or SBCL. And if Clisp, what the issue with Libffcall is. <jgrant>mark_weaver: Seeing that my main box is now 95% GNU GSD (Bootstrap partition still intact just in case) -- I see myself contributing a lot more. <jgrant>At least until I get the rest of my day-to-day software stack in the repo. Namely stumpwm and related tech. <jgrant>The latter is a much further goal, due to where Lispkit is in it's development. *davexunit wonders about how to package Drupal and its many modules *jgrant has noticed that CLisp (he thinks) comes with a CLX module. That just means he'd need to figure a system for cl-ppcre ... but he doesn't think that uses any existing build systems. <davexunit>man, guix takes a long time to build these days. <jgrant>Drupal, Mediawiki, Mediagoblin, Gitlab, and similar are all very important I think ... but too, all very large and have many depends where such things are daughting. <jgrant>davexunit: About 2 minutes, on my realtively high-end laptop. <jgrant>Does it compile the most of the source tree, or /just/ changed modules. <mark_weaver>because we already have gcl, and we use it to compile maxima, which uses asdf I believe. <jgrant>mark_weaver: No, it's evidently not compliant in a number of ways. <jgrant>The only known GNU implementation of CL that works is Clisp. <jgrant>And even then, people say it's fallen out of favor. <mark_weaver>jgrant: I see a lot of mentions of 'gcl' in the stumpwm source code. e.g. code that's conditionally included for gcl. <jgrant>Hm, maybe the few that responded in #stumpwm were not knowledgable of the codebase ... I know how that works. *jgrant goes to ask again, a few days later. *jgrant makes a note. He's about useless right now. <mark_weaver>of course, it would be good to add clisp and sbcl, but since you seem to be impatient enough to want to import the package from nix somehow (which I think is not really feasible, short of installing all of Nix including its base libraries and programs), trying stumpwm on gcl might be better than that :) <jgrant>This was a little over a week ago when I asked, I can't remember if it was in #stumpwm or #lisp... <jgrant>That was the issue, clx is doubtful to work on it anymore. <mark_weaver>it might simply be that none of the stumpwm devs use gcl <jgrant>mark_weaver: Bike in chat, says the only reference he see's in the source are "run-shell-program" which he says isn't relevant. <mark_weaver>and maybe the users don't either, so it may be that it hasn't been tested in a while. but the source code suggests that it has worked in the past. <mark_weaver>well, fwiw, I would normally run GNOME 3, but have put it aside for a long while so that I can eat our dogfood and work on Guix more effectively. However, if you can't wait for the software that's not yet in Guix, then another option is to install Guix on top of another distro. <jgrant>mark_weaver: Nah, I explicitly installed GSD as my main distro to force myself to work on Guix more. <jgrant>I just need to sit down and do it. <jgrant>It's all good, as of Feburary I'm getting in a realtively strict schedule which allots me about 2 hours minimum of hacking daily on whatever -- so likely, much of this time will go to Guix. <iyzsong>I have /run, /tmp mounted as tmpfs with (need-for-boot? #t) in GSD with no problem, perhaps make it default? <mark_weaver>iyzsong: package builds are done in /tmp, so it has to be quite big in some cases, depending on the package. <mark_weaver>also, I think it's suboptimal nowadays. the caching in linux (the kernel) is good enough that I don't think there's a big advantage to tmpfs for those anymore. <jgrant>Not a huge deal for me since I'm no longer using Frog for my blog, but should "raco" failt to install packages? <mark_weaver>and anything in tmpfs needs to be stored in RAM+swap, which is quite limiting. <jgrant>I'm assuming because there is no build-system for racket. <mark_weaver>iyzsong: however, I would be very much in favor of clearing out /tmp and /run on boot. <mark_weaver>jgrant: good question. I don't know much about how raco works. <mark_weaver>however, I think it would be preferable to add a racket-build-system and package importer for racket packages, similar to what we have for pypi packages <jgrant>Didn't we have a partial system for Ruby? <toxemicsquire4>Hi, I'm attempting to package wicd for guix, I don't know how to upload or anything yet, but I just want to be able to package it first. I've