IRC channel logs

2019-04-16.log

back to list of logs

<vagrantc>i've sometimes wondered about re-implementing libreboot in guix ... so rather than a boatload of patches and shell scripts, you'd have a boatload of patches and guile code... :)
<apteryx>oh, is libreboot like linux-libre, taking some upstream and stripping the non-free bits from it through patches/scripts?
<vagrantc>yup
<vagrantc>in my experience, it also showed more information to the user, rather than just a white screen that flashes by in 3 seconds, it actually displays the text for the various key-combinations that do things
<nckx>vagrantc: That's just a setting though.
<vagrantc>maybe so, but there's a lot to be said for default behaviors :)
*nckx prefers the silent default so meh.
<nckx>Apparently flashing *boot without a BIOS (no, *boot is _not_ a free BIOS replacement) makes it impossible to run Windows on the device. \o/ Can't wait.
<apteryx>neat feature!
*nckx quips ‘now that's what I call Secure Boot’.
<vagrantc>well, you can load seabios as a payload, not sure if that would break it for you :)
<nckx>vagrantc: Linux doesn't need it & rescue images can be booted from GRUB, so I don't see the point (not sure about the Hurd though).
<nckx>Plus it adds _seconds_ to boot speed X-(
<vagrantc>so you can upgrade grub without reflashing?
<nckx>I've been less up on GRUB development lately but it seems pretty ‘mature’, no?
<vagrantc>it lacks luks2 support, apparently...
<nckx>Only thing I'd reflash it for is faster LUKS decryption, it's not like new hardware support is an issue.
<nckx>Ah.
<vagrantc>it has been pretty stable, overall :)
<nckx>TBH I wouldn't mind flashing once a year or so (this is all big talk from someone who hasn't flashed it at all, yet).
<nckx>For a big feature such as that.
<nckx>I assume it would involve re-encrypting my drives anyway so it's a Project in any case.
<vagrantc>apparently, guix hasn't yet updated to a cryptsetup version that defaults to luks2, though
<vagrantc>apparently luks1 -> luks2 can be done in-place
<nckx>Shows what I know.
<nckx>Flashing too, if nothing goes wrong 😛
<vagrantc>surprise to me too ... just was reading about all this earlier today
<vagrantc>mostly from https://bugs.debian.org/927165
<nckx>*click* Thanks.
<nckx> https://savannah.gnu.org/bugs/?55093: ‘cryptsetup now defaults to LUKS2 on all major distributions’ - obviously they forgot GNU Guix.
<nckx>Shame there's no response.
<nckx>…since November…
<vagrantc>time for everyone who wants grub and luks to switch to guix
<nckx>(^ hence my general impression of GRUB development.)
<nckx>Which is a shame, since none of the hip new fooloader fads provide a tenth of the flexibility and peace of mind that GRUB does. Bah.
<emacsomancer>`guix package -u` seems to have resulted in local building of 6 different versions of rustc. I haven't explicitly asked for rust/rustc to be built and it's be churning for several hours.
<emacsomancer>not sure why
<vagrantc>have icecat installeD?
<nckx>How does one ’explicitly [ask] for rust/rustc to be built’?
<nckx>Ah, never mind, I imagined a not where there was none.
<emacsomancer>yes, icecat is installed
<emacsomancer>I never know when things will be built locally and when there'll be binary substitutes. I'm not sure why it's trying to build so many different rustc versions though
<nckx>Because (roughly) the only way to build rust@n is with rust@n-1, which can only be built with rust@n-2, etc. You can't just compile the latest rust with gcc, for example.
<emacsomancer>ah, I see!
<nckx>The current path is already the minimal one ☹
<nckx>So I guess someone changed something that changed rust@n-1-1-1... and unfortunately the build farm is still sweating away at it.
<nckx>(Yes, it sucks.)
<tortoise>Hey, if I want to write a package that has a custom license, how do I create the license object?
***jonsger1 is now known as jonsger
<nckx>tortoise: Extremely custom as in unique to that package? (non-copyleft "file://path/to/licence") ; or http[s]://
<tortoise>unique to the package
<nckx>Yup. Just make sure it's really free ☺
<nckx>I'm not sure whether remote URLs or paths into the tarball are preferred. Maybe peek around the current packages using it.
<nckx>Anybody else with creepy low-pitched audio in some videos in Icecat?
