IRC channel logs


back to list of logs

<snape>re cgit and http clones: it's just that cgit only supports being a "dumb" HTTP endpoint
<nckx>Makes sense. Thank you for saving me some time, snape.
***usuario is now known as bzp
<mbakke>I finally had the ssh-daemon startup problem that lfam mentioned.
<mbakke>It was after a fsck, fwiw.
<axd[m]>Hello fellow guix hackers. I've got a question about the tlp service/package in guixSD. I've got it running, and it's been helping me manage battery life, at least in theory. I have a thinkpad, so it's possible to install the kernel module 'acpi_call' in order to have control over one's battery charging thresholds. The issue is that I can't find a package that provides that. Anyone got that to work?
<RetardedOnion>axd[m]: acpi_call, tp_smapi,tpacpi are all not there as far as i am concerned.
<axd[m]>axd: that's unfortunate. Big difficulties with packaging them?
<axd[m]>damn, I'm the retarded onion heh. I meant to go RetardedOnion
<axd[m]>Is my battery going to wear out more if I can't have these charging thresholds?
<RetardedOnion>axd[m]: if you set the threshholds correctly, you are going to wear out faster without them
<axd[m]>RetardedOnion: Yeah, that's the assumption, seeing as that's their main use case I suppose. I used to have it at 85% max and start charging at 80%, I think it was supposed to help out but I don't really understand battery tech. Any ideas if it's hard to package this kernel module for thinkpads? Hard to believe that #1 libre hackers' laptops' battery threshold module is still not packaged on the GNU's distro.
<pkill9>speaking of thinkpads, i'm thinking of getting one, what's a decent cheap portable one with decent battery life?
<RetardedOnion>axd[m]: guix hasnt got too many packages. so dont expect everything to be there. there are headers for the 4.14 kernel in the repos. so it is possible.
<RetardedOnion>pkill9: the latest ones that arent shite are the xx30 series. x230 or t430 id say. t430 can have easy quadcores if you desire. x230 only if you have a bga rework station and a cpu that cannot be found anywhere.
<pkill9>what's a bga rework station?
<RetardedOnion>for a very cheap but still great one get a t420. you have the meme bonus as well. the cpus from ivy (xx30 series) are possible with coreboot
<axd[m]>pkill9: I have an X230 because it is the most modern one that's capable of coreboot flashing (have to use a raspberry pi externally to flash the rom chip. ~$80 investment with all the cables and an SD card if you don't have one already)
<RetardedOnion>pkill9: to replace soldered on gpus/cpus. you dont have that if you ask.
<axd[m]>I think he means that X230 has soldered cpu, so you can't install a quad core.
<axd[m]>My i7 with 3.6 GHz turbo boost is serving me good.
<RetardedOnion>axd[m]: you can. you just have to solder. you cant do that by hand. so....
<axd[m]>$16 of ram. The only downside is the screen resolution is 720p, but I use mine with an external monitor, so I'm good.
<pkill9>i was silly and ordered an x220 tablet, i'm hoping the seller cancels and refunds the order
<axd[m]>heh 16 GB of ram. Payed $240 on craigslist in the bay area. Best deal.
<pkill9>axd[m]: which resolution is 720p?
<RetardedOnion>i have a t420 with a quadcore 2820qm, 16gb ram, 256gb 860 evo. 1600x900 screen. payed 200€ for everything. will buy a modded 1080p ips screen tomorrow. and liquid metal.
<RetardedOnion>pkill9: 1280x720
<pkill9>ahh yah that's small
<RetardedOnion>pkill9: you can get a 1440p screen on the t420/430.
<axd[m]>yeah I'm not a huge fan of the tablets since the screens scratch a lot on the keyboard. I've seen some in bad condition. And keyboard input is superior anyway imo
<pkill9>how heavy is the t420?
<pkill9>yeah also apparenlty the tablets are heavier and don't support 9 cell batteries
<pkill9>is it light enough to carry around?
