<nckx>amz3: I don't see any rudeness or an intention to bury the problem. <nckx>If the status of upstream is ‘dead’, then that's a valid comment IMO. I don't read Ludo's (or Ricardo's) reply as mandating an upstream response, just a best-effort attempt to reach them. <nckx>But then that's just me. <nckx>The fact that neither maintainer replied to your last message isn't malicious; my *personal* impression reading between the ML lines for the last few months is that there's simply too much work to respond to everyone in time. <nckx>I understand that it's unpleasant but I don't think ‘rude’ is fair at all. <nckx>(Looking at the patch: all that is missing, if upstream has no opinion, is a comment saying so and an explanation of what the code does. That would be required for any patch like this: it takes someone who knows the code (you!) to tell clueless readers (me!) what the hell's going on there 😃) <nckx>amz3: IMO that's literally the only blocker, and it's something only you can do. I'd merge it then, if that's any help… <ArneBab>Warning: versions are different between png.h and png.c <quiliro>Dynamicmetaflow: saluton is in Esperanto <quiliro>neutral language for international communications <Dynamicmetaflow>Ah, Esperanto! Yes I've heard about it before. I find it fascinating <quiliro>i made 120 fotos and 120 labels...made it a game to pair them <Dynamicmetaflow>hmm, gracias! I've always wanted to in the past. May look into it again! <quiliro>Dynamicmetaflow: la única manera de aprender es escribiendo o hablando al menos 10 minutos al día <quiliro>no leyendo o escuchando...eso es extra <minall>Dynamicmetaflow, Hello!, Hola, Bonjour! <quiliro>que chiste... hablando inglés entre latinos! <minall>Recuerdan que tengo un sistema aparte, para probar el rendimiento... <minall>quiliro, pasara por saslibre? genial!!!! <minall>Qure bien!! hacia tiempo sin verlo <minall>Dynamicmetaflow, es un lugar de reunion, yo y quilito nos conocemos <quiliro>pero nos vemos por acá casi a diario <minall>Bueno, resulta que revise el log de Xorg en el sistema en el que el rendimiento si es bueno, privativo lamentablemente, asique no podre usarlo bien <minall>Dynamicmetaflow, Wowww massachusets <quiliro>estás cerca de la oficina de Stallman? <minall>quiliro, Y edison? porque no le invitamos? <Dynamicmetaflow>Si, yo he sido voluntario de FSF en el pasado y ha hido a los eventos y libreplanet etc <minall>quiliro, Resulta que tampoco cargo el module openchrome. <minall>ahora esto genera varias preguntas <quiliro>recuerda que el viernes hay la reunión del free software directory a las 11h -utc-5 <quiliro>Dynamicmetaflow puede ir personalmente! <minall>Ahora, porque va mejor aqui a pesar de que no cargo el modulo de igual manera? lueg ode eso procedi a instalarlo, el paquete es xserver-video-xorg-openchrome <minall>Al final no estoy notando tanto rendimiento de diferencia, pero igual, el rendimiento ES bueno <minall>En cambio en guix no, a pesar que estoy corriendo un WM, algo menos pesado <minall>Porque parece que aunque no cargue el modulo, igual tiene mejor rendimiento. <quiliro>Dynamicmetaflow: fuiste al LP de este año? <quiliro>Dynamicmetaflow: has hecho traducciones? <Dynamicmetaflow>eh, no, mas ayudado mandar cartas, ayudado con el video y el audio y siendo moderator, y ayudado unos speaker cuando vienen a Boston mostrandole la cuidad etc <quiliro>Dynamicmetaflow: debes haber conocido gente increíble! <Dynamicmetaflow>oh actualmente avia una vez que si hice una traducion a un grupo que vino al FSF en un libreplanet tiempo atras <minall>Dynamicmetaflow, Tu eres una persona increible! <minall>quiliro, Lo descargare enseguida, casualmente tengo un disco duro para pruebas :D <minall>quiliro, Puedo preguntar porque trisquel exactamente? porque no, por ejemplo, clearos <Dynamicmetaflow>Ahola cuando tengo chamba y estoy en un lugar mejor pudiera ser voluntario y ayudar mas, particurlarmente minoridades y nuestra comunidades <quiliro>minall: ir a boston sería chévere experiencia...por qué no haces méritos? quizá te inviten! <minall>Dynamicmetaflow, Eso es hacer mucho, es increible que puedas ayudar a la causa de esa manera! <minall>quiliro, Meritos te refieres, ayudar a la fsf directamente, o ayudando en el proyecto de guix sserviria? <Dynamicmetaflow>mi sueno es de crear un co-operativo de trabajadores que ayudan los negocios pequenos, pero para la justicia social <quiliro>minall: trisquel tiene drivers libres y puedes ver si dá indicios de solución tu problema de vídeo <quiliro>también puedes probar parabola que también es libre y es basado en arch que ya conoces <minall>quiliro, es que parece que ni en guix, ni en parrot se me cargaron los modulos openchrome, en cambio, en parrotOS, basado en devian pero privativo, me va mejor, mas