<jmi2k_>Reading the mailing list I noticed neovim was added along with its dependencies... <jmi2k_>Well, at least it's way better packaged than my attempts. <ng0>there's also neomutt or whaterver it was called, if someone wants to review it :) <ng0>all I did was inherit original, and replace any "openssl" with "libressl" in the new package <ZombieChicken>anyone know what the program name is that provides terminal info so that programs like vim know what the terminal is capable of? <cantstanya>although ``programs like vim'' probably don't need to system() for that. <ZombieChicken>yeah. I'm reasonably sure ncurses is looking for something itself <efraim>libelf failed configure on aarch64 <ng0>hardened tor build looks functional <ng0>expected nothing else <ng0>I'll send the patch now <ng0>how could I express "version without dots" for a file name? do we have that? <ng0>of course I could just call it version 715 instead of 7.15 <ng0>ah, I can take the gnurl example <adfeno>ng0: About version without dots: As far as I know, you have to use a regular expression to match only "." and replace with nothing. <adfeno>Assuming "." is properly escaped, that is. <efraim>I think until we get something working for coreutils-8.26 it might be best to just revert the update ***ShalokShalom_ is now known as ShalokShalom
<ng0>now i have to construct something to prevent the gnu build-system from thingking that file COPYING is the source <adfeno>ng0: Care to explain the situation in detail? <ng0>no I can handle it myself <ng0>ok... why can't I replace the unpack phase in the gnu-build-system? <adfeno>ng0: Is it by chance a (tar/zip)bomb of somesort? <ng0>just a zip file lacking any directory structure <adfeno>Hm.... so you open the zip file, and you immediatelly see lots of files? <ng0>I just need to replace the unpack phase <adfeno>We might be able to explore the mailing lists to take examples from my previous issue with the Crypto++ zipbomb. <adfeno>I remember that the example given to me worked. <ng0>but zpaq is not really a zipbomb.. it's just a single directory <adfeno>Then perhaps you have to move it to inside another directory. <ng0>i tried by replacing in the unpack phase <nckx>ng0: are you packaging zpaq? You seem to have the same problem I did. <ng0>yeah I kinda need it.. :) <ng0>I already submitted one package which had the same problem (no directory) but for this one it just worked <ng0>but the other one had a directory inside <ng0>so I just had to cd .. <nckx>I think I have a working package, it just needs a snippet to remove some pre-compiled binaries. <ng0>now I wanted to replace unpack and add a mkdir src; cd .. but it didn't work <ng0>would be cool if you can submit it :) <adfeno>Now we just need to adapt to the zpack case. <ng0>The problem here is that this .zip file is a “tar bomb”, meaning that it <ng0>unpacks everything in the current directory, which the apply-snippet <ng0>logic isn’t prepared to deal with. <ng0>so it really is a * bomb <ng0>thanks for clearing that up :) <ng0>as I'm no longer doing this, it'd be useful for nckx <ng0>the package I submitted had the advantage of having one directory inside (doc?) so I could just chdir .. <nckx>adfeno: yes, I've already packaged it. I just replaced the unpack phase with my own. Yesterday I bookmarked an e-mail from Ludo' in which he mentions the /*bomb variants, but haven't tried that yet. <ng0>I need it in Guix only when I have the 0.1 version of my fish-guix done. I prefer zpaq over tar <nckx>ng0: I'll see if there's a /zipbomb unpacker to make it more succinct, then send-mail it today. <adfeno>If I recall correctly, that mailing list topic should have resulted in a improved dowload.scm. <nckx>adfeno: not yet. The patch in that e-mail was never applied. I'll include it if it solves the problem. Thanks for pointing me to that thread! <ng0>Is it somehow different from other system when you want/need to develop with Java on GuixSD? I might need it in the near future <adfeno>ng0: Oh... that is... somewhat interesting question. <ng0>I'd hate to run a VM with Gentoo when I