*nckx is watching Satanic Announcer Baseball.
<emacsomancer>(my Icecat is still building, of course)
<nckx>Agh… The guix-daemon on my build farm had crashed and it wasn't building anything. That'll be a fun catch-up. ☹
<nckx>No new Icecat for me either.
<emacsomancer>I didn't have that video audio issue that last time I noticed.
<nckx>It's only certain files. Youtube's fine, my not-super-legal MLB stream is not. But it's not only live streams or dodgy sites.
<nckx>Let's go, Satan. ♪
<emacsomancer>I don't do a lot of video-watching inside of Icecat. I'm using pulling down things via youtube-dl. Doesn't work as well for streams, I suppose. Could you use VLC for the stream?
<emacsomancer>*usually
<nckx>emacsomancer: Youtube-dl chokes on what I presume is encryption. So MPV is out as well. VLC hadn't occurred to me to be honest.
<emacsomancer>ah
<nckx>Weird how my first choice on borrowed $proprietary_os boxes is my last on GNU.
*nckx installs the Cone.
<emacsomancer>I find VLC doesn't perform nearly as well as mpv (and I presume you have as well), but I've occasionally used it for streaming.
<nckx>Yup. I'm used to extremely low-end laptops that can't afford VLC. I just got one that probably can. Strange new world.
<emacsomancer>even on high end machines, I find that VLC isn't as good as mpv. Sometimes it stutters; and skipping in videos works much better in mpv. Plus, I think mpv's (admittedly minimal) interface is slick.
<nckx>access stream error: HTTP 401 error | local stream 1 error: Cancellation (0x8) | no suitable access module for `https://playback.svcs.mlb.com:443/silk/events/e10890c…
<nckx>When I paste the m3u8 URL ☹
<emacsomancer>ah, too bad
<nckx>Thanks for the idea. VLC hadn't occured to me and it was a good bet.
<nckx>^ agreed.
<emacsomancer>can you try in chromium?
<nckx>Sorry, no.
<nckx>That's a hard line.
<emacsomancer>even the ungoogled chromium?
<nckx>Yah.
<emacsomancer>I just like having a second browser (just in case). Ethics/privacy/etc. aside, Firefox/Icecat is technically superior to Chrom{e|ium} anyway and more pleasant to use.
<nckx>Using Epiphany (my backup GUI browser ;-) gives: Your browser does not support the playback of this video. Please try using a different browser.
<nckx>What are the odds there's a whitelist Chrome|Firefox|Edge somewhere.
<emacsomancer>I mean, the problem is that for anything that might not work in Firefox/Icecat, it's unlikely to work in Epiphany.
<emacsomancer>Does Epiphany have user-agent spoofing ability?
<nckx>If anyone feels like helping me with their Icecat results: tobias dot gr / mlb is where the illegal content at.
*nckx don't even care.
<nckx>emacsomancer: Not exposed in the UI AFAICT.
<nckx>But I'm very unfamiliar with it. Luckily I hardly ever have to fall back to a different browser.
<emacsomancer>nckx: I get stuck at "just a moment" with a cloudflare ddos screen
<nckx>♥ that modern Web.
<nckx>Oh well.
<emacsomancer>even after unblocking the javascript blocker
<nckx>Thanks for trying.
<nckx>It magically broke, it might magically fix itself.
<nckx>🤷
***vagrantc_ is now known as vagrantc
<tortoise>how does guix package -i work on files that don't evaluate to the package in question? whenever I write my own and do guix package -f file.scm or guix build I need to end a package file with it's name so that the program evaluates to the package instead of just having a definition
<tortoise>That might be unclear I'm not sure. I want to define my own packages that only have the package definitions, but guix package -f only seems to work if I end the file by evaluating the package I need.
<civodul>Hello Guix!
<dadinn>I found an issue with the binary install script... it keep failing with error "Could not obtain list of Guix releases."
<dadinn>this seems to be due to wget on an https:// url
<dadinn>Also, the init system is incorrect... it thinks it is sysv-init... this is because systemctl doesn't work well under chroot
<lprndn->Hello guix
<civodul>dadinn: could you paste the output that you got from the install script?
<civodul>hey lprndn-
<lprndn->civodul: Thanks for yesterday. With your advice, I succeeded what I was trying. It's still hacky but it works!