eestable, y con un desktop environment, el cual es mas pesado <quiliro>minall: creo que puedes conseguir hacer la ingeniería inversa de tu tarjeta de vídeo! <minall>quiliro, Ire por trisquek, aunque tambien estaba pensando en clearos, me encanta la presentacion de su pagina! <minall>quiliro, seria un trabajo interesante!, aunque no estoy seguro si es necesario, creo que el problema esta en otro lado... <minall>Dynamicmetaflow, Intentate trisquel! <quiliro>puedes también preguntar en algún sitio web o sala de chat que traten de openchrome <madage>nckx && quiliro, thanks so much for your help, I've been able to rescue my guix system using guix install image grub <minall>Dynamicmetaflow, no se oprque, pero el package management de guix es tan... disfrutable? <Dynamicmetaflow>He brincado por varios diferentes, knoppix, ubutu, arch, manjaro, antergos, fedora, qubes, etc <quiliro>Dynamicmetaflow: no propongo cambiar a trisquel o a parabola. propongo contrastar para entender como reparar este problema de vídeo <madage>and now I'm doubling down on fs corruption and faulty disk, I'll do a fsck/smartctl <quiliro>minall: otra opción es donde hablen de Xorg <Dynamicmetaflow>bueno quiza te puedo hechar la mano un dia, no soy experto ni nada pero tengo un backup x230 <minall>Basicamente, el sistema va bien en sistema privativo, pero si va igual de bien en un sistema libre, seria ver que esta cargando que guix no <Dynamicmetaflow>I did an initial search of using QubesOS has the base installation and then using Guix as the package manager, I would think it would be interesting combination to say the least. <minall>En cambio, si simplemente no funciona... bueno, sera que hay algo que solo podria correr privativamente <madage>I thought it would be hard but it was just a matter of two grub commands really - cryptomount -a ; configfile $file <quiliro>madage: cool...it'd be nice if you documented it and posted your experience on the mailing list <quiliro>minall: pero dicen que también hay #openchrome <minall>enserio, croe que ya tengo que considerar que hay freenode de todo jaja <minall>Es que, no cargo el modulo, pero por alguna razon va rapoido el sistema... <quiliro>minall: sin embargo quiza no sea problema de openchrome...por eso mejor preguntar primero en xorg <minall>Instalare PureOS, ya que solo contiene software gratis.. <madage>quiliro: for sure, I'll do that! <minall>Si funciona lento, significa que algo solo funciona igual de bien con privativo, osea que tendria que buscar otras opciones <minall>Sale en la lista de distros libres en gnu <quiliro>lo conozco, es software libre no software gratis...es muy diferente <quiliro>podrías venderlo, eres libre de hacerlo <quiliro>minall: apura que en #xorg pueden no estar luego <quiliro>minall: Svetta se manifestó en #xorg, no vas a entrar? <minall>quiliro, en camino, aunque no podre estar mucho.. <minall>quiliro, Aunque la verdad no se que preguntar... porque ya el problema parece no ser de xorg... <quiliro>in order to attend this friday's free software directory meeting at #fsf , run: date --date='TZ="America/New_York" 12:00 this Fri' <quiliro>minall: usaste tu disco bueno o el dañado? <minall>quiliro, no creo que sea necesario, ya que no creo que el problema sea del xorg <quiliro>buena información la que les proporcionaste en #xorg <minall>ya que, por ejemplo, este sistema no cargo el modulo al igual que guix, pero aun asi va bien de rendimiento <quiliro>minall: te digo porque puede ser que linux cargue algo privativo del video antes que el driver <minall>Mhhh, quiza sea eso... y al no poder cargar ese malware en un sistema libre, no va tan bien <minall>(minall ha visto varios videos de stallman) <quiliro>así podremos ver si tienen alguna información para mejorar el rendimiento del núcleo con guix *quiliro cree que el de ted talks es el mejor <quiliro>ya solamente me concentro en las preguntas <minall>aunque tengo muchas, muchas preguntas cuando nos veamos <minall>Y si invitamos a edison? el viernes ya que nos reuniremos <quiliro>él lo resolvería con el driver privativo y punto <quiliro>nosotros queremos solucionar ese problema y no salir del problema <minall>Resolverlo es mas dificil pero, se resuelve, en vez de bueno, salir como usted dice <minall>Jejej intento entender el problema <quiliro>minall: estoy seguro de que es Xorg porque no te dá problemas si no lo inicias <Dynamicmetaflow>Greetings, I have a program I had to build from source, hugo, since I couldn't create a guix package. I want take hugo and put it inside ~/bin and then have it be recognized when I use the terminal. How should I go doing this? I don't want to accidentally screw up Guix paths etc <quiliro>Dynamicmetaflow: you must make a guix package <Dynamicmetaflow>I tried for a few hours and couldn't figure it out, I will try again in the future but I have a website I'm trying to make a website for a friend who works at a non-profit and I sadly can't figure it out in time <Dynamicmetaflow>Yeah, I'm familiar with that blog post and site, haven't been able to get it to work <quiliro>you musto read and do the steps on the tutorial several times....