start university (that's the current plan for winter 2017) <ng0>all you learn here is C++,C, and Java <nckx>The problem [I had] with zpaq wasn't the unpack phase (that was trivial to override), but the fact that snippet application in guix/packages.scm repeats the same mistake (a very optimistic ‘first-file’, which leads to ‘chdir COPYING’) with no obvious way to override it. <ng0>dropping the attempt to learn C back in 2007 was a mistake though <ng0>nckx: ah, that's the problem :) <ng0>I've picked up the debbugs package I made a while back. Doesn't look so terrible incomplete <nckx>ng0: It would have been, once you'd fixed the unpack phase, so I jumped in to spare you the inevitable disappointment. ;-) <ng0>well I was able to unpack <ng0>but I ran into can not chdir to COPYING <nckx>ng0: that's still done in 'unpack though. <ng0>rekado_: did you apply any further changes with darcs I could integrate in my darcs branch? Otherwise I'd just use that branch (the state I originally submitted) to test wether https will work one day <ng0>my guix checkout feels like the literal image of 260 keyboards placed on one desk. too many branches <ng0>oh, I found a webbrowser hidden in branches <ng0>I have mbedtls here.. but as a proof of concept I want hiawatha to work, because I have no idea if it actually works. <ng0>I think I will soon be done with everything I ever maintained or wrote on Gentoo (outside of portage) <csanchezdll>civodul: I am trying to test cross-compiled guile on ppc as you suggested on the mail <csanchezdll>but for some reason getting my guix up to date with core-updates makes tcsh testsuite to fail, so I am blocked solving that <ng0>I think I will submit mbedtls now.. the hello binary works <ng0>I will make a note for myself that I need to continue the conversation with upstream <efraim>Tcsh hangs on the test suite for me on aarch64 on core updates <efraim>csanchezdll: I would just disable the test suite as a work around <ng0>that's still outdated tcsh though and not my work in progress, so nothing I should consider I guess <ng0>*take in consideration <civodul>csanchezdll: there should be substitutes available for everything, no? <efraim>Most of the failures seem related to coreutils-8.26 so I was planning on reverting that upgrade <jmd>What is different in that version of coreutils which makes things break? <civodul>coreutils.git likely has the fix, it's just a makefile issue apparently <efraim>I saw manolis patch on the coreutils mailinglist, he wrote that it didnt actually work for him <catonano>given a file in the store, how do I know which package it belongs to ? <jmd>catonano: In general you can't. <jmd>Althought you can get an idea from its name. <jmd>Some store files don't belong to any package. <catonano>jmd: there are some conflicts when installing geiser. It randomly chooses one off the conflicitng libraries and I'd like to know why those libraries are there <catonano>this is not gonna kill me, it is just a curiosity <jmd>There are some screenshots on the web page. <jmd>But if you're going to install it yourself you can get your own. <avoine>I'm trying to fix guile extension in gdb but building it with: guix build -f gdb.scm --with-source=./gdb-7.12/ (the extracted tar + patch) <avoine>I'm getting errors like this: echo: write error: Bad file descriptor or ./config.status: Permission denied when the configure script <phant0mas>civodul: efraim I was hoping coreutils devs would comment on my patch <sneek>phant0mas, you have 1 message. <phant0mas>I found the exact coreutils commit that caused the issue <avoine>Ignore me, that must be because /tmp/ is full <phant0mas>civodul efraim I will send the updated patch to coreutils, patch our coreutils package, link to that discusion and push the patch to core-updates <phant0mas>sneek: later tell GNUnymous_IRC not yet, soon :-) <phant0mas>sneek: later tell GNUnymous_IRC the longer answer is you can, you just need to place the tarballs in the correct directory while using core-updates branch <efraim>phant0mas: sounds good :) I'm planning on regenerating my aarch64 bootstrap binaries in preparation of sending in some patches <phant0mas>efraim civodul Have a quick look at the coreutils patch for guix. It works for us, but if the coreutils guys have a different fix I will just update it. <efraim>Im on my phone, I'll try to take a look at it later tonight when I'm at home and the kids are down :) <ng0>hi, gmime fails some tests <ng0>or at least the one from 2 days ago :) ***kelsoo1 is now known as kelsoo
