IRC channel logs

2015-08-20.log

back to list of logs

***Mathnerd314_ is now known as Mathnerd314
<sbidin>Is there a way to generate a manifest file from the current generation?
<davexunit>not currently
<ecbrown>which linux kernel is in 8.3 alpha? my vm software is asking...
<davexunit>4.0.8 I believe
<ecbrown>ok, the highest is 3.x, so i will go with that. thanks
***ecbrown- is now known as ecbrown
<davexunit>ACTION packages sassc
<davexunit>that was a weird one, but it works. I was able to compile bootstrap CSS from source with it
<davexunit>which makes me feel a lot less bad about using bootstrap
<davexunit>now that I can actually compile their stuff without needing crazy node things.
<sbidin>just in time for the alpha, eh?
<davexunit>heh, yeah. :)
<davexunit>it motivated me to take another look at libsass
<davexunit>I like when people make things that avoid needing huge language runtime environments + special package managers to make them work
<sbidin>doing guix package -u starts upgrading my git, which it claims is 225 MiB
<sbidin>seems fishy somehow :)
<marcus>hmm, libinput seems to be too old to detect my touchpad correctly. is there a way for me to updated it on my own, or should i contact the package maintainer (if there is something like that in guix)
<phant0mas>marcus: send an email to guix-devel with what your problem is
<phant0mas>and good morning :-)
<marcus>yeah, good morning ;)
<iyzsong>marcus: our X don't use libinput by default, I think you have to add a section as 'extra-config' to xorg-configuration-file.
<iyzsong>btw, I just update libinput to 0.21.0 :-)
<marcus>iyzsong, thanks a lot. how can i do that extra-config? i am looking for that for a while now ;=)
<marcus>i mean in system configuration iyzsong
<marcus>is there a way to display the session log, like .xsession-errors? because that does not seem to be available
<remi`bd>you never know; is there a function to extract the hash from a path in the store?
<remi`bd>for instance: /gnu/store/zxfivagcmg1r76v137cfjiyzbqcv8bxh-attr-2.4.46/lib/libattr.so → zxfivagcmg1r76v137cfjiyzbqcv8bxh
<remi`bd>if not, I’m gonna simply remove the %store-dir prefix, call dirname, and use the entire dirname instead of the hash
<remi`bd>I would use that when there’s an error publishing a file from the store, to identify the store item that hasn’t been correctly uploaded
<remi`bd>(a simple string-match could also work, though)
<iyzsong>marcus: here is my config.scm: http://paste.lisp.org/display/153947 (I haven't test it yet..)
<marcus>great, thanks. just a small typo in lable. guess it should mean label?
<iyzsong>marcus: uh, I never find that 'lable typo, since it works :=
<marcus>hehe
<marcus>does not yet build http://paste.lisp.org/display/153947
<iyzsong>what error? I'm fetching linux-libre with it.
<marcus>ERROR: In procedure scm_lreadr: /gnu/store/h10nmh6r3r09cjqncdg3ij91dmk6jqnv-slim.cfg-builder:1:159: Unknown # object: #\\<
<marcus>maybe it's the comment in your libinput
<marcus>hmm nope
<iyzsong>I read something same on the mailling list, if you use git checkout, try 'make clean-go'
<marcus>i did a guix pull
<iyzsong>so it's really a bug ;-
<marcus>is there a way to work around it?
<iyzsong>marcus: can you send a report with the slim.cfg-builder file to the ML?
<iyzsong>it look like the same issue with "Set screen resolution in X" from Camel.
<marcus>yes, looks similar
<iyzsong>ok, wait, it's the wrong use of 'extra-config', I got the same error, sorry for that :-(
<civodul>marcus: could you paste /gnu/store/h10nmh6r3r09cjqncdg3ij91dmk6jqnv-slim.cfg-builder?
<marcus>iyzsong, could you paste a corrected config when you got it, please?
<iyzsong>civodul: I get the same. http://paste.lisp.org/display/153947#1
<civodul>iyzsong: this #<procedure ...> thing suggests you got a monadic value somewhere and forgot to bind it
<civodul>namely the 'startx' argument of 'slim-service'
<iyzsong>civodul: yes, I just figure it out..