( i have to do the same) <quiliro>and you could report your problem on the mailing list <quiliro>there are more chances to get help there <Dynamicmetaflow>Bueno, I will try again soon, at the moment I would like to get clarity on adding a path, ~/bin so commands are recognized <quiliro>export GUIX_PATH=$GUIX_PATH;"~/new/additional/path" <quiliro>Dynamicmetaflow: test it... i am not sure that is correct...but it will be erased when you log out <quiliro>then you can include it on you .profile <Dynamicmetaflow>ok, bueno, that's what I imagined I had to do, I wanted to confirm that before doing so, gracias quiliro <quiliro>too bad...your program...what is needed to compile it? <Dynamicmetaflow>bueno hugo works, I'm just unsure about adding ~/bin so that it's recognized in the terminal <quiliro>Dynamicmetaflow: creo que no es GUIX_PATH <quiliro>oh! ya no es como era antes... creo que lo hice en trisquel antes <quiliro>pero eso lo debes meter en .bashrc o en .bash_profile <quiliro>no entiendo que tiene que ver emacs con hugo <quiliro>ya lo he escuchado todo de emacs con esto! <quiliro>Dynamicmetaflow: cual es la ventaja de usar hugo si se puede exportar a html directamente desde emacs <quiliro>no es mejor solamente crear el código con emacs, el que hugo crea cuando corre? <Dynamicmetaflow>los themes que tiene hugo, y tambien que es un static site generator <Dynamicmetaflow>uno podria usar lo que esta integado en emacs para hacer algo parecido, me gusta hugo por los themes que ya estan hecho y despues de ahi solo tengo que enadir los contenido y hacer unos cambios <Dynamicmetaflow>The ~/.bash_profile would be used once, at login. The ~/.bashrc script is read every time a shell is started. <quiliro>pero después de terminar mis otras tareas <quiliro>but that is not guix on topic...sory <nothingmuch>when using ./pre-inst-env is giving read access to guixbuild group appropriate when seeing permission denied on nix/script/download? <nothingmuch>(seems a bit ham fisted so i thought i'd check for pebcak first) <fps>hi, i just installed guix from the 1.0.1 installer image into a virtualbox vm and i installed mate (along with awesome, i3 and xfce) <fps>logging into awesome had only mod4-shift-q functional (to exit out of it) <fps>logging into mate just has the login screen frozen <fps>hmm, killed mate-session from the console. now the login screen is broken offering me no choice of DEs anymore.. will reboot the thing <roptat>sneek_: later tell quiliro mi tradukas guix en la francan. Se vi havas demandojn, eble mi povas helpi vin :) <nly`>guix-manual 3.7MiB 3.38GiB/s 00:00 [##################] 100.0% <nly`>wow, is that real? 3.38 GiB/s <str1ngs>I specifically needed shutdown-hook for bookmarks lol <str1ngs>ahh I think that file is to small to get a proper download speed <str1ngs>nly`: user-init-hook is now startup-hook keeping with emacs, but not using elisp namespace <str1ngs>encase you were using user-init-hook <nly`>probably won't need write-bookmarks, they are kept up-to-date automatically <str1ngs>okay that's good I should have read that first <nly`>this hook looks super useful <str1ngs>shutdown-hook is still useful so no biggy <str1ngs>I hope over time I can make the C code much thinner <str1ngs>also I think I can move init into app. and webview into buffer <str1ngs>we probably have to many scheme modules <str1ngs>what do we need for a alpha release do you think? <str1ngs>I really think documentation is kinda lacking and needs updating. <nly`>there is dead code some modules (renamed or replaced procedures, some naming conflicts) next-line in minibuffer collides with next-line from (emacsy emacsy), probably more things collide too <str1ngs>yeah, and some incomplete code, that's commented out that's where next-line and prev-line come into play <nly`>Then the code would make sense to me to write documentation for, because i can't tell what's there for a purpose vs leftover <str1ngs>okay tomorrow I will clean up dead code. keymap, minibuffer come to mind and eval. and then update our current readme to reflect gtk <str1ngs>and i'll draft a RFC for guile-user and guix-patches. if you want to start on a guix patch . I'll prepare for another alpha release. <bluekeys>Is there a reason we don't have node/npm? Is it not free? <str1ngs>bluekeys node does exist. the package name is node <str1ngs>nly`: I'll also update code comments and add FIXME where possible will help with