<lprndn->I'm asking way too late but did someone worked on a lightdm service for guix?
<lprndn->*work
<civodul>i don't think so
<civodul>it would be welcome!
<lprndn->cool thanks! so it should be on the way...
<dadinn>civodul: the https problem is due to not having the cacerts installed on debian... it is a freshly debootstrapped instance
<dadinn>civodul: I think it is fair to ask that I install those beforehand, I already install info, and dirmngr just for guix, so ca-cacert is an additional requirement
<civodul>oh right, i see
<dadinn>civodul: also xz-utils
<dadinn>civodul: but regarding the systemd init config... unfortunately I am not able to run the necessary systemctl commands in a chroot... any workaround for that?
<civodul>dadinn: you can always run the appropriate commands manually as described at https://www.gnu.org/software/guix/manual/en/html_node/Binary-Installation.html
<dadinn>civodul: does the `guix archive` command talk to the guix-daemon?
<dadinn>civodul: because that seems to be the only reason the script needs it to be enabled
<dadinn>civodul: sorry not enabled, but started
<dadinn>civodul: `systemctl enable` is just a symlink AFAIX, so it should be doable even in a chroot :/
<dadinn>*AFAIK
<dadinn>civodul: nevermind I think I should just stick to executing the the install script in a running system
<civodul>ok!
<roptat>hi guix!
<civodul>hey roptat!
<roptat>I think I won't be able to bootstrap scala after all, it's just too hard
<roptat>I wonder if it's acceptable to send a patch series to add a binary scala and sbt built from that scala?
<roptat>it's a bit hard to know what to do especially when the first thing that sbt does is to download itself and another version of scala...
<civodul>sending what you have could at least bootstrap the discussion :-)
<roptat>I see
<civodul>but yeah, it sounds like a really hard problem
<civodul>you had a Scala parser, right?
<roptat>kind of, but I found out that the scala documentation was not complete (and even contradictory), so my parser is not complete and probably wrong too
<civodul>it would be nice not to waste it though
<roptat>sure
<civodul>you should share what you did
<civodul>perhaps there are people in Scala circles that would be interested
<roptat>it's available here: https://framagit.org/tyreunom/scabo
<roptat>I'll work on a plan to merge some of my own packages too, I have a lot of them that I've been just too lazy to send patches for
<roptat>working on a separate collection of packages is nice because it's easy to develop, but it's more work to merge in guix afterwards too...
<civodul>heh
<civodul>ideally it shouldn't be harder to get them into Guix for someone already familiar with the conventions and all
<roptat>it's only harder because I don't pay too much attention to conventions in my own package set
<civodul>ok :-)
<linarcx>Guys, Can anyone help me to dualboot guixsd: https://www.reddit.com/r/GUIX/comments/bcpx25/how_to_dual_boot_guixsd_with_another_osdistros/
<roptat>linarcx, hi :)
<roptat>we don't support auto-detection of other distros, instead you have to either not use grub from Guix System, or declare your other distros in your config file
<roptat>if you're using windows, we don't support that (yet) unfortunately :/
<civodul>linarcx: https://www.gnu.org/software/guix/manual/en/html_node/Bootloader-Configuration.html#index-dual-boot provides hints
*civodul is unsure whether this is "unfortunate"
<abcdw>`guix system reconfigur config.scm` reports that system is 8 days old and I need to do `guix pull`, but I already did it right before reconfigure. What am I doing wrong?(
<roptat>civodul, well it prevents me from booting into my haiku system too
<roptat>abcdw, you probably forgot to update PATH or run `hash guix`
<roptat>(or both)
<civodul>roptat: which means you can contribute more to Guix! :-)
<civodul>seriously though, aren't you able to specify a menu-entry for Haiku?
<roptat>no because I need to write "chainloader +1" somewhere and that's not possible with our grub abstraction
<roptat>so I can boot from the grub console, but not from an entry
<roptat>I'm going to eat, see you
<civodul>i see
<linarcx>roptat: Hi, i have windows7, arch linux, debian, nixos on the same hdd. A question: How to declare other distros in my config files? which config files?
<civodul>roptat: enjoy your meal!
<civodul>we should discuss it with Danny and Mathieu
<civodul>linarcx: for other GNU/Linux distros, see menu entries at https://www.gnu.org/software/guix/manual/en/html_node/Bootloader-Configuration.html#index-dual-boot
<civodul>the problem roptat describes is for OSes other than GNU/Linux
<abcdw>roptat, message appeared because I did sudo -s, hash showed path to user's profile. Thanks for help.