read some of the manual online, but I can't find any useful package definitions, is there a folder on the system which stores those. Or how can I get them? <rekado>civodul: about the logo: I centered the spiral thingie and the text vertically, because it looked weird in my opinion to have the text higher than the spiral. <rekado>looked as if someone just removed the baseline text and didn't bother with the alignment. *rekado goes afk for a while <civodul>you really mean "higher", or you mean with the top of the text aligned with the top of the spiral? <civodul>maybe you're right, i haven't checked what it would look like actually <civodul>just thought it might be useful to keep the placeholder <rekado>I aligned the vertical centres. Previously the vertical centre of the right part was lowered by the baseline text. <rekado>Anyway, I don't mind moving the text up a bit, relative to the spiral thingie. <civodul>yeah, we could try and see what it's like <civodul>or find another baseline or something <jgrant>I mocked up a G that looks like a ying-yang to mimic the eval-apply "symbol" in SICP if anyone is interested. :^P <civodul>i don't think it would be a good idea, for reasons taylanub very well explained <jgrant>What was the above refering to then? <civodul>the one without "the GNU system" on it <jgrant>A GNU System, not make the cut? :^P <civodul>"talking points" is the term i think :-) <jgrant>civodul: I do really want to thank you for your amazing paitence you've had regarding this issue. Hopefully RMS will budge and be "okay" with GSD as a name, and we can finally move past this big mess... <jgrant>I'm hoping we are just about done. <civodul>i hope the FSF/RMS kind of people will understand this <jgrant>civodul: When you do mention the name during your talk, please, please, please make sure you make it clear it's "GNU GSD" not just "GSD" -- this is one of RMS's biggest areas of contention, that some people will think the G in GSD is GNU. <jgrant>I think it's clear and I think you know that already... but, still, very important me thinks. <Sleep_Walker>it's nice to see so many people here, <name to be decided> is growing nicely! <Sleep_Walker>I didn't have time for that recently but at least ML has good number of patches/month <civodul>Sleep_Walker: the name *is* decided :-) <Sleep_Walker>civodul: it's not important for me, I'll just check in 3 months what is current :b <jgrant>Sleep_Walker: I think it's pretty universally accepted as "good enough" for people working on Guix, RMS is really the only opponent on this. Not sure if he has the authority to do anything though. <jgrant>His arguments are pretty weak, I think, in any case on this matter. Though to be clear, overall, I strongly respect the ma -- just not his view on this point of contention. <Sleep_Walker>I'm no longer interested in this topic, sorry, I had blood boiling already some time ago :b <jgrant>Sleep_Walker: I mean, it's hopefully "over". <adhoc>i wish folks would just get hacking on important stuff instead of boiling blood over a name <jgrant>We'll see if/how RMS reacts to use somefactor of pushing a name he isn't 100% okay with ... but yeah. <adhoc>i wonder if GSD is too close to BSD for him ? <jgrant>adhoc: That's one of his arguments, but I see it no worse than "GNU is not Unix" as a playful jab. <adhoc>mark_weaver: yeah. there are better things to do =) <jgrant>I mean, there's not much really to talk about more. It has been fomrally chosen, documentation and other realted bits have been and will continue to be changed to created to reflect that. The only thing I thing that can be done to change that if RMS was able to take action and I'm not sure if he has the power to do so. :^P <jgrant>I'm expecting and hoping the issue is done. ***civodul changes topic to 'GNU Guix | http://gnu.org/s/guix/ | 0.8.1 is out! | FOSDEM: https://fosdem.org/2015/schedule/event/the_emacs_of_distros/ | This channel is logged, see <https://gnunet.org/bot/log/guix>.'