<ng0>nvm, I had an old checkout <rekado_>civodul: would you be opposed to using lalr-parser in bournish to support more complicated syntax? <bavier`>rekado_: a past gsoc project was to implement a more complete shell implementation for guix, but we didn't get to the point of deciding where it would live, as a replacement for bournish or as its own guile language <civodul>phant0mas: cool! did you send the patch to guix-devel? <civodul>rekado_: definitely not opposed, it's the right thing to do <civodul>bavier`: like for (guix gcrypt) and so on, it can start its life in guix.git, that usually makes things easier <phant0mas>civodul: the coreutils patch got accepted. Yes the guix related patch is in the ml <efraim>phant0mas: civodul: I'm testing the coreutils patch now, trying to compile aarch64 bootstrap binaries <efraim>phant0mas: you missed which package in base.scm the patch applies to, and I would mention that it reverts an upstream commit <phant0mas>efraim: oops :P you are right, I will send an updated patch <phant0mas>and it actually doesn't revert the upstream commit, just adds something they missed <phant0mas>I will also update the patch to remove the extra ":" to mirror the updated coreutils commit they accepted <jmd>I'm getting this error doing guix package --upgrade --fallback: guix package: error: reference to invalid output 'debug' of derivation '/gnu/store/giyh9cv0avvybhfa0imay4qwj1xsqwli-yelp-3.20.1.drv' <Apteryx>Could anyone give me a hint about how to connect a bluetooth device? I'm using the %desktop-services default services, which FWIU includes a bluetooth service called bluez. It seems though that I should add myself to the group "lp". <ng0>civodul: I will look into the shallow clone patch soon, thanks for the link <rekado_>Apteryx: yes, your account should be member of the “lp” group. <rekado_>Apteryx: check with “hciconfig” if your bluetooth hardware is up. If it is not run “sudo hciconfig hci0 up”. <rekado_>Apteryx: then you can use bluetoothctl to pair devices or control the device some other way. <Apteryx>rekado_: Thanks. I have to install "bluez" in my user profile to get this "bluetoothctl" executable, rigth? <rekado_>Apteryx: yes, it’s part of the bluez package. <efraim>From when I used to bluetooth tether my phone from the cli I had 'hciconfig hci0 up && hcitool cc $phonecode' <Apteryx>Currently, running "hciconfig" justs waits "to connect to bluetoothd". Maybe this is because I haven't reconfigured my system with group "lp" yet. <Apteryx>I'll do this and report of success. Thanks efraim & rekado! <ng0>ah so it's just a joke again :) <nckx>‘Hopefully we'll see other Linux distribution vendors do a similar maneuver this year!’ — it has to be, right? <rekado_>Why is this considered to be a good thing? <lfam>"Hopefully, we'll see other Linux distros drop support for some of the last trustworthy hardware available." <nckx>ACTION is talking through a i686 VPS right now :-/ <jmd>I hope guix doesn't do this too soon. <lfam>jmd: I doubt we'll do it *ever* <lfam>Or at least as long as one of us has 32-bit hardware <lfam>Honestly, I wouldn't look for leadership from these sites. <lfam>I don't like to say negative things here, so I just filled my quota for the quarter. <lfam>LWN has an article about package managers today. Maybe some subscribers would like to pipe up if functional package management is discussed <ng0>I don't like phoronix. <ng0>or was this another site? someone showed up here one day and requested to package a huge benchmarking suite and disappeared again <nliadm>phoronix seems to mostly be lots of low-quality posts <nliadm>and the in-depth articles are annoyingly paginated <adfeno>I don't like both. No clear license for content to allow sharing. :) <rekado_>adfeno: when in doubt it’s all rights reserved. <Walakea>adfeno, you don't like content that ain't free culture? <adfeno>Walakea: I don't like content that isn't at least shareable. <adfeno>(Free/libre culture is way beyond what I said) ***contrapumpkin is now known as copumpkin