<linarcx>civodul: Yeah, i see it before, but i don't have any idea about it. what is : `linux "/boot/old/vmlinux-2.6.32"`? How can i find read address to another distro kernel addresses?
<linarcx>or this: ` (initrd "/boot/old/initrd")`
<civodul>linarcx: you first need to find the "boot parameters" of that other distro: kernel file name, initrd, kernel arguments, etc.
<civodul>this is what 'menu-entry' specifies
<linarcx>civodul: How can i find these informations?
<civodul>it depends on the distro so i can't really help you much here
<civodul>it could be in /boot/grub/grub.cfg
<linarcx>civodul: Thank you, let me try :)
<abcdw>I have %base-desktop-service and on each boot on first tty I see `New session cNNN of user gdm. Removed session cNNN.` for 5 hundred times. and nothing happens after that. I can login, start sway and so on. But those lines are very annoying. Why they appears?
<roptat>They're from gdm I think
<civodul>abcdw: it's printed by elogind when the gdm user "logs in" (i.e., when GDM is started)
<civodul>but you should see it only once, when gdm starts
<civodul>and then probably also when you log out
<civodul>definitely not 500 times, though
<civodul>roptat: i was thinking that it'd be nice to have fr_FR@tutoiement, as well as a variant that would explicitly use gender-neutral wording
<civodul>are these things discussed by the French translation team?
<civodul>are there guidelines or is it really up to each translator?
<roptat>none that I have seen so far
<roptat>we could ask though :)
<abcdw>civodul: I have pretty minimal setup https://github.com/abcdw/configs/blob/master/guix/qemu.scm and don't know why it happens.
<civodul>abcdw: but when you wrote 500 times, maybe that was actually less than that? :-)
<abcdw>civodul, every time the number in the message increased by one and last number that I saw was c505.
<civodul>abcdw: could you see if you can reproduce it in a VM (with 'guix system vm'), and then send the config and steps to reproduce to bug-guix@gnu.org?
<abcdw>civodul, sure, will try to reproduce.
<roptat>civodul, about gender-neutral/tutoiement, were you talking about guix, the manual or both?
<roptat>how do you select a variant as a user?
<civodul>roptat: i was thinking about translations to French, generally speaking
<civodul>locale names can have variants after the "@" sign
<roptat>that's the "modifier", right? like euro, etc?
<civodul>yes
<roptat>and gettext can load different files depending on the modifier?
<civodul>i think so
<civodul>i'm not quite familiar with the details though
<roptat>civodul, I just sent a mail to the traduc mailing list to ask about that
<civodul>roptat: heh cool, thanks!
<civodul>oh, the [m] folks are back!
<abcdw>civodul, on system vm message appeared only once, probably I will need to reinstall guix from the scratch.
<abcdw>I wrote simple package definition: https://github.com/abcdw/configs/blob/master/guix/alacritty.scm but the build falis with:
<abcdw> https://gist.github.com/abcdw/80ff5253f22bcceb454caab9169f423c Can't understand the reason from stack trace. Any ideas?
<divansantana>Updating guix system, keeps failing with https://paste.debian.net/1077779/
<roptat>abcdw, maybe it's because sources are read-only in your case
<divansantana>perl-io-socket-ssl / git fails to build. Not sure if this has been asked yet, if so apologies.
<roptat>I actually got a similar problem with an OCaml package
<divansantana>I've been trying over a few weeks and same issue.
<roptat>divansantana, can you show us the build log at /var/log/guix/drvs/99/9jzy3bx8aahj2h9fkc0xag6729pzq9-perl-io-socket-ssl-2.038.drv.bz2 ?
<divansantana>roptat: thanks. https://paste.debian.net/1077780/
<divansantana>I'm not sure why it's trying to build from source too. I thought, at what of the times I would try, there would exist a prebuilt package.
<roptat>what does "guix describe" tell you?
<roptat>also, have you authorized substitutes?
<roptat>(check in /etc/guix/acl)
<roptat>I just guix pull'd and got a substitute for perl-io-socket-ssl
<roptat>also make sure .config/guix/current comes first in your path and run "hash guix"
<abcdw>roptat, what does it mean? And what can I do with it?