understanding what's going on for documentation <str1ngs>nly`: also all doc strings.. so much work lol <str1ngs>nly`: and I'll merge feature-emacsy into master and cleanup branchs for alpha release <nly`>i suggest this key: 'FIXME: REMOVE' -> no longer useful, remove calls etc <nly`>'FIXME: REPLACE foo' -> find and replace instances with foo <nly`>etc, idk. that way maybe i can help <str1ngs>I generally use FIXME to indicate a hack <dfeng>Hi! I would like to install the source of Emacs. Does anyone know how to procceed? <dfeng>For example, I can't find the definition of `car' using "C-h f". <roptat>dfeng, if you want the sources of emacs, you can run "guix build -S emacs" <dfeng>roptat: thanks. Unfortunately, I need to rebuild Emacs from sources to display C definitions from *Help* <roptat>dfeng, I'm not an emacs user, but you can probably download the sources with "guix build -S emacs", unpack them, enter the source directory, run "guix environment emacs" and them ./configure && make? <roptat>guix environment will create a new shell where every build dependency of emacs is available which should help you compile it *rekado-bad-hotel is on vacation, wanted to hack on wip-texlive, but hotel wifi blocks everything but HTTP. <rekado-bad-hotel>I just found that guix-daemon won't work with torsocks. Is there any other way of running it over Tor or over an SSH socks proxy? *rekado-bad-hotel reads a little of the backlog <rekado-bad-hotel>amz3 nckx is right, a comment about upstream being unresponsive is just fine. <rekado-bad-hotel>amz3 we only want to make sure to keep the number of custom patches in our package collection low. *rekado-bad-hotel --> afk <dfeng>roptat: thanks. I looked at the manual but I am not quite comfortable with a manual installation. <roptat>dfeng, are you talking about installing emacs manually, or guix manually? if the former, you just shouldn't <roptat>why do you want to install it manually? <dfeng>roptat: using ./configure and make <roptat>I don't really understand your issue... can't you run what you built without installing it? <roptat>if you don't find enough help here, you can always try and ask on help-guix@gnu.org <dfeng>roptat: of course, I can build a local version of Emacs even without Guix <dfeng>roptat: ok. I will rather study my issue later after reading the manual. <efraim>how can I check which service is pulling in network manager? <efraim>oh, it's probably network-manager-applet <nothingmuch>still stuck with error: failed to run download program '/home/user/guix/nix/scripts/download': Permission denied <nothingmuch>only 1 search result on the intertubes (shebang is #!/bin/bash, but changing it did not help) <civodul>nothingmuch: you installed it "by hand", right? <nothingmuch>i'm trying to run inside of ./pre-inst-env, to try and fix an error with perl-5.28 as a native dep when building using --target <nothingmuch>also i just realized i'm an idiot and that i've been reading a stacktrace as if it was for a warning, not an error ((gcrypt hash) module missing), i bootstrapped and configured guix in guix environment --pure guix <nothingmuch>config.log claims (gcrypt hash) is available, -lgcrypt and gcrypt.h also found, and the actual error is s/missing/no code for module/ sorry <civodul>hey, how's your stay, rekado-bad-hotel? :-) <civodul>nothingmuch: to run the daemon from your build tree, you should do: sudo -E ./pre-inst-env guix-daemon ... <rekado-bad-hotel>I thought I could at least hack a little, but the wifi here's so bad and all but HTTP ports are blocked... <nothingmuch>good news: error seems to be gone now, bad news: i can't reproduce it anymore <nothingmuch>tried in a new shell, new env created with 1.0.0, as well as 3c2bbbb both work <nly`>I wanted to `guix package -f klisp --system=i686-linux`, but there is no system flag :( <bricewge>How can I append a module to /etc/pam.d/login? I already asked it a few days ago but didn't had any answers. <ArneBab>I found my problem: in /gnu/store/jyp8fmipyc9gjh6l465dzc95yx7whg23-pngcrush-1.8.13.drv there is <ArneBab>("/gnu/store/nr087v1hy3x2byjvvvmbqk6nhrzcxrmg-libpng-1.6.34.drv",["out"]),("/gnu/store/qz49ik2lvbff7xrl5685kvng6iv6qa7x-libpng-1.6.37.drv",["out"]) <ArneBab>so the pngcrush derivation uses two different dependencies for libpng. <roptat>ArneBab, what version of guix are you on? <rekado-bad-hotel>bricewge you'll probably need to specify a value for `pam-services` in your operating system configuration <bricewge>rekado-bad-hotel: The thing is that a pam service for login is already defined in `login-pam-service`. If I do so would it be append or replace the origianl one? <civodul>rekado-bad-hotel: rainy weather is good for mumi (or Cuirass) hacking <civodul>so, i plan to work on the web site migrations <civodul>jonsger: there are two things: one is migrating the guix-hpc web site to bayfront, and another one is migration the guix web site to guix.gnu.org <jonsger>ah so for guix-hpc it's about the hardware where it's running <roptat>civodul, I'd like to work on the website, but I have two questions: should I commit the pdf files to the guix-artwork repo? and how to configure redirects? <civodul>roptat: cool! i have preliminary work for the redirects and all, i'll push it shortly so you can take it from there <civodul>we could have them in /srv/something <civodul>i mean i don't know if it makes sense to have them in VCS <civodul>so i'd lean towards just having them copied to some directory on berlin for now <roptat>looking at it, it seems to be tied to github... <roptat>like, there introduction is "with git lfs, it's easy to work with big files on github"... <civodul>roptat: maybe that's just a way to market it? <roptat>indeed, it seems that "gitlab supports git-lfs" <roptat>so it's not tied to one platform <nly>thanks i didnt notice <ArneBab>roptat: guix (GNU Guix) e5b810f755d5ba37b7eae3a3580f6778aa7dacf2 <roptat>ArneBab, I'm just a few commits more recent than your version and I don't have both libpng versions in pngcrush's derivation <ArneBab>roptat: at the company we use git lfs with gitlab, but it’s only worth using it over regular git if you expect to have really repositories. If your <100 MiB or so, the added complexity isn’t worth the saved storage. <nly>I meant git merge or apply, oh well <ArneBab>roptat: I’m not sure where it got them from, but I have several programs installed which use the old libpng <roptat>that shouldn't affect the derivation though <roptat>note that mine is named /gnu/store/gpw25q14lsjgllvpyg7957f17hzm1d1p-pngcrush-1.8.13.drv <ArneBab>yes, it has a different hash — must have, if it has different dependencies <roptat>yes, let me guix pull to your commit <ArneBab>… could this happen with channels that have different versions? <roptat>maybe, if they redefine a variable named libpng or one of its dependencies <roptat>after my guix pull, I have /gnu/store/gpw25q14lsjgllvpyg7957f17hzm1d1p-pngcrush-1.8.13.drv <roptat>yes, it means that no dependency changed between your version and the version I was using <roptat>and it only refers to /gnu/store/r4gp1qzqqqilrshy9azs667pcwv36rri-libpng-1.6.34.drv <roptat>oh wait! it's because I didn't use grafts <roptat>now that I look at the code I see that libpng-1.6.37 is grafted on libpng-1.6.34 <roptat>so the code in pngcrush probably recorded the version that was used to build it, but after the graft, it's using 1.6.37 instead <civodul>roptat: i've pushed to guix-maintenance.git, notably a new static-web-site-service-type <ArneBab>wow, jut checked, it shows (define-public libpng <ArneBab>so it now finds the new header but still uses the old png.c <ArneBab>no, it still has the old header, but has the new png.c <efraim>So perhaps the two PNG versions aren't as similar as they could be <civodul>rekado_: could you have hpc.guix.info point to bayfront? <ArneBab>efraim: or pngcrush is just too pedantic? <ArneBab>Well, maybe not. I now installed with guix --no-grafts pngcrush, and I get <ArneBab>While measuring IDATs in <image> pngcrush caught libpng error: <civodul>roptat: i've just sent an update about the guix.gnu.org migration to guix-sysadmin <mbakke>rekado: Do you remember what the 'pywrap' stuff in libselinux does? <mbakke>I wonder if it can go to a separate output: libselinux adds almost 100 MiB to the glib closure (because of Python). ***atw` is now known as atw
<civodul>mbakke: do we really want libselinux at all? <mbakke>civodul: libselinux appears to be a mandatory dependency of newer glib <mbakke>Adding a python output was a two-line change that did the job, so I went with that. <civodul>SELinux is not the kind of security tool i'd like to see on Guix System <mbakke>Interesting... What kind of security tool would you like to see? :-) <civodul>it depends on the context, but if it were me, i'd rather provide tools that make it easy to enforce the "principle of least privelege" <civodul>like the container wrappers we've discussed, make-forkexec-constructor/container, etc. <civodul>it's more prospective and probably not one-size-fits-all though... <mbakke>civodul: I agree, I really dislike that every program has unlimited access to my home directory (and system in general). <mbakke>I have played with the idea of a QubesOS-like window manager, where each workspace is a Guix container with different access controls. <civodul>we could even make it finer-grain :-) <civodul>and i hear we should default to Wayland too, because on X11 any client can fiddle with any other client <civodul>mbakke: BTW, i was traveling and i just realized the July 9th