<alextee[m]>anyone know what guile module "substitute*" is in? <nckx>Temporary users would be a nice addition to Guix, pkill9 (minus the IMO odd special-default-guix-thing 🙂). <nckx>pkill9: How are you [thinking of] implementing it? <pkill9>nckx: my particular idea, i'm not, im just guix-pulling now :P <vagrantc>hrm. core-updates didn't bother to build the mesa update ... maybe "not change on x86_64, nothing to build" sorts of problems <pkill9>nckx: you can already add temporary users with 'useradd' <pkill9>on next reboot i don't think the user will be there <nckx>Sure, that's more ‘exploiting’ Guix than ‘leveraging’ but it totally works. <pkill9>since guix creates them each time on boot? maybe that's wrong <janneke>civodul: tested pushed that mingw fix to master <joshuaBPMan>Hey guix, I'm not trying to be a buzz kill, but I'm been having a really hard time using any web browser that guix provides. I'm on a T400 with 8GB. (recently librebooted yeah!). I can use a web browser reliable for about 5 or 10 minutes at a time. Then I have to close the browser and re-open it. Essentially after a few minutes of use, the stock web browsers, start to move very jerkily. If I keep using it, the whole computer <joshuaBPMan>If I am using the stock browsers on a simple site (like debian) or wikipedia, I can use it for hours. <joshuaBPMan>But when I do my online banking I have to use ungoogled chromium, because the bank I use won't accept icecat, and icecat would probably freeze if I try. <joshuaBPMan>Also ungoogled chromium also freezez after 3 minutes of use on any banking site I use: paypal, chase, etc. <joshuaBPMan>anyway, long story short, I am going to have to dual boot. Debian will probably be the other distro that I use, and I will only use it for banking. Everything else icecat works most of the time. <pkill9>joshuaBPMan: so chromium/icecat don't freeze on debian, but do on guix system? <alextee[m]>(guild build utils) is the most awesome guile module ever <grumbel>How do I enter a 'pure' profile? Doing '. "$GUIX_PROFILE"/etc/profile' leaves me with a poluted environment. 'guix environment --pure' doesn't have an option that takes profiles. guix environment --manifest=$GUIX_PROFILE/manifest doesn't work because manifest is an s-expr, not scheme code <nckx>grumbel: Something like env -i "$(which bash)" --init-file <(echo ". '$GUIX_PROFILE/etc/profile'") ? <alextee[m]>joshuaBPMan: it has a ton of useful procedures like invoke, mkdir-p, dopy-recursively etc <joshuaBPMan>Blackbeard: I suppose that I could try using gnome (which probably is probably better tested), but I really wanted to get on the wayland train. <leoprikler>I tried getting on that train several times, but it travels slowly through GNOME land. <Blackbeard>joshuaBPMan: that honestly seems more of a sway problem :/ I've had problems with sway several times and I always give up <vagrantc>mbakke: so, what i'm discovering is i'm just hitting enter with <vagrantc>as i don't really know meaningfully what i should do different <vagrantc>at which point, it's essentially like silentolddefconfig only with me having to wear out my keyboard :) <raghavgururajan>How can I use antlr3-3.4? The variable is not declared public, in java.scm. <brendyyn>It's just antlr3 isnt' it? look at the (name ... ) <nckx>And there is no antlr3-3.4 nor antlr3@3.4 anyway. <nckx>If your package insists on 3.4, you'll have to create a new (inherited) antlr3-3.4 package first. <nckx>raghavgururajan: No, it's not. Your Guix must be out of date. <nckx>Ah. And that's not an option? <nckx>raghavgururajan: What do you mean by ‘use’, though? <nckx>‘Install into a profile’, ‘use as a dependency in another package’, …? <nckx>Then you must simply make antlr3-3.3 public. <nckx>And include that as part of your patch. <nckx>Note that we often first try to ‘convince’ packages to build against our package version(s) by relaxing overly tight dependencies (e.g. substitute* "==" → ">="). Whether that's safe depends on the package and the culture of the language. I'm not familiar enough with Java. <raghavgururajan>I am gonna try 3.3. If that doesn't work, I have to inherit and create 3.4. <nckx>raghavgururajan: Also try just plain ‘antlr3’, which is version 3.5. <raghavgururajan>nckx I did. It appears 3.5 has it's own bugs. It doesn't even compile grammars provided by the package. <Veera>directly using url by lint check says all the source URIS unreachable and permanent redirect <nckx>brendyyn: To? I think all lists have been fixed. <nckx>To their usual blazing speed, but mail should arrive eventually. <brendyyn>nckx, the emails i sent yesterday haven't appeared on guix-patches. do i have to send them again <nckx>brendyyn: Oh, that's different. <nckx>The lists have been fixed, but the old mails have to be dug out of a 2.3 GiB archived file by a man named Bob and re-sent to the list manually. <Veera>though guix download or as part of guix build it downloads and works <nckx>So, I'd wait, unless this is really blocking your mojo. <brendyyn>i guess i no i wont cause them more trouble by sending duplicate emails <nckx>That depends on the mails. <nckx>Veera: That's expected, since that /download URI is just a HTML page that uses JavaScript to point you to the real URI (nice…). What I do is click the ‘Problems downloading?’ button before the page redirects to a helpful survey and copy the ‘direct link’. <Veera>nckx: that links works fine with wget or guix download <Veera>nckx: only lint is complaining <Veera>nckx: one is permanent redirect which happens with wget and guix download also <nckx>Then turn it into a mirror URI: mirror://sourceforge/xplanet/xplanet/1.3.1/xplanet-1.3.1.tar.gz <Veera>nckx: but could not make out for what or why it says all the source URIS unreachable <Veera>nckx: tried some. ok i try this <nckx>Doesn't matter, as I said, the redirect is unacceptable anyway. <Veera>nckx: not working 404 not found <Veera>nckx: some existing pkgs are using without mirror; shall I also do so <nckx>Please try that exact URI. <nckx>Veera: no, mirror:// works. <nckx>Please share the command that caused the 404 status and the complete output. <Veera>nckx: yes it is working. thanks <Veera>nckx: wrongly missed one directory <apteryx>raghavgururajan: I'm a bit late but going to check your Linphone patches now! <nckx>I got disconnected without notice. Veera: glad to hear it 🙂 <nckx>raghavgururajan: Making antlr3 public would be part of the same patch that adds the package, it doesn't really make sense as its own patch/bug (without explanation). Which package are you adding? Then I'll change the title of #40285 and you can send the ‘gnu: Add…’ patch to it as well. <nckx>Which package takes antlr3 as input? <apteryx>today I did a quick Btrfs RAID1 test (yes, Guix System can boot off Btrfs RAID), and noticed no performance improvement over a single drive. Somewhat surprised that it can't make better use of the two disks for reads. <nckx>And you'll send that to the same thread you posted in your previous message? <nckx>Then I'll merge the two. <nckx>apteryx: Haha no that is unimplemented /*TODO*/ my friend. <nckx>IIRC btrfs still balances reads per PID. <nckx>Yes, that ‘PID’, not some magical sane definition. <apteryx>nckx: what do you mean unimplemented? In Btrfs? <nckx>So if you launch 2 processes, and they happen to have both odd & even PIDs, they'll read from different drives. <nckx>If both odd or both even, not. <nckx>This is the Facebook Enterprise File System. <raghavgururajan>nckx I have a doubt. My belle-sip's package require a package called 'tunnel', for tunnel support. The source doesn't mention which tunnel software. How do I find the right package? It appears there is no software with the name just 'tunnel'. <apteryx>nckx: well at least it's not coded in PHP <raghavgururajan>There is stunnel and ptunnel. But both doesn't provide 'tunnel.c' and 'tunnel.h' files. <nckx>I cloned the repository and… hm. Wow. That is obtuse indeed. <nckx>configure.ac does say it's optional though; does --disable-tunnel work until you find out more? <nckx>I'll keep looking too, but it's not like the source is hiding any magic. It's literally just ‘tunnel’ everywhere 😒 <nckx>Are you using the CMake build system? <raghavgururajan>So they have it within the source and also look for outside package? weird. <nckx>OK, I was looking at #include <tunnel/client.hh> <nckx>Then I don't understand why the build system doesn't find it. <Veera>what is "autogen" ,autogen is for? <nckx>But where is the ‘Tunnel’ CMake stuff? find_package(Tunnel) needs… something. <apteryx>raghavgururajan: do you have the other patches somewhere I can grab? <nckx>raghavgururajan: No, it just moves it inside the if(ENABLE_TUNNEL) above…? <nckx>I'm pulling out of this mess before I get sucked in 😉 Bed time. <apteryx>raghavgururajan: do you know how many dependencies we're missing before getting to linphone itself? <raghavgururajan>apteryx In the NextCloud link, the patched for deps are placed in /others <raghavgururajan>apteryx Oh you were internal deps of linphone? Those are part of linphone project. They are numbered in my link. <apteryx>so these are linphone core deps. And under deps/ are indirect dependencies? *brendyyn just got some emails <brendyyn>id