<civodul>ok :-)
<iyzsong>marcus: this one should work. http://paste.lisp.org/display/153947#3
<davexunit>the news about the hurd port has been going around on reddit and stuff, so I posted it to HN, too https://news.ycombinator.com/item?id=10091371
<davexunit>welcome back civodul!
<iyzsong>rebooted, it works. welcome civodul and davexunit too.
<civodul>howdy! :-)
<civodul>davexunit: nice that you posted it to HN
<davexunit>need a few helpful upvotes to get it to the front page ;)
<davexunit>hurd stuff usually gets to the front page for the novelty factor, I think
<civodul>heheh
<marcus>iyzsong, just to let you know: works fine for me too, now. Thanks for your help
<davexunit>phoronix wrote about it, too. http://www.phoronix.com/scan.php?page=news_item&px=GNU-Hurd-Gets-Guix&utm_source=twitterfeed&utm_medium=twitter&utm_campaign=Feed%3A+Phoronix+%28Phoronix%29
<iyzsong>sure :-)
<civodul>Phoronix is our friend, it seems
<davexunit>yeah :)
<davexunit>thanks Michael, if you're reading this chat log ;)
<civodul>:-)
***gamabunta is now known as pikos
<davexunit>I think HN is penalizing me for some reason
<davexunit>the post has 12 upvotes, yet it has only increased my karma by 1
<davexunit>and it isn't anywhere near the front page despite having more upvotes than other new posts
<phant0mas>davexunit: wow :-D
<davexunit>phant0mas: fantastic job on the hurd port. congrats!
<phant0mas>thank you :-)
<remi`bd>phant0mas: your update mail is very impressive :D
<mark_weaver>thank you for all of your excellent work on the hurd port, phant0mas!
<mark_weaver>I'm looking forward to running it on one of my machines!
<mark_weaver>welcome back, civodul!
<civodul>hey, mark_weaver
<civodul>yeah, thumbs up for phant0mas!
<mark_weaver>I'm typing this on GuixSD running on a Yeeloong.
<davexunit>:)
<civodul>GuixSD on mips, woow!
<mark_weaver>not sure if anyone will use it, but at least it's work toward generalizing GuixSD to other architectures :)
<civodul>that also sounds like quite an achievement!
<civodul>nice
<mark_weaver>it's all on the wip-loongson2f branch
<xentrac>it's very important!
<mark_weaver>running a full XFCE here
<civodul>mark_weaver: so you're using GRUB?
<xentrac>that's fantastic!
<davexunit>exciting times :)
<mark_weaver>xentrac: thanks :) I do feel that it's important to support non-Intel architectures, especially since both Intel and AMD seem to be headed in the direction of making machines we can't use without non-free software.
<xentrac>yes, and also making non-auditable machines
<mark_weaver>civodul: yes, that was the last major challenge
<mark_weaver>full grub including the eye candy
<civodul>woow, exciting
<mark_weaver>also, qemu now builds on mips but the test suite fails.
<mark_weaver>so, for now, I'm disabling tests and building qemu with the aim to build a USB installer
<mark_weaver>I rewrote the grub image conversion code to avoid inkscape, instead using rsvg-convert
<mark_weaver>the challenge turned out to be that there's a layer in our SVG files with a checkerboard pattern that inkscape specially avoids rendering when exporting to png, but other tools render.
<mark_weaver>so I wrote a little emacs batch program to remove that layer from the files during the build.
<davexunit>you probably could have used sxml with guile :)
<xentrac>heh
<civodul>fun, you dug deep :-)
<mark_weaver>davexunit: you know, that was my first attempted approach, of course.
<xentrac>QEMU or equivalent is pretty important
<davexunit>mark_weaver: I figured ;)
<mark_weaver>davexunit: but alas, I found that our handling of XML namespaces is problematic, at least when using sxml->xml
<mark_weaver>maybe I'm not just using it right.
<davexunit>I should check it out sometime
<mark_weaver>basically, xml->sxml drops all of the xmlns:... attributes and instead creates symbols like: http://blahblah/blah:foo
<mark_weaver>and then sxml->xml raises errors on those symbols, complaining of more than one colon
<mark_weaver>xml->sxml accepts a 'namespaces' keyword argument to convert those to something like blah:foo instead, but then sxml->xml lacks any mechanism to do the right thing with those blah:foo symbols.