<RetardedOnion>do you use the touchpad? the screens are worse, they are heavier and less robust
<RetardedOnion>pkill9: i use my t420 on the bus. grab it by the 9cell.
<RetardedOnion>i meant to say touchscreen
<axd[m]>it's just over that at 1366x768. You can do a 1080p mod on an X230 though: search 1080p X230 screen on youtube. See vid by Louis Rossman. Interesting motherboard repair channel.
<RetardedOnion>though you have to solder...
<pkill9>RetardedOnion: i was thinking of using it like tablet for media consumption as well as for laptop usage, but it was somewhat of an impulse purchase and i didn't research enough to find out the downsides
<axd[m]>pkill9: that's a killed feature, honestly. The utilitarian design also helps with anti-theft ;)
<RetardedOnion>if you want a t420 or a x2x0 is completely up to you. you can have a quadcore on the t. tablet on x
<pkill9>but the seller seems to be away, hoping they refund me, i was the only bidder anyways
<pkill9>otherwise i'll have to try to sell it >.<
<RetardedOnion>honestly: the touchscreen doesnt work as well as i want it to. so touchscreen thinkpads are not of interest for me
<pkill9>1080p is 1960x1080 isn't it?
<axd[m]>RetardedOnion: I actually just received a new 9-cell battery for my X230. It's supposed to be new old-stock, that are genuine. What's the proper way to initialize a battery on thinkpads? Should I go starting-% -> 100% -> 0% -> 100%, and then start using it, do the reverse or do neither and just don't think about it after plugging it in?
<axd[m]>RetardedOnion: also do you use GuixSD as the main OS on your X220?
<RetardedOnion>axd[m]: boot into a tlp and tp_smapi/tpacpi_call compatible distro. boot it up, plug the battery in. sudo tlp recalibrate bat0. let it charge until 100%. set up thresholds AFTERWARDS
<RetardedOnion>axd[m]: t420 with quadcore. no i dont. i have it on a secondary ssd. it has so many missing packages which makes it really unsuitable for me at the moment. i run arch.
<axd[m]>RetardedOnion: you mean set up thresholds using tlp on debian for e.g. (that's my backup OS that's on the drive). Will they keep working even once I boot into guix that has tlp without the kernel module?
<pkill9>RetardedOnion: how much did you buy your thinkpad for?
<RetardedOnion>axd[m]: no they wont. they get cleared after a reboot. the ec does remember it in sleep
<mbakke>axd: Lithium-Ion batteries do not need any sort of "initialization".
<RetardedOnion>pkill9: germany. 80 with 900p screen. really good price. 150 is fine as well. 40 for cpu, 40 for ram
<RetardedOnion>mbakke: its for the ec. its not really good.
<mbakke>Rule of thumb is don't let it go to 0%, and don't keep it at 100% for too long, and you'll be good.
<mbakke>For long term storage, ~50%,.
<RetardedOnion>pkill9: is german(google translate will help), but for the quadcore(not necessary) download link for the modbios . you need a 90w psu or higher for the battery to not discharge when running. so.... get a 90w or 170w psu if you intend to quadcore. cooling is an issue. so if you want 8m cache go for the 2820qm. not too
<RetardedOnion>much higher
<axd[m]>mbakke: RetardedOnion: sooooo, should I do the whole tlp recalibration and setting the thresholds in debian? Do the thresholds get set in the battery firmware and would remain valid on next boot even in GuixSD? Or did mbakke mean that I don't have to do any of this at all. I don't really know how not to keep it at 100% all the time though. I'm a student so it does get some use, but sometimes it may sit in the dock for a few days
<axd[m]>without coming out, constantly plugged in. That's where the thresholds are nice from what I understand.
<axd[m]>oh I see that you said the thresholds won't stick. Is there any reason to do the `tlp recalibrate` at all then?
<RetardedOnion>axd[m]: it is not persistant between reboots as far as i know. you should recalibrate for the ec. especially if you have a xx30.