deadline was behind us :-) <civodul>so i'm really eager to get core-updates merged, so i'll do some testing, and i hope others will too! <mbakke>civodul: I'm currently writing a summary email and call for testing :-) <mbakke>Though we should wait for cbaines gobject-introspection fix before the Big Rebuild. <Dynamicmetaflow>hi mbakke, out of curiosity have you executed your idea? Yesterday I had a similar idea of using QubesOS and the Guix package manager in the template VM's. I actually am workin on getting my backup machine (X230) running QubesOS to tinker with the idea. Been trying to find other posts related to this but not much is showing up <emacsen>What is the guix approach when packaging to when a source goes away? like if you downloaded it from https://foo.org/.... and foo.org goes away? <mbakke>Dynamicmetaflow: Cool! I have not explored it further, the implementation is still up for grabs ;-) <Dynamicmetaflow>lol, I was and still am excited that there was someone else thinking about this after hours of searching online <mbakke>emacsen: Guix will try a few content-addressed mirrors, such as the CI systems of Nix and Guix, as well as the recent "software heritage" effort. <Dynamicmetaflow>The roadbump I'm hitting now with the x230 is related to the sys-net and sys-firewall that Qubes uses, prevents other vm's from loading. Hopefully once I figure that out I can experiment with Guix in and see what we can do <emacsen>mbakke, so there's no place where guix caches old versions? <mbakke>emacsen: Caching old versions of software is really the job of software heritage (which Guix neatly integrates with). <mbakke>The CI system is rarely garbage collected, so in practice old versions are cached there for a long time too. <civodul>"guix upgrade: warning: at least 10,345.0 MB needed but only 5,748.9 MB available in /gnu/store" *civodul tries an upgrade to core-updates <civodul>it would be around 20G with gzip i guess <civodul>there are many tarballs and checkouts there, so it's probably not representative ***amiloradovsky1 is now known as amiloradovsky
<lispmacs>hi, having been trying out Guix for the first time over the last few days. my understanding is that Guix uses a rolling release model, but I am wanting a pretty stable production environment, where interfaces and such don't change often. Is there anything I can do toward that end without ending up with an insecure system? <lispmacs>And on a related note, does the package database have something like version release tags, so I focus on only updating to version points believed to be well tested? <minall>lispmacs: Such a nice question, I'll give you what I would do, but I'll make that question too <minall>I'll install everything I'll need and use, and then update all the system, and since I already updated it, it has the latest security updates, so I'll remain without updating unless I encounter a problem <minall>guix gives you the option to do that!, but that's my opinion of course, I've never heard of release tags, maybe someone has? <lispmacs>my thought was that I could just only update packages that become vulnerable, but I'm not sure where to easily get notified of just vulnerabilities. I looked on the site for something like Debian Security Announce but didn't see it <minall>Yep, the reason that debian is stable, is because they use not new packages, but very tested and patched ones, I had that question too, How example, If I can replicate a debian like stable system on guix, and, If needed, updae something <minall>Actually, I'm making that question too... teach me guix-irc! <Fzer0>Hello everyone, how can I make Fish the global shell? <roptat>we have a rolling-release model, so security fixes come in with every update we make to packages <minall>roptat: But is there a version of any package that would be considered more stable? <roptat>there's really not much more testing being done on releases compared to just any other commit in master, so the 1.0.1 commit for instance is a bit arbitrary <roptat>there isn't, mostly because we don't have enough power (in terms of people and build machines) to maintain both models <roptat>especially for system configuration, it's mostly all or nothing: you either upgrade your whole system to a newer package set, or don't upgrade it at all <roptat>but that's not a problem at all, because you can always roll-back in case of troubles <mbakke>lispmacs: You can use `guix lint -c cve` to check for vulnerabilities in packages, but there's currently no way to use it on just a profile. <erudition>sure it is. what if you want a new version of package A, where a vulnerability is patched, but you want to keep your old version of package B, because the latest one is broken to you? <minall>Well, that's right, guix would be