<mark_weaver>disappointing
<mark_weaver>so, I had to do it in a less nice way for now
<davexunit>oh bummer
<mark_weaver>but it would be good to get that sxml namespaces stuff sorted out
<davexunit>yeah
<davexunit>we should fix that
<mark_weaver>definitely
<mark_weaver>civodul: so, here's something I'm not sure how to deal with: %base-file-systems includes a filesystem that is not supported on MIPS, namely the cgroups hugetlb filesystem.
<mark_weaver>or at least I couldn't seem to figure out how to enable that in my kernel config.
<mark_weaver>in most of these cases, we can change things to procedures that accept a 'system' argument, but %base-file-systems is just a variable and part of our API effectively.
<mark_weaver>one idea I had is to add a flag to <file-system> that allows for graceful failure, or perhaps a 'supported-systems' field or something like that.
<mark_weaver>it would have to pretend to succeed, since 'user-processes' depend on all filesystems being mounted successfully.
<mark_weaver>well, we can do this on the ML. sorry, I have a bad habit of asking you too many questions on IRC.
<rekado_>mark_weaver: here's a patch to fix the sxml->xml behaviour: http://lists.gnu.org/archive/html/bug-guile/2015-04/msg00009.html
<mark_weaver>rekado_: ah, perfect! sorry for not responding to that. I remember looking and not fully groking the issues at the time, so I postponed ... too long.
<mark_weaver>but now I understand the issues (at least partially), and the need.
<mark_weaver>bah, I'm stretched too thing
<mark_weaver>*thin
<mark_weaver>too much to do, not enough time
<davexunit>time to recruit and delegate ;)
<phant0mas>remi`bd mark_weaver civodul thank you guys :-)
<mark_weaver>davexunit: quite true! I've never been good at that.
<mark_weaver>davexunit: would you like to take a look at that bug report and DTRT? I suspect you have more knowledge in that area than I do.
<davexunit>yeah I can take a look. I don't know a ton about XML namespaces, but I can test out the issues with SVG and see if the proposed patch fixes things.
<davexunit>feel free to comment about style guidelines and stuff
<davexunit>you know more about that stuff
<mark_weaver>davexunit: no need, I have confidence in you :)
<davexunit>okey dokey! :)
<mark_weaver>but if you feel uncertain about something, feel free to ask me for advice.
<davexunit>sure thing
<mark_weaver>thank you!
<davexunit>I want to make future enhancements to the sxml modules to better handle html
<mark_weaver>sounds great :)
<davexunit>for now I have my own sxml->html procedure that I ship with haunt
<mark_weaver>I really need to take a look at haunt. sounds like something I could use.
<davexunit>mark_weaver: I'm sure you could find many things to improve in it, too.
<mark_weaver>and we should add it to guix. you're still waiting for a response on that, sorry.
<davexunit>mark_weaver: I pushed it.
<mark_weaver>ah, great!
<davexunit>ricardo looked at it.
<davexunit>it's pretty primitive right now, but with some more love and thought I think it can be quite nice.
<mark_weaver>speaking of things starting out primitively, I took at look at the initial commits of the guix repository, just for fun. a humble beginning to a vibrant project :)
<davexunit>I did that a few weeks ago
<davexunit>was cool to see
<civodul>mark_weaver: haha, that's the least we can say :-)
<mark_weaver>:)
<paroneayea>mark_weaver: davexunit: https://identi.ca/cwebber/note/y0hdfNZaS0uIPJFtCEfIJg
<paroneayea>potentially a bit flamey
<davexunit>I'm always down for some flams
<davexunit>flames*
<davexunit>I mean, flams too, because I'm a drummer. ;)
<paroneayea>:)
<davexunit>I'll read after I get myself some lunch
<paroneayea>debating an x-post to my blog or not
<davexunit>paroneayea: good read.
<davexunit>npm really is the worst.
<davexunit>the nix folks have npm2nix which will spit out the entire dependency tree for a node package
<davexunit>which I think is about as good as we can hope for.
<paroneayea>davexunit: admittedly npm is much more feasible in nix/guix than elsewhere
<davexunit>paroneayea: speaking of npm, do you know of a JS minifier that doesn't run on node?
<davexunit>yesterday I found and packaged a SASS compiler that has no ruby or node dependency.
<davexunit>now I need a minifier.
<davexunit>this is to make guix-web a sane thing to include in core guix.
<davexunit>I really need to get comfortable enough with the dmd to hack on it.