<civodul>i second mark_weaver, let's stop talking about it <jgrant>Is there any way to view a GNU's page traffic? I think it would be interesting to see the spike before and after FOSDEM. <civodul>jgrant: i think the FSF people don't publish this info <civodul>i'm not sure there'd be such a spike, though <jgrant>I mean, I would hope. I don't know how much traffic the page get ussually -- but I suspect a fair amount of people who are going to a talk have barely heard of it and/or have not heard of it at all and were drawn in by the talk title. <jgrant>Maybe not a huge spike, but I would think a notable one. <Sleep_Walker>oh, 0.8.1, I have to update packages to be ready for the peak :) *jgrant needs to figure this ACPI issue eventually ... getting very annoying to use a tty. <jgrant>Anyways, it's edging to 4am and have class at around noon. If I wasn't formal enough with that smile before, congrats on 0.8.1! *taylanub needs to change his last name some time :P <taylanub>I pop up twice in the commit list, once with just "B" for my last name (a legacy measure) and once the full "Bayırlı/Kammer" <civodul>↑ anyone willing to send this to HN? :-) <taylanub>civodul: is it fine to mention 'GNU GSD'? <taylanub>(this is technically a Guix release I guess; I'd mention it in the text anyhow) <taylanub>hm, the spelled-out name is on the page; I'll do the same then <civodul>yeah, either the full name or "GNU GSD" <taylanub>doh, I must submit either the URL or text <atheia>Morning Guixers, what a bright and shiny day now 0.8.1 is out :-) <atheia>done — had to make an account first, which I guess was about time :-) <atheia>hmm, so what's the best way of installing 0.8.1 to replace 0.8? <atheia>From what I can see, guix pull, pulls in new scm files, but not for example emacs.el etc. <atheia>so presumably a new tarball install? <rekado_>how long does it usually take until a completed build job on Hydra is marked as successful? <rekado_>the builds for icedtea6 on i686 and x86_64 are complete according to the logs, but the builds are still marked as "in progress". *davexunit reads new email from rms and screams *rekado_ had the same reaction to the two rms emails. <rekado_>old and busted: "Linux" vs "GNU/Linux"; new hotness: "GNU GSD" vs "GuixSD". <davexunit>apparently the preceding "GNU" doesn't satisfy rms <davexunit>people will think that it's "GNU GNU System Distribution"? <davexunit>taylanub: please don't call the guix distro "avant-garde". rms likes to use it but it just dooms our project. thanks. :) <atheia>yeah, pretty frustrating… I guess he really doesn't like the punning on BSD. Which is a shame: I definitely think that that's fun. <davexunit>apparently what's "fun" is to be decided by rms <rekado_>is it about "fun"? Or about his second reason: "G" is too similar to "B"? <davexunit>well he thinks it could be interpreted as a "dig" at BSD, so we better not say it. <rekado_>how could this abbreviation possibly be interpreted this way? <davexunit>are are going for "playful cleverness" as he says, but he doesn't see it. <rekado_>It's not like we're calling it BSD but with a different expansion. <rekado_>(by the way: I paid my taxes already.) *davexunit took the day off, will be working on a guix patch or two ;) <taylanub>davexunit: you think it sounds too .. old-timey? <davexunit>taylanub: it's not that, but things that are "avant-garde" are things that people didn't discover until well after their time has passed. <davexunit>like an artist being appreciated only after they die. <davexunit>I hope people will appreciate Guix while it's actively maintained! <taylanub>OK, didn't know of that connotation. the phrase was recommended by RMS... <davexunit>yes, he did recommend it. I don't blame you for using it. <davexunit>he's advised that people don't use words that are totally fine, while promoting words that are truly detrimental. <davexunit>release day is a happy day though, so let's stay happy! <davexunit>on the bright side, I am very pleasantly surprised at the HN comments for the new Guix release <civodul>yeah nice comments on HN, that's cool <davexunit>many times I was like "oh yeah, that happened! cool!" *davexunit updates inkscape <davexunit>release isn't official yet, but building from what should become the release tarball. <civodul>it's at the top of HN, so it must be almost official <davexunit>the folks on #inkscape told me that despite it being spread everywhere, that the release is not official yet. <davexunit>it looks like they won't be using sourceforge for this release, either. so gotta figure out what the new canonical download place is. <civodul>i upgraded Racket yesterday and it was terrible too <civodul>and of course, it would fail close to the end <civodul>because it dlopens stuff and doesn't check beforehand whether these are available <civodul>Guile is so much better in this respect ;-) <davexunit>inkscape