***snape` is now known as snape
<Walakea>how far would you say GuixSD is from a stable release? <lfam>Walakea: What do you mean by "stable"? <lfam>We have full disk encryption now <Walakea>i mean, i would like to try it on my computer some day <lfam>Well, that's up to you :) <bavier`>Walakea: there are many here who run GuixSD as their primary OS for quite a while <jmi2k>I have a laptop running Guix almost 24/7, and I carry it in a USB everywhere. Something I miss from Arch Linux are the packages, but I help whenever I can :) <Walakea>do you need a Guile script for everything? <adfeno>I use Guix on top of Trisquel, but I'm still trying to gain courage to install GuixSD. <adfeno>And, I even contribute to some packages :) <rekado_>Walakea: you need a Guix package expression as much as you need a *.deb for a package on Debian or Ubuntu. <snape>jmi2k: you won't miss AUR packages :) <rekado_>Walakea: i.e. you can build software without it, but to have them managed by Guix you need package expressions. <jmi2k>Something I love of Guix is how *easy* is contributing and hacking into it. Here was the first time I sent a patch to a project. <Walakea>are Guix packages principially easier to package? <nliadm>I've found the average case to be about as easy <nliadm>I sometimes have to go digging in the scheme or guix docs if I need to do something a bit more complex <Walakea>do you think Hurd will ever be usable for normal work? <rekado_>I find rpms more difficult. (I packaged a lot of software with rpm before finally switching to Guix at work.) <nliadm>but I packaged most of the proprietary stuff I use for $dayjob with little hassle <jmi2k>Well, I'm still getting used to Guile and there are some packages I tried to package and failed. But <rekado_>Walakea: it is already usable for some work. But it doesn’t run on commonly available hardware at this point. <Apteryx>Back from a successful system reconfigure from Guix master. Bayfront server provided valuable substitutes such as the Linux-Libre 4.9.5 kernel! <nliadm>I'm still looking for an easier way to specify packages we build from git master <Walakea>what is the reason for Linux code base being so huge? <jmd>Walakea: Mostly, yes. <Apteryx>rekado_: Shouldn't the bluez service be started already if I'm using the %desktop-services list? <Apteryx>I'm getting "sudo systemctl start bluetooth" when using bluetoothctl which apparently means the service is not running <ng0>that's systemd though? <ng0>oh.. bluetoothctl outputs that <adfeno>Or is my connection slowly dropping? <rekado_>Apteryx: the service is called “bluetooth”. You can check its status with “sudo herd status” <Apteryx>rekado_: Ah, I understand my mistake now. I thought the "bluetooth-service" was part of the %desktop-services list, but it's not. It has to be added explicitly to my config.scm. <Apteryx>Maybe my reading comprehension is bad, but I got that impression from this sentence in the user manual: " The actual service definitions included in ‘%desktop-services’ and <Apteryx>provided by ‘(gnu services dbus)’ and ‘(gnu services desktop)’ are <civodul`>jmd: the "reference to invalid output" problem is post 482fda2729c3e76999892cb8f9a0391a7bd37119, right? <civodul`>jmd: do you have an easy way to reproduce it? <civodul`>"guix build yelp" works for me, for instance <ng0>108 letters in unixpath length maximum error in an out of tree package i have.. regardless of my further findings and solutions, is this something guix has considered, maximum file name length and maybe shorten the build dirs in /tmp/ to point to a symlink with the long name or smth like that? <ng0>I can't post the link to the bug report right now <ng0>"ubuntu has already started