<paroneayea>davexunit: I don't know of one!
<davexunit>it has greatly improved things for me at work where I have to manage a large number of daemons that run under my user account.
<davexunit>it just needs a REPL.
<paroneayea>davexunit: I posted a followup to my own post https://identi.ca/cwebber/comment/BrV18gbcRAqx9CQeYsi5wQ
<civodul>bavier: FYI, the 'slepc' package keeps references to all the build-time tools, which is guess is a mistake
<taylanub>civodul: BTW if you find time, would love input on SRFI-123: http://srfi.schemers.org/srfi-123/ it pains me that I can't support alists (since alist-ref is so common in Guix), but I really see no clean way.
<civodul>ok
<civodul>i'm not quite convinced by the dynamic dispatch approach, but we'll see
<davexunit>ACTION feels similar
<taylanub>because of performance, or?..
<civodul>when the reader or the compiler sees 'vector-ref', they know that what follows is a vector
<mark_weaver>ACTION agrees
<mark_weaver>there are times when dynamic dispatch is quite useful, but this is not one of those cases.
<mark_weaver>paroneayea: it's really great that you're working on this and writing about it :)
<mark_weaver>davexunit: perhaps I'm naive, but it seems to me that a minifier wouldn't be too hard to write, especially since we already have a partly-complete javascript front-end for guile.
<mark_weaver>okay, qemu-headless is built (without tests) on my yeeloong. now to try to build a usb installer
<davexunit>mark_weaver: that's a good point. I haven't explored Guile's JS capabilities.
<davexunit>I was looking for something was already written.
<davexunit>paroneayea: someone showed me the original jsmin http://javascript.crockford.com/jsmin.html
<davexunit>whose source is a single C file
<davexunit>so yeah, maybe I should just write the Guile equivalent.
<mark_weaver>davexunit: or use that C program!
<mark_weaver>whatever's easiest
<daviid>^^ Copyright 2001 Douglas Crockford. All Rights Reserved Wrrrldwide.
<davexunit>hmmm
<davexunit>yup, seems to be proprietary
<davexunit>wait, no it's free.
<davexunit>the source file has the expat license
<davexunit>plus this: "The Software shall be used for Good, not Evil."
<daviid>ok, i was wondering
<daviid>when I see "All right reserved worldwide", my 1st reaction is to step back
<mark_weaver>hmm
<mark_weaver>I'm not sure it's a free software license. looks like a restriction on use.
<xentrac>there's been a long discussion of this on debian-legal
<davexunit>I will just not touch it.
<xentrac>daviid: under some non-Berne copyright conventions, the "All rights reserved" notice is required to avoid abandoning the copyright
<davexunit>why do people add weird stuff like this to licenses?
<xentrac>as a form of self-expression.
<davexunit>s/this to/this in/
<daviid>xentrac: why do we do not use/add this for the fsf ? and the fsf has lawers who surely tok care of this right?
<davexunit>the FSF doesn't have any lawyers on staff
<davexunit>well, actually they have 2 JDs
<mark_weaver> http://www.gnu.org/philosophy/programs-must-not-limit-freedom.html
<davexunit>mark_weaver: I had interpreted it as more of a "cute" thing that had no legal standing because it's far too vague to be taken seriously.
<daviid>anyway, unless clearly stating what license, which afaic must be gpl compatible, it's not good. the copyright is just who's going to defend the license terms in case of violation
<mark_weaver>xentrac: if that were true, I'm sure Eben Moglen and the Software Freedom Law Center, who worked on making sure GPLv3 worked well internationally, would have pointed out that we need to add "All rights reserved" to the standard GPLv3 copying permission notice.
<mark_weaver>and it's not just the GNU GPLv3. there are several other well known free software licenses that don't include that language in their copying permission notices.
<xentrac>mark_weaver: I think that at this point there are no countries that are signatories to the Inter-American Convention or whatever it is that aren't also signatories to Berne.
<mark_weaver>the FSF may not have lawyers on staff, but they have some brilliant lawyers providing massive amounts of pro-bono services.
<davexunit>yes, this is true.
<xentrac>it was the Buenos Aires Convention
<xentrac>WP explains, "The requirement to add the "all rights reserved" notice became essentially obsolete on August 23, 2000 when Nicaragua became the final member of the Buenos Aires Convention to also become a signatory to the Berne Convention."