<axd[m]>Because I'm using GuixSD on my machine primarily. Debian on 20GB on the SSD for extreme cases and another debian in a VM, in libvirt
<RetardedOnion>you should recalibrate for the ec. lenovos genuine hunt hurts more than it should
<RetardedOnion>thresholds just help the battery to survive longer. recalibrating will help the ec to not give you bad information.
<mbakke>I had upwards of 20 hours for two years on a x250 9-cell without any kind of calibration.
<RetardedOnion>mbakke: he said the battery was used. also the ec improved on the xx50 series. was still shite at the xx40 series
<mbakke>It was probably good for three more years.
<mbakke>RetardedOnion: I see.
<mbakke>With bad information, you mean "don't suddenly shut down at 15%", right?
<RetardedOnion>mbakke: yes
<RetardedOnion>turn off. not shut down
<mbakke>Yes indeed :P
<mbakke>Can TLP send a sleep event when it reaches a certain threshold?
<RetardedOnion>mbakke: i dont know why it should. and i dont know why it shouldnt. i will have a look
<mbakke>Or hibernate, probably more useful.
<mbakke>Somehow I managed to not try TLP in all these years :P
<RetardedOnion>mbakke: tlp doesnt have a background daemon. it has a service that sets thresholds. (systemd, dont know about anything else). so it cannot do that. there are udev rules if you like. it also gets triggered when connecting power, disconecting and so on.
<mbakke>RetardedOnion: Oh, okay.
<axd[m]>RetardedOnion, so EC where the battery firmware lives? What do you mean when you say "lenovos genuine hunt hurts more than it should"? The battery isn't used, but it's new old stock, meaning it was manufactured sometime 2014-16 (is there a way to check by the serial), and lost some capacity, just sitting in storage, according to the seller.
<RetardedOnion>axd[m]: the battery firmware lives mostly in the ec. as in every notebook. yes. it has lost capacity. that is why you should recalibrate for the ec to notice.
<RetardedOnion>lenovo tried to verify geniune batteries since the t61(i think) (so forever) and they fucked up badly. that is why you pretty much have to recalibrate them before using them. the ec isnt really analyzed at all on the thinkpad. even with libreboot it doesnt get replaced/changed in any way
<pkill9>axd[m]: did you mean to say 'killer' instead of 'killed'?
<pkill9>RetardedOnion: which packages are you missing from GuixSD?
<RetardedOnion>pkill9: tlp stuff, kde, qt stuff, notepadqq, headers for 4.17, syncthing-gtk, kdocker, newer xfce, networkmanager
<axd[m]>pkill9: you're correct. I /did/ do an `s/killed/killer` after that message though ;)
<axd[m]>RetardedOnion: when you say headers for 4.17... isn't is
<axd[m]>... isn't it possible to already install 4.17 on guix? Would the headers have to be packages for that to be possible?
<RetardedOnion>axd[m]: no. you dont need headers for much. pretty much just dkms. 4.17.2 is in the repos
<axd[m]>RetardedOnion: also, would it be possible to do the `tlp recalibration` in guixSD? I don't think the acpi kernel module is needed for that, but you'd know better.
<pkill9>axd[m]: i heard the tablets have parallax issues (like upto 1cm inaccuracy near the screene dges). Do you have one of the tablets?
<RetardedOnion>axd[m]: no. its not possible at this point.
<axd[m]>or you would still recommend to boot to a proper Debian?
<RetardedOnion>axd[m]: i do. still. recalibrate it once, you are fine then
<axd[m]>RetardedOnion: oh really? Interesting, well, I'll be back then. Thanks for the suggestions and the discussions.
<RetardedOnion>pkill9: that is true. also if you use it extensively and it gets warmer, you get noticable latency. its more fragile as well. id strongly recommend against tablets. they are fine for collectors. i actually have an x220t and an x200t.