send him a thank you email but he's probably really not loving emails at the moment. <apteryx>raghavgururajan: by the way, to generate the patch you can use `git format-patch`, it gives them prefixes that make it easy to 'git am' them later <apteryx>weird, the error CU_trace_handler. Should work with ENABLE_BASIC, which you did, it seems. <apteryx>raghavgururajan: I think bctoolbox can have bcunit as an inputs instead of propagated-inputs. Same for belr's bctoolbox. <apteryx>also, the static libraries maybe are not really needed to be built; if built we often move them to an explicit output 'static', to reduce the size of the installed package. <raghavgururajan>apteryx I gonna leave internal dependencies as propagated for now. I will be revising this later. <raghavgururajan>apteryx Yeah, I have planned to move static libs to separate output. :-) <apteryx>about the bug, it's nice to add a comment in the package refering to the upstream issue (URL). <apteryx>I'm curious about all these test support files, which are built, but not explained how to run? <apteryx>Probably something to do with bcunit <apteryx>I tihnk to run belcard tester one must do: ./tester/belcard-tester or similar (will try) <apteryx>we can just run bin/belcard_tester after install :-) <apteryx>raghavgururajan: other little comment: our convention for commit subjects are "gnu: Add belcard." rather than "gnu: packages: linphone: Add belcard" for example. <apteryx>I've checked your package definitions so far and they all built OK. The comments I'll have about those for review are minor. Seems you're almost done with linphone-sdk :-) It's shaping up nicely. Keep up the good work. *apteryx heads off to bed ***Veera__ is now known as Veera
<Veera>sent to patch submissions; one for mtpaint #40281 and another for libqalculate & qalculate-gtk #40320 <Veera>have to send another one fro xplanet <bavier`>I was also wondering whether packaging some images and data files would be useful <Blackbeard>anyone wants to help me give a last-minute check to my GSoC proposal before I submit it <Blackbeard>guix-vits: hopefully I will be selected, but I'll start from tomorrow, I would like to help even if I don't get accepted <brendyyn>Blackbeard, i asked a systemd related question on my Piper submission on guix-patches. perhaps tomorrow you could comment. good night <Blackbeard>brendyyn: ohh that's nice, I need to learn a lot so it is a good start <brendyyn>Blackbeard, my goal for guix is to make 1-click configurations for services. with my submission a user just wants to configure there mouse. if they put in their sudo password, we should be able to insert the appropriate herd service <brendyyn>but then i noticed that herd doesnt seem to handle dbus services directly , although i havent researched much <brendyyn>The whole reason I started working on guix was to create the best end user It Just Works distro <brendyyn>not some Emacs-like only for geeks thing <brendyyn>since operating-system definitions are composable, and services too, it should be possible to have a Default operating-system defined by us in the guix git repo, but the a GUI configuration manager could make tweaks to that <brendyyn>like "add a tor service", "add support for configuring gaming mice" <brendyyn>Blackbeard, no not yet, but i think first we need to create the abstractions that the GUI would use <guix-vits>i need to check why NetworkManager does not start <brendyyn>Blackbeard, take this service: (simple-service 'ratbagd dbus-root-service-type (list libratbag)) <brendyyn>this service doesn't appear in herd status, so you can't herd start/stop it <brendyyn>im still working on another project at the moment, improving my programming skills <guix-vits>it's works, reboot solved. seem that now i can discard ./dhclient_restart.sh <brendyyn>Blackbeard, I work at the hospital, but I don't have many shifts. I'm working tomorrow. <Blackbeard>brendyyn: you work at a hospital? What do you do? I hope you are safe. <brendyyn>I'm a hospital aide. mostly cleaning and restocking. <brendyyn>I clean infected rooms, but not COVID-19 since you need some extra training and stuff for that. ***rekado_ is now known as rekado