<roptat>abcdw, in my case, it meant that the install phase installed a file with read-only permission (it didn't change it) and then tried to overwrite it
<roptat>abcdw, see the end of patch 2 here: http://issues.guix.info/issue/35280
<roptat>"add write for user, to prevent a failure in the install phase"
<roptat>the reason for the failure is that in the case of a git checkout, the files are directly in the store, so they are read-only. in the case of a tarball, the tarball is read-only but uncompressed files have write permission
<roptat>in my case, that lead to the install phase failing with the same message as yours, and as you're also using a git checkout, it might be the same reason
***tilpner_ is now known as tilpner
<divansantana>roptat: thanks. Will check that.
<abcdw>roptat, I obtained an archive from github and now it's not failing in that place, thank you.
<roptat>you shouldn't use a github archive, because they are not stable
<abcdw>roptat, ok, I will keep it in mind, but now I just want to build a package somehow)
<roptat>ok :)
<abcdw>I'm gettin "error: no matching package named `euclid` found location searched: registry `https://github.com/rust-lang/crates.io-index`", but actually euclid exists in registry. Can someone explain how cargo-build-system works? I can't understand it from source code yet. https://gist.github.com/abcdw/3f36475acb4f6eaf25f6ba09f8f45b6c
<roptat>sorry, no idea, but one thing is sure: it doesn't have network access when building a package
<abcdw>roptat, hm, that's mean that I should specify those rust packages as inputs somehow
<roptat>I think so
<pkill9>does guix have a /var/log/syslog equivalent?
<pkill9>i've never actually used /var/log/syslog before though
<civodul>/var/log/messages i guess?
<pkill9>that looks like it
*abcdw Imported around 20 crates and still importing :/
<dustyweb>wow
<civodul>matrix.org is DDoS'ing Freenode
<dustyweb>civodul: well not quite I think
<dustyweb>civodul: those were all people who *were* on freenode through the matrix bridge I think?
<dustyweb>but then they disappeared
<dustyweb>because matrix's bridge server (and a bunch of other infrastructure) got attacked
<dustyweb>so now it's coming back online
<dustyweb>afaict.
<galaxie>Huh. I was wondering why there was so many messages.
<dustyweb> https://twitter.com/matrixdotorg/status/1116304867683905537
<civodul>dustyweb: yeah, that's my understanding, i was kidding :-)
<civodul>but it certainly has an impact on Freenode
<dustyweb> https://matrix.org/blog/2019/04/11/security-incident/
<dustyweb>I'm still not sure why there's a *central* bridge server
<dustyweb>why not have each separate federated matrix node do its own bridging
<dustyweb>I find many things about matrix to be very strange
<civodul>yeah
<abcdw>how to do such thing: (define (string->symbol "symbol-name") "symbol-value") ?
<bavier>nckx: yasnippet troubles? commit 6e36c6ed8
<bavier>civodul: there's a typo in guix.texi. Section "Limitations", 'dekstop' should be "desktop".
<civodul>bavier: i don't see that typo in current master, is it still there?
<bavier>civodul: ah, sorry, must have been fixed. I guess I was browsing the pinned version
<civodul>could be
<civodul>i recently removed a few things from that section
<nckx>bavier: I remember that from guix-commits@, it's just been fixed.
<bavier>nckx: ok, great
<jonsger>dustyweb: I think almost all matrix users are on the matrix.org instance, so this is centralized either ways...
<nckx>bavier: Eh. Whoops. Sorry.
<nckx>I launched epiphany last night and it seems to have hijacked URLs in my terminal.
<katco>hi, can irc users see this message?
<nckx>katco: o/
<katco>yay! sorry for any spam i've been causing while i figure out matrix's irc bridge
<katco>dustyweb:
<katco>dustyweb: that's how it's supposed to work. but many of us are using a central matrix server, not our own federated one.
<katco>(yet)
<dustyweb>katco: aha
<katco>i was actually watching the fosdem talk about matrix, and the goal of the matrix team is to actually get rid of matrix.org as federation improves. i don't think they have a desire to admin their own instance -- at least not one this large
<dustyweb>I certainly understand that
<dustyweb>I've come to believe it's an antipattern for people who work on decentralized systems to host a "flagship" node
<dustyweb>that's a whole other job
<dustyweb>and it's healthier for the network for the core people to not be tied up in that every day
<katco>yes, agreed hehe
<jfred>I'm a bit divided on that. On one hand it promotes centralization, but on the other hand when the community is small enough not having a flagship node makes it harder for new people to join
<kmicu>nckx: happy mimeapps configuration.