<axd[m]>pkill9: I just realized that I messed up. The `killer` comment was actually in reference to what onion said about holding a thinkpad by the battery hump of the 9-cell battery. I have never used a thinkpad tablet in person. Although, I've seen someone use one a few years back, and I thought it was the coolest thing ever, seeing how they had the proper stylus and everything setup and were able to just "flip" and have a tablet on
<axd[m]>such a workhorse of a machine. They were using microsoft's os, so had good driver support. The tablets, I think, are bit more "proprietary" than a usual X230/220 and harder to install coreboot on, so I didn't bother. I use Emacs as my Window Manager, so it's not like I was gonna get much use out of the touchscreen, although with GNOME, it could be something. But the screen resolution remains a bit lacking.
<RetardedOnion>i dont know about gnome on a tablet. xfce is not good. kde eats a lot of power but works kinda ok. i guess. the hardware issues, coreboot compatibility and not so good screens really dont make them good machines for usage imo.
<RetardedOnion>also: batteries are smaller and not easy to get. expensive as well.
<RetardedOnion>i have a fetish for the older thinkpads. i dont own any thinkpad, but a shit ton. if you got questions about them, just ask
<happy_gnu[m]>RetardedOnion: how do you fight overheating
<happy_gnu[m]>I have a thinkpad x60t
<happy_gnu[m]>it gets hot
<RetardedOnion>happy_gnu[m]: liquid metal and liquid rubber. what cpu?
<RetardedOnion>doesnt really matter, its max 17w afaik. should be fine if you clean it. i got rid of my x60t a long time ago. it wasnt too warm
<RetardedOnion>good thermal paste should be fine though. thermal grizzly kryonaut should already be overkill, but its not that expensive. so get that. make sure you clean everything
<apteryx`>hello! given that the 'doc' output of gnucash is a symlink to the gnucash-docs output, shouldn't "gnucash-docs" be part of gnucash's inputs rather than native-inputs? (otherwise a gc might remove it after GnuCash is installed, right?)
***payphone_ is now known as payphone
<g_bor[m]>Hello guix!
<g_bor[m]>I've seen a bug about java-bouncycastle test problems, but could not reproduce it. Could you please confirm, that it still persist.
<g_bor[m]>Also where are we in the staging cycle?
<civodul>Hello Guix!
<g_bor[m]>I've checked the references in icedtea-packages. It seems, that stripping the binaries removes the references. I did not check, if this is safe to do though. WDYT?
<civodul>g_bor[m]: was stripping disabled until now?
<civodul>normally it's enable by default, with --strip-debug though (not --strip-all)
<Misha_B>hello, I'm interested in trying out guixsd, but I have a wifi card non-free that is non-free. Is there a way to be able to use the card?
<Misha_B>disregard that first 'non-free'
<civodul>that was fast :-)
<g_bor[m]>civodul: It seems, that these binaries are not stripped. I will have a look at the definition. Stripping with --strip-debug also removes the references.
<civodul>note that 'file' reports binaries as "unstripped" unless you use --strip-all
<roptat>hi guix!
<civodul>any ideas regarding UEFI doc?
<roptat>g_bor[m]: I think the bug with bouncycastle tests is that there's one test that takes a long time without doing anything and reports a failure, but it's ignored (the phase returns #t)
<roptat>I need to fix my server, I don't receive any mail anymore :/
<civodul>oh that's bad
<civodul>or good? :-)
<roptat>hey, my smtpd listens on a weird ipv4 address
<roptat>I think I fixed the issue by restarting the service
<roptat>regarding UEFI, i remember my computer was able to boot the installation media in efi mode, but I installed it in bios mode
<roptat>it can boot either, but of course there's only one possible mode for each disk
<roptat>but it could be the case that some computers enforce one mode or the other
<civodul>hmm yeah
<roptat>but I guess the majority of users will want to install in the same mode as they booted the installation media
<civodul>right, and you can't really do otherwise
<roptat>so we can boot both modes from the same iso, right?
<roptat>how do we do that?
<civodul>roptat: yes, it boots in the "right mode"
<civodul>basically it has a EFI system partition that's ignored when in BIOS mode
<civodul>something like that
<civodul>i'm trying to fix the "" issue
<civodul>weird thing is install-info strips the ".fr" part in the dir file it generates
<civodul>but only under some circumstances
<roptat>was the plan to generate a with everything (including English manuals when untranslated), or only French manuals?