the same" <ng0>the resulting builds i have in store for this particular package is: <ng0>zsqqrklysrh25n8kfdbq7qc7aar97qpj-g <ng0> nunet-git-full-tests-0.10.1-2.15719ea/ <Apteryx>Any suggestion as to how use my config.scm interactively the same way I can use the guix scheme files? I can't associate a running geiser REPL with it. <ng0> /gnu/store/zsqqrklysrh25n8kfdbq7qc7aar97qpj-gnunet-git-full-tests-0.10.1-2.15719ea/ <ng0>this is way beyond what we do normally, right? <bavier`>ng0: we don't usually have package names that are that long <Apteryx>It seems the only problem with my config.scm and Geiser was the "implementation" being set to Chicken Scheme rather than Guile for some reason. I feel silly now ;) <Apteryx>There most be some dir-locals within the Guix tree which was setting this to Guile. <Apteryx>OK, so I managed to pair & connect the bluetooth keyboard... but I can't input any text yet. Is there something more I should do? <Apteryx>I think I need the hid-apple module. <Apteryx>it's one of those mini apple bluetooth keyboards. <Apteryx>Seems to be GPL'd and included in Linux-libre so I'm not out of lock yet (despite the Apple in the name). <ng0>you would also be out of (caps)lock if you would be out of luck <Apteryx>I have no idea how to get a list of the running kernel modules on GuixSD <Apteryx>there is no "--list" option to modprobe. <ng0>lsmod or what it was <Apteryx>Oh, right! I had forgotten that one. Thanks. <Apteryx>OK. So the device is paired & connected, and the required hid* modules running. Not sure why I still can't input anything using it. <ng0>restart X / init? i have no idea <Apteryx>That's a good idea! I wonder if the "paired" state will continue to hold. We'll see! <Apteryx>(I paired interactively using the bluetoothctl CLI tool). <Apteryx>Hmm. Still no joy (apple wireless keyboard) -- connected & paired but doesn't register input. <Apteryx>ng0: I tried restarting the xorg-server service with herd but it didn't seem to help. The log at /var/log/Xorg.0.log doesn't contain anything interesting. <Apteryx>It works! Not sure which of the following did it: 0) connect & pair using hcitool + bluetoothctl, 1) "sudo modprobe hid-apple" 2) Restart Xorg 3) Disconnect & Reconnect keyboard with bluetoothctl. <Apteryx>After that, the A1314 apple wireless keyboard got registered in the kernel (visible in dmesg) as an input. <adfeno>Apteryx: Try seeeing if it stays that way after reboot, just in case :) <adfeno>But before that, take note of the steps done, and try jumping one or two. <Apteryx>Probably not, unless the bluetooth controller will stay on (I had to do "sudo hciconfig hci0 up") <rekado_>Apteryx: you can add a udev rule to start the bluetooth interface automatically. <rekado_>(I don’t know the magic incantation, but I know it exists.) <Apteryx>rekado_: OK! Thanks for mentionning it. I'll look into it tonight. I guess this can all be done using gexps from my config.scm file? <rekado_>Apteryx: yes, you can add the udev rule in your config.scm <rekado_>it’s just a custom string delimiter because I wanted to play around with strings without having to escape quotes. <Apteryx>Seems GnuCash (accounting software) is not able to save data it keeps across sessions (for example, it the tips pop-up always come back, no matter how many times you untick the "don't show me again" checkbox). <Apteryx>It also doesn't remember about the last opened file (it's usually re-opened automatically). I'll file a bug tonight. <jlicht>Does anyone have weird input issues using the developer console in IceCat 45.5.1 using recent master guix? <jlicht>to be specific, after entering an expression such as `2 + 2', and presssing enter, the expression is evaluated, but the console input line now contains `2 + 2' and a newline <Apteryx>jlicht: I have that too. It's ugly but it doesn't seem to cause an issue. <Apteryx>Other than you have to manually clear it sometimes. Is this only in IceCat? <jlicht>Apteryx: I have no other installed browsers with developer console, so I couldn't say <Apteryx>I'll try Firefox 45 ESR I have on Windows