<nckx>kmicu: xdg-settings set default-web-browser icecat.desktop did the trick, but finding that out took time I would've rather wasted on something else. 😠
<nckx>PSA: You're not helping anyone by making your programme do things like that.
<nckx>I expect that from Chromium & other commercial crapware.
<kmicu>Better do not look inside xdg-open script then xD
<jonsger>diaspora had a similar issue, where the biggest instance wasn't maintained well anymore (old software, slow, etc...)
<katco>i agree with this. i definitely wouldn't look at matrix as favorably as i do had i not been able to try it out on a well maintained instance first
<nckx>Today, on Overdrive: ERROR: In procedure dynamic-link: In procedure dynamic-link: file: "/gnu/store/dd8ld51ilm098kj2apx3wr53y124w8yi-libgcrypt-1.8.3/lib/libgcrypt", message: "file not found"
<nckx>During ‘guix pull’. Which is broken. Which is bad.
<nckx>I'd expect a failed ‘guix pull’ not to break ’guix pull’ (since it hasn't switched guixes yet) so what'm I missing.
<lprndn->Hey. Does anyone know what are 'c1', 'c2' etc. sessions on guix? It
<lprndn->*It's elogind related right?
<nckx>Problem seems to b the missing extension on libgcrypt. /gnu/store/dd8ld51ilm098kj2apx3wr53y124w8yi-libgcrypt-1.8.3/lib/libgcrypt.{la,so,so.20,so.20.2.3} do exist.
<nckx>lprndn-: Yes.
<nckx>You can list them through loginctl, that's all I know.
<lprndn->nckx: ok thanks! I need to go but i'll start with that.
***rekado_ is now known as rekado
<rekado>hey Guix
<rekado>I’m still alive, but I’m not currently following the IRC channel or mailing list
<rekado>still recovering from (planned) surgery; I’m only online some of the time when the pain killers kick in :)
<bavier>rekado: recover quickly! :)
<rekado>trying!
<rekado>had surgery on Saturday; I’m on sick leave for two weeks, but recovery can take three weeks or more.
<nckx>rekado: Get well soon & take it easy :)
<OrangeShark>rekado: get well soon
<nckx>sneek: later tell civodul: Thanks for fixing http-parser.
<sneek>Okay.
<apteryx>rekado: Give it the time it needs! But I hope you get well soon :-)
***maddo_ is now known as maddo
<abcdw>Created a package definition for alacritty terminal, but the build process hangs even with 40G of ram. https://github.com/abcdw/configs/blob/master/guix/alacritty.scm What am I doing wrong?(
<nckx>abcdw: First (and total) guess is a circular reference in your code. Does the *build* hang, or the *computation* of the derivation (♪)?
<nckx>Although I don't see an obvious problematic reference.
<nckx>Does RAM usage actually have anything to do with it?
<efraim>Could it be a problem with the build system?
<abcdw>nckx, I created a swap file and set GC_INITIAL_HEAP_SIZE=40G, without setting this var to big enough value build fails. And htop says that all 16G of ram and 30G of swap used.
<nckx>abcdw: By Guile?
<abcdw>nckx, yes
<nckx>I'm not great at debugging circular refs myself; my usual method is ‘guess and poke until you hit something’.
<nckx>since there's no info to go on.
<abcdw>Can I plot a graph with some guix flags/commands?
<nckx>🤷 Maybe ask in #guile.
*kmicu wishes rekado quick and painless recovery.
<dongcarl>Wondering how I might inherit from gcc but remove this substitute block: http://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/gcc.scm#n235
<dongcarl>(And yes I know it's stupid and against everything Guix stands for but for my particular use-case I need 0 rpaths)
<nckx>dongcarl: You'd replace the pre-configure phase with a copied phase that omits that block.
<nckx>Yes, it sucks, but that's the granularity you have.
*nckx prefers small 'do-one-thing phases over 'pre-foo phases for that reason.
<dongcarl>nckx: Thanks!