<roptat>ah there's maven-3.5.4
<civodul>oh, got it
<civodul>roptat: for some reason, /gnu/store/4lk7j5693ikj7yiw2qpsbva5vkicxmp4-guix-0.14.0-13.7af5c2a/share/info/ contains dir entries like "(guix)Foo" instead of "("
<civodul>any idea why?
<roptat>I thought I fixed that
<civodul>oh ok, maybe that's because it's an old snapshot
<civodul>in which case, problem solved
<roptat>but probably not completely
<roptat>some part of the texi file contains these (guix)Foo, and I translated them as (
<roptat>but maybe I didn't find all of them
<roptat>on master, there's still (guix)Invoquer guix gc. for instance
<civodul>roptat: here's what i have:
<civodul>it generates a dir.LANG, which contains only manuals written in LANG
<roptat>ok, but if I speak LANG, will I be able to read English manuals for untranslated ones?
<g_bor[m]>civodul: I guess there were some improvements regarding the efi doc. I would really like to back that with system tests though. Had no time yet do it.
<roptat>and not have English manuals for translated ones?
<roptat>civodul: in the French manuals all @direntry nodes contain references to ( only
<roptat>but the .info file still contains some (guix) references
<g_bor[m]>It would be really nice, if we could generate parts of the manual from system tests. That would ensure that they are in sync. WDYT?
<civodul>roptat: it generates dir with only English manuals, and dir.LANG with only LANG manuals
<civodul>it half a solution anyway because the dir.LANG convention is not recognized by Info readers
<civodul>we'd need proper support from Info readers
<civodul>snape: i'm getting bounces from you: " Network error on destination MXs"
<snape>thanks for letting me know
<civodul>interesting comment on 2FA on GitHub:
<efraim>I'm working on the qt-5.11.1 update, I assume I should push it to a new branch?
<civodul>efraim: probably a qt-update branch, to isolate any issues
<snape>civodul: fixed: for some reason my primary MX (opensmtpd) didn't start after reboot. But my backup MX was working and did receive the emails, so I don't really understand why the bounces were sent
<efraim>civodul: sounds good, so far 5.11.1 is looking good
<civodul>efraim: and then just guix build everything guix refresh -l mentions, or wait for berlin/hydra
<civodul>snape: cool! this is opensmptd on GuixSD?
<snape>yes, for both MXs
<efraim>owncloud-client got fixed with 5.11.1
<efraim>blogpost and code snippets! inquiring minds want to know
<civodul>+1 :-)
<civodul>snape: ↑ hint, hint :-)
<snape>oh you are talking to me... :)
<snape>when I find some time!
<snape>but inquiring minds can already see the code snippets, it's public
<civodul>heh, true
<civodul>so are there more people experiencing the 'guix substitute' hang now?
<snape>bounce-warn ... will generate a failure report when an envelope is in the queue for more than one hour, six hours and two day. The default is 4h.
<snape>makes sense. civodul: did you receive the bounce 4h after you sent the email?
<rekado_>snape: I also got a warning for an email to you.
<snape>yes, rekado_, you were in the queue too, sorry about that :/
<rekado_>no worries
<snape>I'll try to disable it or increase the time delay. 4h seems a bit short
<rekado_>since the addition of the “install-license-files” phase we sometimes end up with packages that do nothing but install their license files
<rekado_>(e.g. because the build or installation failed in an odd way)
<rekado_>previously, these builds would be marked as failed because they did not result in any output.
<civodul>sneek: later tell snape looks like i received it 4h later; note that it just says "delayed"
<sneek>Got it.
<civodul>rekado_: such packages should be rare though, no?
<civodul>fun, there's a bug report/support request here:
<rekado_>here’s a dangerous idea: implement the reckless-build-system for software that does not conform to any conventions.
<rekado_>the complete build output would just be dumped in a squashfs image, and the image would be installed.
<snape>hey sneek
<sneek>Welcome back snape, you have 1 message.