<mark_weaver>okay
<xentrac>I certainly don't mean to cast any aspersions on Eben.
<xentrac>I hold him in the highest esteem.
<mark_weaver>:)
<paroneayea>iirc "all rights reserved" is basically a licensing no-op
<paroneayea>but it's certainly not friendly sounding
<sbidin`>davexunit: where's guix-web? gitorious is down...
<davexunit>sbidin`: https://git.dthompson.us/guix-web.git
<sbidin`>thanks!
***sbidin` is now known as sbidin
<mark_weaver>hmm, I wonder if this kernel for loongson2f machines will work in qemu. I guess we'll find out!
<alezost>hm, did anyone receive my yesterday's patches for updating "magit-svn" and "git-modes"? I see only "magit-svn" one on guix-list: <http://lists.gnu.org/archive/html/guix-devel/2015-08/msg00383.html> and none on gmane.
<alezost>so I'm not sure if I should resend "git-modes" patch or both or none
<mark_weaver>alezost: I patch I posted to grub-devel over a day ago is also shown on lists.gnu.org but not gmane
<mark_weaver>s/I patch/a patch/.
<mark_weaver>I guess gmane is having issues
<alezost>yeah, I think so too; I'm just a bit concerned that my second patch did not appear on guix-devel list
<mark_weaver>I suppose the other possibility is that something is wrong with the gnu lists, and some of the subscribers (including gmane) didn't get the message.
<yenda>do we have an utility for screenshots yet ?
<alezost>yenda: there is 'scrot' package, if it's appropriate for you
<mark_weaver>alezost: when you did send the messages?
<mark_weaver>I want to see if I received copies via email.
<alezost>mark_weaver: <http://lists.gnu.org/archive/html/guix-devel/2015-08/msg00383.html> shows the right time. The second patch was sent right after that
<mark_weaver>alezost: thanks
<alezost>thank you for responding!
<mark_weaver>alezost: neither of the messages were delivered to me.
<alezost>ouch, thanks for looking, I think I'll wait for a couple of days, and then resent them. I recall there were similar problem with gnu lists in the past
<mark_weaver>alezost: I've notified nully via PM
<mark_weaver>(although she's also here on channel)
<alezost>thanks for noting
<paroneayea>This is the year of the Hurd desktop!
<paroneayea>^-- reddit comment :)
<paroneayea>"year of the Linux(sic) desktop" is a classic
<paroneayea>in all seriousness, awesome work to phant0mas
<bavier>civodul: regarding slepc, the references are kept in installed makefile fragments that are meant to be used by developers building against the library
<bavier>I'll take a look though, and see if there are any references that aren't strictly necessary
<civodul>bavier: ah ok, didn't know about the makefile fragments
<civodul>then it's probably fine
<sbidin>Is there something like a (write-string-to-file s f) function?
<sbidin>I could use (copy-file), but then I need to create a temp one.
<davexunit>(call-with-output-file "foo" (lambda (port) (display str port)))
<sbidin>Great, thanks!
<sbidin>ok, I've made lua build with its /lib/pkgconfig file now
<sbidin>but am not sure whether I've done it the right way
<sbidin>this is what I've added: http://lpaste.net/139298
<sbidin>it builds fine, but am wondering whether I made common errors
<sbidin>ACTION isn't very familiar with scheme
<sbidin>note: lua doesn't come with its own *.pc file
<sbidin>not sure whether just embedding it like that is acceptable
<cehteh>iirc there was some dicussion about a .pc file on the lua mailinglist some time ago
<cehteh>someone maintains one or so
<cehteh>oh lol .. that was from the guix maintainer :D
<sbidin>the thread taylan started? looks like there's a patch there for the makefile, which would take care of generating the makefile
<sbidin>thanks!
<cehteh>yep
<sbidin>*generating the pc file
<cehteh>lua has its own goals, but still a very supportive community
<cehteh>i like it a lot
<codemac>in case anyone was following along with my bitlbee drama yesterday, it was the stupidest of mistakes. I was doing 'bitlbee -n' but not 'bitlbee -D -n' in the systemd unit file.
<mark_weaver>ah, I'm glad you sorted that out :)
<mark_weaver>I was the one who added the bitlbee package and service, and I've used it quite a bit, but lately switched to pidgin