<dongcarl>Yeah 'do-one-thing phases ftw
<nckx>Or maybe you could do some lispy code-as-sexp manipulation. ‘Probably.’ Haven't tried.
*nckx is being super helpful here yw.
<dongcarl>Hahaha
<nckx>dongcarl: If you have the time, I'm curious about what you're building. Bitcoin something Guix something?
<dongcarl>nckx: yeah I'd love to explain if it's not too offtopic for this channel
<dongcarl>Basically we have a deterministic build system already
<nckx>If it involves Guix in any way I don't see how it can be.
<dongcarl>It's called Gitian
<dongcarl>But it basically just builds inside an Ubuntu VM with libfaketime and our own set of patches
<dongcarl>I think that the Ubuntu distribution is too large of a trusted binary
<dongcarl>So Guix became a natural choice
<nckx>I agree.
<dongcarl>I hesitated between NixOS and Guix, but Guix seems more focused on bootstrappability
<dongcarl>So when I first started, I thought the easiest way would be to ship `guix pack` tarballs
<dongcarl>But I thought about it some more
<dongcarl>And I knew that no one would use that
<dongcarl>So I'm on a journey to produce the same tarballs as our normal Gitian builds
<dongcarl>Not deterministically the same, just functions the same.
<dongcarl>Same `ldd` output and such
<harding>dongcarl: you should link your WIP PR. :-)
<dongcarl>Hello harding! Good to see you here.
<nckx>Hence the desire to de-guix the rpath.
<dongcarl>I've posted on the mailing list already, but here's my WIP PR: https://github.com/bitcoin/bitcoin/pull/15277
<dongcarl>nckx: Right! Exactly
<dongcarl>Right now I've solved that problem, and the only problem is the interpreter path
<dongcarl>Once I solve the interpreter path, we're golden (at least for x86_64-unknown-linux-gnu)
<nckx>While I'd prefer to see it done ‘properly’ (not that I know how), maybe running patchelf on the result would be an acceptable hack for now?
<nckx>It's been ages since I've used it but it could modify interp paths last I did.
<dongcarl>nckx: Yup! I've already got it working with `patchelf`, so the journey is to do it 'properly'
<nckx>Oh, OK, cool. 👍
<dongcarl>nckx: would you know how to append to inherited configure-flags?
<nckx>Nice to hear peeps doing that in this age of DockeraaBuildSystem.
<nckx>dongcarl: Yes, but off the top of my head.
<nckx>*not
<dongcarl>`DockeraaBuildSystem` nice one I'm gunna use that
<nckx>dongcarl: make-glibc-locales in (gnu packages base) seems to be a pretty minimal example of that. If it fits into your current way of inheriting things.
<dongcarl>nckx: Ah! `substitute-keyword-arguments`!
<dongcarl>Danke schon
<nckx>Graag gedaan.
<tune>does anyone here use xmpp on guix? can't figure out how to enable encryption in dino, can't get logged in at all in gajim
<tune>oh dear I think it was just lagging massively. I'm now logged in several times over
<nckx>eolie on Guix defaults to opening google.com.
<nckx>Is this something that we want?
<dongcarl>Does anyone know how to configure spacemacs to reformat my guile code the way the Guix repo does?
<dongcarl>I think spacemacs' Guile/Scheme layer uses geiser in the back, but not sure if that's used to reformat things
<civodul>dongcarl: Geiser is used to communicate with the Scheme implementation
<sneek>Welcome back civodul, you have 1 message.
<sneek>civodul, nckx says: Thanks for fixing http-parser.
<civodul>yw!
<civodul>dongcarl: but i think Emacs-Guix provides scheme-mode customizations
<civodul>also, the Git repo has a .dir-locals.el file
<civodul> https://www.gnu.org/software/guix/manual/en/html_node/Formatting-Code.html
<dongcarl>civodul: Thank you very much, I need to check that out
<dongcarl>does .dir-locals.el have special meaning?
<apteryx>it's a Emacs builtin way to define variables 'per-directory'
<dongcarl>Understood, thanks!
<dongcarl>It seems that I need to build guix first to use ./etc/indent-code.el.in though?
***paroneayea is now known as dustyweb
<apteryx>correct, that .in is a template used by the build system
<civodul>dongcarl: but since you're using Spacemacs, you probably don't need indent-code.el at all?
<civodul>i suppose Spacemacs honors .dir-locals.el, doesn't it?