a free option, when you can test the latest technologies o packages in this case as fast as possible <minall>But it isn't unstable as archlinux for example <lispmacs>practical question, then: what is easiest way to view master branch of package database, so I can get a view of what is coming down the pipe? <minall>It let's you fall back so... I think the system is actually pretty stable!!! <roptat>for user profiles, you can always keep some packages to what they were, but for the base system, it's not easy (you can use inferiors though) <erudition>quick recovery is nice, but not to be conflated with stability <roptat>so you could have your system on a specific commit of guix, and use some packages from a newer guix, as an inferior package <minall>So it's not stable, but quick recovered? <minall>I assume that if one user is infected by something, it won't do anything to the system, right? <erudition>minall: Yeah, instant rollback makes instability tolerable haha <minall>Actually it is rare for a rolling release to have problems, for example, archlinux is not TOTALLY unstable, it just isn't stable <minall>But things may happend, the difference is that archlinux won't allow you to roll back <erudition>But as far as I know you can still downgrade a package in arch <minall>So if you have important information, you won't be able to get to the system on arch <lispmacs>what is easiest way to view master branch of package database, so I can get a view of what is coming down the pipe? <minall>erudition: Yes, it's possible, but it may break something, a program for example, also, just in guix you can test a whole new system safely! <minall>Is there a way of testing a config.scm after build it? like a Virtual system or something <erudition>That's one thing I'm struggling with in guix - I thought I'd have more freedom and flexibility with packages across the board, but things like upgrading and downgrading individual packages seem to be designed against <erudition>minall: I guess it could break something if it's a system package, but most user packages shouldn't matter. Like if I want an older version of blender, for instance. On Debian-based I could just downgrade it. <minall>erudition: as easier as downgrading in guix? but you're right, that's an option! <minall>But I like more guix since I can test the latest of everything, but being 'stable' <minall>On debian some packages are not upgraded, BUT, as you said, you could install a debian based distro, or go to debian unstable <sneek_>quiliro, roptat says: mi tradukas guix en la francan. Se vi havas demandojn, eble mi povas helpi vin :) <quiliro>roptat: Ĉu vi uzis la Emacs-an pomode-on por traduki la Guix-on? <quiliro>roptat: Dankon por oferi vian helpon. <quiliro>minall: Ĉu vi skribis al la liston de poŝto? <minall>quiliro: I haven't fixed anything yet, the module is now loaded, BUT, the performance is not... I'll test some things after going on my house... Then I am planning on updating the mailing list, so I can make a problem, but also the fix <nly`>> what is easiest way to view master branch of package database, so I can get a view of what is coming down the pipe? <nly`>lispmacs: the comprehensive way would be to keep a local guix checkout <roptat>kaj offlate-on, sed ĝi nur ne estas tre uzebla <nly`>and point your guix channel to the local checkout <roptat>minall, you can build a system with "guix system build" (without executing it), you can create a vm with "guix system vm" or "guix system vm-image" (either read-only or read-write image) and they are all going to be the same system <roptat>now usually, I don't install too many things in the system config, and use my user profile to install everything I need <roptat>the user profile is much more flexible because it doesn't have to be built in a declarative way <roptat>it's easier to lose track of things when it's not declarative though... <roptat>and as I said, you can use inferiors to get packages from another guix commit (either past or future compared to your current guix installation) ***nly` is now known as nly