<sneek>snape, civodul says: looks like i received it 4h later; note that it just says "delayed"
<snape>cool thanks
<roptat>\\o/ I've fixed the "mvn can't compile java8" bug :D
<roptat>I guess the issue was mvn had too many things in its CLASSPATH
<civodul>rekado_: this is eviiil!
<civodul>i'm sure even Docker, Inc. hasn't planned for this ;-)
<civodul>(maybe Anaconda, Inc. did?)
<rekado_>from what I understand about how some Conda packages are built, something close to this is already done.
<civodul>arf, see? we can't compete
<rekado_>they also leave the build environment connected to the Internet, IIUC
<civodul>more convenient i guess
<rekado_>and it runs in a Docker container! So it must be safe.
<civodul>sure thing!
<jonsger>guix can also run in a docker container, so guix is save :P
<civodul>in Nix there was (is?) an escape hatch: you could add a special attribute to a derivation to have it built outside of the normal chroot environment
<roptat>\\o/ and the "maven fails to run tests undeterministacally" bug is a bug in upstream maven-surefire-plugin :)
<roptat>I can reproduce with a binary version of maven
<roptat>and that's fixed by downgrading maven-surefire-plugin
<rekado_>the stats for Guix here are really not correct because they just limited the packages to the bioinformatics module.
<rekado_>hmm :-/
<civodul>everything but the acknowledgements is behind a paywall
<jonsger>civodul: you can click on figures, there is the graph
<civodul>hmm, small figures :-)
<jonsger>rekado_: how many bio packages do we have?
<rekado_>oh, I’m accessing this from the campus network, so I don’t see the paywall
<rekado_>I really don’t know.
<rekado_>depends on how you count.
<rekado_>they say: "Bioconda provides packages from various language ecosystems such as Python, R (CRAN and Bioconductor), Perl, Haskell, Java, and C/C++"
<rekado_>so … I guess anything goes?
<civodul>here we go:
<civodul>(behind captchawall this time)
<rekado_>their counts are also inflated because they keep old versions.
<rekado_>apples and apples.
<civodul>"number of explicitly bio-related packages"
<civodul>maybe we have too many implicitly bio-related packages
<rekado_>hah :)
<civodul>looks like anything goes to Nature
<civodul>"for each Bioconda package, our build system automatically builds a minimal Docker image"
<civodul>"the desired level of reproducibility can be chosen freely"
<civodul>as opposed to those who impose just too much reproducibility!
<roptat>we can also build a minimal docker image :)
<rekado_>“For reproducible data science, it is crucial that software libraries and tools be provided via an easy-to-use, unified interface, so that they can be easily deployed and sustainably managed. With its ability to maintain isolated software environments, integration into major workflow management systems, and lack of requirement for any administration privileges for use, the Conda package manager is the ideal tool to ensure sustainable
<rekado_>and reproducible software management”
<rekado_>wait a minute…
<rekado_>the first sentence is about ease of use; the second just claims the reproducibility crown.
<civodul>boils down to: "because everyone uses it it's ideal"
<rekado_>gotta go to a meeting, bbl
<civodul>rekado_: take advantage of that meeting to draft a rebuttal :-)
<jonsger>do they have anywhere raw numbers or do I have to extract the information from this micro chart?
<roptat>hey! maven is also reproducible and easy to use: every time you download a binary, you get the same one :)
<civodul>jonsger: the latter
<civodul>roptat: hey that's good! :-)
<roptat>(but don't try to build anything, it's not user-friendly anyway)
<nckx>civodul: I read that comment too. It says many things. <>
<nckx>Which did you find interesting?
<civodul>nckx: for one, i don't have a "smartphone", so i wouldn't be able to contribute
<civodul>second thing, it's creates a bias towards non-freedom-conscious contributors because of that
<civodul>and then reliance on SaaS
<civodul>and questionable security gains
<nckx>Nor do I. I just set up GitHub 2FA using my GuixSD laptop.
<civodul>using the zbar thing?
<nckx>Just oath-toolkit (package incoming).