builds and run just fine with only tweaking the version and hash. :) <davexunit>I'll keep this in my back pocket until the release is official and I have a proper download URL. <davexunit>I hate PHP, but so many free software web applications run on it. <davexunit>if I could figure out a sort of drupal-build-system thing, that would be cool. <rekado_>(for comparison: icedtea6 builds in 57mins (without tests) on my workstation) <davexunit>what do you do when a configure script tries to do things like this? <davexunit>configure:23184: gcc -o conftest -g -O2 -fvisibility=hidden conftest.c <davexunit>this somehow fails when building, but works in my 'pure' guix environment. <civodul>davexunit: could you check config.log with "guix build -K"? <davexunit>configure:23184: gcc -o conftest -g -O2 -fvisibility=hidden -Wl,-rpath,/gnu/store/dng0rvcw4264k0r9fp7618bxsh20bi9v-libxml2-2.9.0/lib -L/gnu/store/dng0rvcw4264k0r9fp7618bxsh20bi9v-libxml2-2.9.0/lib conftest.c <davexunit> -lrt -lm -ldl -lnsl -lxml2 -lz -lm -ldl >&5 <civodul>rekado_: the result is being transferred from the build machine to hydra.gnu.org <davexunit>couldn't find zlib. I guess my pure env isn't that pure. :/ <civodul>davexunit: yeah, perhaps because there's an ld.so.cache or something <davexunit>a lot of optional dependencies to add in now, though. <civodul>rekado_: BTW, the CUPS client works pretty well <civodul>at home, the CUPS-enabled GTK+ magically discovered the printer plugged in my partner's computer <civodul>at work the GTK+ print box doesn't work so well, not sure why, but lpr does *davexunit watches all the PHP tests fail <rekado_>meh, why do a lot of authors of scientific applications have no discipline when it comes to building their software? <rekado_>here's one that depends on bzip2, jansson, and zlib, but it includes slightly patched versions of these libraries as tarballs and then builds them in-tree. *rekado_ notes that these continuing unproductive discussions about agreed-upon names mostly attract those who do not contribute to Guix. <dfh>great to hear you're working on php davexunit! i i totally get you re hating it. wish i could help, but i don't have the skills to (yet) <davexunit>dfh: I think I'm almost there. the issue with php is that there are a *lot* of optional dependencies to turn on with flags and pointers to the lib directories <dfh>doing a wordpress plugin atm & feel like i'm stuck in a Kafka novel. <davexunit>I spend most of my time working with drupal/civicrm at the FSF <davexunit>I'm beginning to learn how to use 'drush make' to wrangle the complexity of our setup <davexunit>but in my ideal world, drupal + all the good modules will be readily available in guix <davexunit>and you'd just install them to a profile and voila. :) <dfh>really, drupal would exist in your ideal world? ;) <_`_>In my ideal world cms be written in scheme. <davexunit>okay, halfway between here and my ideal world <davexunit>that's where I fix the deploying web apps problem <davexunit>and then I work on making good web apps written in real programming languages <davexunit>we can do better than PHP. free software devs and users deserve it. :) <dfh>little by little, bit by bit :) *davexunit found a problem with httpd recipe <davexunit>I see that it's in the store directory for every profile <_`_>doesn't glibc provide that <davexunit>this doesn't make sense. the include should be just fine... <davexunit>looks like php is expecting libintl.h to be provided by gettext <ams>civodul: is there a reason why you named the usb images for gsd? <davexunit>I can't find a definitive answer here. plenty of people on the web said the gettext provides libintl, but that doesn't seem to be true <DusXMT>davexunit: afaik, it depends on whether you have a libc that supports gettext. Glibc, if I remember correctly, does support/contain gettext, so it should be provided by it (while on some systems, like w32, gettext has to be explicitly provided) <ams>what was my last msg? <davexunit>DusXMT: so I guess PHP's configure script is doing the wrong thing. <_`_>davexunit: are you just passing --with-gettext? <davexunit>_`_: yeah, 'with-gettext=/gnu/store...', but I've removed that switch for now. <davexunit>I'm going to give compiling this another few tries and if I can't get it I'll move on to something else. <davexunit>maybe I can finally figure out my 'guix publish' problems today... <davexunit>ld: ext/opcache/.libs/ZendAccelerator.o: relocation R_X86_64_32S against `.text' can not be used when making a shared object; recompile with -fPIC <davexunit>ext/opcache/.libs/ZendAccelerator.o: error adding symbols: Bad value <davexunit>this is a library within php that it's building, so I don't know if it will have any effect <bavier`>just looks like libtool is trying to create a shared library when the objects weren't built with -fPIC <bavier`>davexunit: can use see the compile statement in the build logs? <davexunit>bavier`: yeah but I don't really understand it <bavier`>usually libtool takes care of the -fPIC then *davexunit said he would punt on php but keeps trying anyway <davexunit>bavier`: the build completes when I turn off some configure flags, so I'm bisecting until I find the culprit <DusXMT>Does anyone know how does cross-gcc (from gnu/packages/cross-base.scm) know what dynamic linker to use? <DusXMT>mark_weaver: already found it, thanks <davexunit>bavier`: the switch that builds the apache module is what's breaking everything, somehow. <davexunit>the other annoying issue is that the whole test suite fails, but returns a status code of 0 <bavier`>mark_weaver: the python-pillow failure was caused by an upgrade of openjpeg. <bavier`>If I upgrade openjpeg again, to the 2.1.0 version, python-pillow builds fine. I'm building openjpeg's dependents to make sure they're ok with the upgrade. <davexunit>hmm, there must be something about the default make flags or something that breaks the php build <davexunit>I built it in a guix environment without issues <davexunit>ld: ext/opcache/.libs/ZendAccelerator.o: relocation R_X86_64_32S against `.text' can not be used when making a shared object; recompile with -fPIC <davexunit>ext/opcache/.libs/ZendAccelerator.o: error adding symbols: Bad value <davexunit>Makefile:738: recipe for target 'ext/opcache/opcache.la' failed <mark_weaver>davexunit: I would save a log of the build output when running in guix environment, and compare that with the build log whe building in guix-daemon. <davexunit>when building php with the --with-apxs switch to build the apache module, compiling this Zend stuff fails. <mark_weaver>I've managed to debug some inscrutible problems on MIPS/ARM by finding the first occurrence of a notable difference in config output compared with the output on intel, for example. <mark_weaver>missing propagated outputs will also cause problems in a chroot environment but work properly in guix environment. <davexunit>this seems to be some static vs. shared object problem <mark_weaver>note that to comparing logs, it is sometimes necessary to remove all nix hashes from both logs before comparing them with diff. <mark_weaver>though in this case, it might not be needed if you local profile is up to date. <mark_weaver>time to fix the backlight in my X200 and get it flashed. <jxself>Because the announcement's not been posted yet. *jgrant 's next box will probably be a Gluglug or if he has the money a Librem/Novena. <jgrant>I did not expect, by any means, to have 3 or-so viable mostly FOSS compliant to the bios level laptops floating about in 2015. <jgrant>I have no real reason to upload now, now that I've gotten GSD on my realtively high-end laptop -- but my next box will certainly be of this stock. :^) <davexunit>I wonder if the novena will ever get RYF certification <jgrant>davexunit: Did they resolve whatever problems they still had (thought something with boot?). <davexunit>I don't know what remaining issues there are. <mark_weaver>davexunit: I vaguely remember reading something that implied that the Novena would not get RYF certified, but I've forgotten the details. <jgrant>Don't know enough about Librem, but I've not also heard loud outcries ... so I assume that it's at least "sufficently foss" for most people. Don't know if that is actually free-software level, ryf-compilant. <jxself>I suspect it has to do with the distro they ship, at least partly. <jxself>And graphics unless the 3D acceleration can work now without the non-free parts. <mark_weaver>it might be related to the GPU, whose libre driver is still a work in progress, or perhaps with the FPGA, for which there is no free compiler to generate new FPGA bitstreams. <jgrant>At the very least, such attempts make me hopeful that there is a decent market for companies that want to provide all-free software capable systems -- to sustain themselves. <jxself>Ah yeah, the FPGA would be a