<Formbi>actually why are things in Guix built with GCC 5.5.0? <mbakke>Because upgrading to GCC 7 was really difficult. <mbakke>Formbi: ^ but the 'core-updates' branch has GCC 7 now. <mbakke>I hope the switch to GCC 9 goes smoother :-) <ArneBab>erudition: you can actually get the local git clone of the package database, change the version there, and then install from your local checkout. <Formbi>actually what were the problems with GCC 7? <Formbi>(not that I could resolve them, just curious) <erudition>ArneBab: I'm not saying it's impossible, but even that solution sounds hacky compared to a single command to install a desired version of a package. <lispmacs>what is guix command if I want to see what commit of guix repo I last pulled? <ArneBab>I’m used to Gentoo where I can just get a testing version with a single command, and that indeed feels cleaner. With additional people working on packaging, there could simply be multiple versions of packages (for some packages there already are), but until then I’m happy that my system won’t break due to a mid-update power-outage. <rvgn>civodul It would be better to migrate guix-hpc to guix.gnu.org as hpc.guix.gnu.org? <Formbi>ArneBab: you can use package transformation options <Formbi>ex. «guix package -i foo --with-source=foo@bar=./foo» <ArneBab>Formbi: wow, I did not know that. It should be a really easy way to test whether a new version requires more than updating the version. <Formbi>yeah, it's one of my favorite features of Guix <ArneBab>Do we have a cheat-sheet for contributing commits I created to Guix without having to know anything (I forgot most, but I have some clean commits I’d like to share). <nckx>madage: I'm very happy to hear that. Have you found any clues as to what went wrong? <ArneBab>With the many tricks I see here that each are huge timesavers: Is there a Guix packaging best-practice tutorial? <quiliro>roptat: Kio estas via preczo por traduki? <mbakke>Formbi: The troubles were mostly related to how GCC 6 and later handles "system headers". The fixes were rather trivial, but the hard part was figuring out why things were failing. <madage>nckx: not yet, I'm planning to do a disk/fs checkup and see if I can find any logs on the previous generation, but that will have to wait a bit. I'll post here if I find the culprit. <quiliro>roptat: Pardonu min, mi petas: Kio estas via procezo por traduki? <roptat>quiliro, ĉu vi registriĝis en la tradukan projekton? <quiliro>roptat: mi ne memoras. mi kontrolos. <nckx>quiliro: manlibron de Guix aŭ Guix per si mem? <roptat>mi elŝutas la po-dosierojn per offlate-o, sed vi povas ankaŭ elŝuti ilin el la retejo <nckx>(Tamen ambaŭ uzas la TP.) <roptat>poste, vi devas traduki ilin. Mi pensas ke vi povas uzi la po-mode-on aŭ poedit-on <roptat>kiam vi volas alŝuti la dosierojn al la projekto, vi devas uzi la roboton <pkill9>would guix be able to be run on the pinebook pro? <roptat>maybe? you should check if we have a pinebook-pro-u-boot or something <roptat>and whether there is a particular target to build a proper u-boot for that board <roptat>vi devas elŝuti la dosierojn en via lingvo (esperanto aŭ hispana?) <roptat>se ili ne ekzistas, vi devas elŝuti la la pot-a dosieron <roptat>"The current template for this domain is guix-1.0.1-pre1.pot." <quiliro>roptat: oni ne uzas VCS-on kiel git aŭ aliaj <roptat>vi devas sendi retpoŝtaĵon al la roboto. kiam vi faras tion, mi aŭ ludo denove elŝutas vian dosieron kaj enmetas ĝin en guix <roptat>civodul, I actually don't have access to berlin where the website is hosted right now <roptat>should I create a patch for berlin.scm and send it to the mailing list? <civodul>roptat: you can push the patch for berlin.scm directly <roptat>I'm not sure how to test it though... <civodul>you can roughly test in a VM, but the thing that's boring is that you have to comment out networking and mapped devices at least <civodul>then that won't allow to test the whole setup anyway, because you don't have certs and so on <civodul>so that part of the test will happen on the actual machine i guess :-) <civodul>also, the VM allows you to check that the generated nginx config file is good <civodul>so a lot is scheduled but the binaries are presumably already available <nckx>It's just happening a lot slower than I'd expect, but if you say so I am happy 🙂 <quiliro>nckx: why do i get squares with numbers at the end of your last two messages? <nckx>If you'd find out how to teach emacs how to properly fall back to different fonts, I'd be all ears. <nckx>-nw is just a hack to tell emacs ‘don't try to render fonts, you suck at it, let my terminal render fonts for you, it can’. <nckx>quiliro: I've read that, but was unable to write a working config that falls back from Hack to Google Noto only for 'mojies. <nckx>quiliro: ‘All ears’ = ‘very interested’. <quiliro>oh! i thought you were telling me how to do it <nckx>I wish I knew 🙂 (← smiley). Dankon! *nckx now really → 😴 (a sleeping face saying ‘zzz’) <nckx>(Thanks for not making me speak Esperanto while exhausted.) <null_radix[m]>is there anywhere that list all of the services that are avaible to use? <null_radix[m]>i guess i was hoping to find something that was automated though... like the package lists <quiliro>nckx: "I *think* this package will try to configure the best fonts for various character ranges, including emoji, and I *think* it will work on Linux [sic]." <Formbi>i have Iosevka in Emacs and in terminal, but emojis work only in terminal 