<civodul>in fact there's already two factors when authenticating to GitHub: ssh key + on-line account
<civodul>so i suppose it's 3FA?
<nckx>I obviously don't really like GitHub or think their ‘2FA’ adds much of value (sure, it adds some more trouble for a would-be attacker, but just by adding more trouble for the *user* which the attacker would want to impersonate).
<nckx>So I agree on the security theatre part. But as a non-phone-owner I almost believed the FUD about needing one to contribute. That's why I don't like that post.
<nckx>GitHub is already SaaS, and GitHub is already unfree, while the 2FA can actually be done with entirely free software.
<civodul>right, that's good, and it's true that the post i mentioned above assumed a phone was needed
<civodul>(which is also what i thought before reading responses)
<civodul>so it's "not that bad"
<civodul>bah, unable to report a bug at :-/
<civodul>the login page directs me to a "oops" page
<Gamayun>civodul: Ah. I get redirected to an 'Ubuntu One' sign-in page. I thought they scrapped that years ago...
<civodul>yes, i created an account there, which went fine, but it failed afterwards
<nckx>civodul: Sorry, didn't mean to rant. When first I read that comment (then without responses) I believed it, searched anyway, and found out it was random speculation. >:-(
<nckx>It was a good dry run for enabling mandatory 2FA for a site I actually care about, though, which is what I'm really afraid of.
<Gamayun>Hm, I can still log in to my old account there it seems.
<civodul>nckx: you mean you're afraid of mandatory 2FA on GitHub?
<nckx>civodul: Nah. With ‘site[s] I actually care about’, I meant sites other than GitHub.
<nckx>Anyway, I asked because I wondered if you were also referring to points like ‘make [a] bot merge all the PRs, requiring at least two reviews’. I guess not.
<efraim>i forgot how hard github is to read from links
<civodul>nckx: ah well, that's interesting too no? :-)
<nckx>civodul: I think so, but I'm not the one with significant experience in running a software project ;-)
<efraim>we had a suggestion before about using aegis(?) to manage patches, where no one could push their own commits, ensuring everything got at least one review
<civodul>i think we need a champion to actually deploy a tool and get people to try it out :-)
<efraim>qttools directory is 16MB, the bin dir is 4.9, looks like a candidate for splitting off
<efraim>perhaps not, only used as an input in like 5 packages of the ~80 times it appears
<rekado_>I have yet another bioinfo package here that consists of dozens of scripts that use the result of `which foo` to figure out if “foo” exists and which “foo” to use.
<rekado_>I have started patching these files, but maybe it makes more sense to wrap the program in a PATH containing these tools.
<roptat>bootstrap says "possibly undefined macro: GUILE_MODULE_AVAILABLE", what do I miss?
<roptat>I have guile-2.2 and pkg-config already
<mbakke>Speaking of tools, I've been eyeing the platform recently, and it looks very good.
<mbakke>Designed for email workflows. And is entirely free software, of course.
***benny is now known as Guest33773
<rekado_>mbakke: I’ve seen this too, but I haven’t figured out how to use it.
***Guest33773 is now known as benny
<lhart>redako_: ?
<roptat>somehow it draws well in my terminal, but not in my browser
<rekado_>oh, it’s an emoji
<nckx>I had support for those in Hexchat for a while, then I reconfigured my system, now it's gone.
<civodul>go figure ;-)
<rekado_>nckx: sometimes this kind of behaviour is due to caches.
<nckx>~ λ grep font- /etc/guix/system.scm → "font-abattis-cantarell" "font-fira-code" "font-gnu-unifont" "font-hack" "gs-fonts"
<rekado_>same with ibus, which sometimes breaks after reconfiguring the system because the cache files contain absolute file names.
<nckx>Neither ‘fc-cache -f’ nor ‘sudo fc-cache -f’ change the output of ‘fc-list’, which is missing those fonts.
<nckx>It found them previously.
<rekado_>do you find any references to /gnu/store in ~/.cache ?