problem with their criteria too I suspect. <davexunit>I will only use free software on my Novena, so no big deal. <jgrant>There's like currently 0 free-software and/or complete FOSS toolchains for FPGAs right? <jgrant>I was talking/asking about this one day in ##fpga, and people thought it was so odd I wanted such a thing. <davexunit>that people using FPGAs don't care about a free toolchain <jgrant>Too, they thought it was odd to ship such a thing in a consumer aimed decive at the level of a laptop. <mark_weaver>on the other hand, FPGAs seem to me to be the only practical way for us to start experimenting with free hardware designs like CPUs, which I think will be crucial. <davexunit>so there's very few people capable and motivated enough to do the work. <jgrant>davexunit: Well, I mean aren't there like 3 big companies that controll a VAST amount of the market? <davexunit>I don't know. I'm pretty ignorant when it comes to FPGAs. <davexunit>but mark_weaver makes a great point about how crucial they are. <jgrant>Froget the other really big one. <jgrant>Had to check their site; they look familar, but not sure if that's in the top 3. <jgrant>I was listed them in ##fpga, so I assume it was relatively trustworthy. THe only one I can say with certainity I remeber is said mentioned Altera. <jgrant>Does anyone know if the lowRisc board is going to make an attempt at a FOSS toolchain? <jgrant>I know the design implemented on the FPGA is FOSH, don't know about the toolchain which supports is... <jgrant>That would be a big bummer, but something I kinda expect at this point... <jgrant>Another argument was that they didn't see any worth for them to be in consumer level devices, which I found odd. <davexunit>I think we'll need to make more of our own hardware going forward. <davexunit>since everything is moving towards lock down. <davexunit>people use mobile phones more than any other type of computer, and they are almost all hopelessly locked down. <jgrant>davexunit: Well hopefully home-fabriaction will be somewhat affordable in like 10-20 years. Ideally if we could amass a fund at hackerspaces to by certain machines... that would do a lot I think. <davexunit>despite all the things I disagree with him on, I agree with this. :) <davexunit>and in the meantime freeing Thinkpads and such is great. <jgrant>Is Chisel at all popular? I saw that lowRISC is using it in their toolchain. <jgrant>Maybe it's just that I really didn't pay attention to such things when I first was getting into Free-Software near 5 years ago now ... but from my (certainly biased) point of view, it seems like there is a lot more full FOSS-capable devices and FOSH designs out there generally. Not just the ones that get popular. <jgrant>Which if try and is not just me being overwhellmed by be outwardly aware of such things, is GREAT. <jgrant>Which if I'm actually trying to be objective and coming up with such things* <davexunit>I give up on php... I'll try to fix the bugs in 'guix environment' instead. <jgrant>davexunit: PHP, for Drupal? Or making like a build-system? <davexunit>jgrant: PHP for anything PHP-related. no one's packaged it yet. <davexunit>I have an almost working package, but the apache httpd module is broken, which is the most important thing. <davexunit>I want people to be able to get their GLAMP on in GNU GSD *jgrant just got reminded, needs to figure out Linode soonish. <jgrant>davexunit: Yeah, that would be nice. :^) <davexunit>AMT was completely disabled, which is super impressive. <civodul>davexunit: re PHP, maybe you should post the current status on the list *jgrant could probably install GSD on Linode the same way he did on his native box, but that seems like a lot fo work for something that should be able to be factorized. <jgrant>davexunit: Also, yeah LAMP or any variant there of would be useful to many. :^) *jgrant looks at the FPGA jxself give ... who was this mysterious "big 3rd" they were talking about in ##fpga when he asked awhile back, he wonders. <jxself>I'm not sure. That chart is from a couple years ago though. *civodul goes get some sleep *davexunit attempts to fix guix environment <davexunit>I just made some silly mistakes when I wrote it. <ams>please refrain from using GSD as the name for GuixSD <rekado>sent a couple of more audio-related patches to the mailing list. I would be happy for a review (especially of that waf-build-system that all of these new patches depend on).