<ng0>the end of a stacktrace:
<ng0>ERROR: In procedure seek:
<ng0>Value out of range -2147483648 to 2147483647: 4770726968
<ng0>cannot build derivation `/gnu/store/a8qvy51jl3r4mz75v9xxsznvsrmgg7xm-texlive-2017.drv': 1 dependencies couldn't be built
<ng0>does this look familar to anyone with very big stores?
<nckx>rekado_: Was that to me?
<rekado_>nckx: yes
<nckx>I haven't tried debugging this yet. Let me check.
<rekado_>ng0: these numbers are familiar
<rekado_>is that maxint?
<ng0>i'm not sure..
<ng0>inodes are 19% left
<rekado_>what file system is this?
<rekado_>I don’t have this derivation. Is it a grafting derivation?
<ng0>it's directly after an import from my builder
<ng0>retrieving 1 store item from ''...
<ng0>importing file or directory '/gnu/store/9rsxwkbzqj4x6scdnd9w5ial1dw3vzis-texlive-texmf-2017'...
<ng0>found valid signature for '/gnu/store/9rsxwkbzqj4x6scdnd9w5ial1dw3vzis-texlive-texmf-2017'
<ng0>then follows the backtrace
<rekado_>could you look at the derivation file?
<ng0>one moment
<ng0>it's my turn at presentation
<ng0>grafts had smaller content, right
<ng0>it's getting late here into 5 hours repeating the same stupid lines practicing, so I'm not so well focused :)
<nckx>Several ‘.cache/fontconfig/*-le64.cache-7’ matches for ‘/gnu/store’, I deleted the directory, tried ‘fc-cache -f’ again, no improvement.
<ng0>eh wrong folder
<ng0>right url
<ng0>sometimes eduroam interrupts the connection to my domain though
<ng0>so it could be just that for large files
<ng0>I'm triyng to retrieve it again right now
<ng0>i don't think this is the problem, but who knows
<ng0>I never had so much time and boredom to offload texlive builds
<ng0>we really should've made the blender rendering to more computers -.-
<ng0>*run on more
<ng0>no change. I'll try again at home
<cryptocat1094>So, for optional features of programs, manual package overrides are the standard way, correct?
<payphone>Hello. I'm having a bit of trouble with running `guix pull` as root. The command procedes without error, but afterwords `guix system reconfigure` will not work. The error is 'guix: system: command not found.'
<payphone>`ls /root/.config/guix/` returns 'latest'
<nckx>cryptocat1094: Yes.
<cryptocat1094>nckx: Thanks
<nckx>There are a handful of foo-with-bar variants packaged, but I assume you're asking about the general case.
<cryptocat1094>nckx: Yes, that was what I was asking.
<janneke>civodul, rekado_: working with Rutger to get gash in a usable state. I have just copied and adapted bournish, into my gitlab fork; hope that's OK.
<janneke>we want to fix some silly things before the 0.2 release and i'd like to figure out together how and where to package shelly guile scheme code
<roptat>payphone: that's due to a change in the way guix pull works
<roptat>payphone: you only have to run "guix pull" once more
<roptat>and adjust $PATH according to what guix will tell you
<payphone>roptat: Ah, thank you. I had read that `guix pull` had changed, but I didn't realize I would have to run it twice.
<vagrantc>it is certainly non-intuitive to have to run it twice
<roptat>g_bor: maven-compat tests should be fixed in 892a5647debc7fe8eafac4f725c46134eb085b63 :)
<civodul>rekado_: i think i'm going to punt on the 'guix substitute' hangs for this release
<jlicht>hey guix
<g_bor[m]>roptat: nice, thanks!
<civodul>we keep hitting synchronization issues once mingetty is started, like this:
<civodul>the PIO_UNIMAPCLR message is the same thing over again
<jlicht>which branch is Tatiana working on again?
<nckx>That's what my MUA just told me.
<jlicht>thanks nckx :-)
<nckx>You're welcome.
<jlicht>I was just happy to see things about pagination + sql come by :)
<jlicht>used to love that stuff when I still had courses about it