<Veera>Help needed: see paste.debian.net/1137339 xplanet lint fails <guix-vits>Veera: probably you need to include a module that provides "definition" of `search-patches` variable. <guix-vits>Veera: gnu/packages.scm define it on L99. It's included in astronomy.scm... <nicecockbro>Hi Guix. What is the analogue to "systemctl suspend" in Guix? Or in other words how do K <brendyyn>nicecockbro, can you please change your nick so something more appropriate <guix-vits>Veera: do you use it as (patches (search-patches "a" "b" "c")) ? <brendyyn>looks like spammers or bots, something weird going on <Blackbeard>@freenode_nicecockbro:matrix.org: loginctl suspend <Blackbeard>@freenode_somethingmorepol:matrix.org: loginctl reboot <guix-vits>Blackbeard: i was using `echo mem > /sys/power/state`. awkward. <Veera>guix-vits: do you use it as (patches (search-patches "a" "b" "c")) ;;; yes <brendyyn>Veera, it looks like the artifact of a syntax error. can you build your package? <Veera>guix-vits: when I keep it separate xplanet.scm it's fine <Veera>guix-vits: when I include it in astronomy.scm then... <guix-vits>Blackbeard: cat /sys/power/state # freeze mem disk <guix-vits>Blackbeard: maybe modern Puppy Linux has loginctl (or i just overlooked it befora just as threre) <guix-vits>Veera, brendyyn: in astronomy.scm there is (gnu packages) in use-modules, but search-patches isn't used. <guix-vits>Veera: and you use it inside of `(source (origin`? <Veera>guix-vits: i am using it inside (source (origin <Veera>guix-vits: as I said it is working fine in separate xplanet.scm <Veera>guix-vits: But when I integrate into astronomy.scm it is erroring <brendyyn>guix-vits, thats code that runs inside the builder <Veera>guix-vits: thanks. gnu packages module is not there <Veera>guix-vits: why they have not pu use-module (gnu packages) in astronomy.scm <Veera>guix-vits: why they have not put use-module (gnu packages) in astronomy.scm <brendyyn>Veera, modules are included when they are needed. if you look at (gnu packages), which is the gnu/packages.scm file, you will see the #:export list of procedures. it just so happend that none of those were needed by anything else in astronomy.scm <brendyyn>so it's up to you to add i as a part of your packaging effort. <guix-vits>Veera: oh, really. i was though that it included (misread "guix packages" as "gnu packages") <Veera>guix-vits: there is guix packages but no gnu packages <bricewge>guix-vits: I did the exact same error as you. <Veera>guix-vits: I thought gnu packages is neccessary before using anyother gnu package <guix-vits>Veera: no: it's gnu/packages.scm module. Probably some modules not depend on it. <Veera>guix-vits: okay. that explains ***apteryx_ is now known as apteryx
<brendyyn>(base16-string->bytevector "a") ;; guix/base16.scm:74:31: In procedure bv-u8-set!: Value out of range: 0 <brendyyn>it makes sense for guix's purpose since one wants a hash of a certain length, so we'd want the prepended 0 <guix-vits>as example for nftables-configuration (as fro iptables)? <rekado>raghavgururajan: you need to escape the " <rekado>this is unrelated to substitute* — it’s what you need to do for any string <raghavgururajan>Apparently , There is something wrong with "#include "liblinphone_gitversion.h" <rekado>so if you want to use " inside of the string you need to escape it with \ <rekado>"#include \"liblinphone_gitversion.h\"" <rekado>otherwise Guile would think that you have two strings and a variable: "#include ", the empty string "", and in between the variable liblinphone_gitversion.h, which is undefined. <rekado>substitute* takes a number of clauses that consist of a match expression and a replacement <rekado>here the replacement is the empty string "" <rekado>guix-vits: it would be good to add an example to the documentation of nftables-configuration. Maybe something a little shorter than what you linked to. <rekado>guix-vits: would you like to submit a patch to guix-patches@gnu.org? <guix-vits>rekado: yes; i'll try to find something shorther. <kmicu>FYI “[Libre-kernel 5.6] introduced 3 new drivers that request and load blobs: AMD Trusted Execution Environment, ATH11K WiFi, and Mediatek SCP remoteproc. The requests for those are inhibited and silenced in our release, and so are those for new blobs in nouveau, AMDGPU and AMD PSP.” <civodul>uh it was 20°C two days ago and now there's light snow out there <brendyyn>im confused about amdgpu, some of it is free isn't it? <kmicu>Tell me about it. My plants are not happy at all. <kmicu>brendyyn: drivers part. Firmware is blobby mcblobface. <brendyyn>does linux automatically install firmware? <kmicu>brendyyn: yes, if it can find it. Kernel Libre‑linux removes messages about requests to not steer folks into Blobs World. <brendyyn>oh ok. I thought firmware was something that was embeded on a device like a wifi card or graphics card. not something that was run by the operating system? <kmicu>brendyyn: kernel loads firmware into a device. That way drivers in kernel stay GPL and firmware stays blobed. Kernel Libre‑linux obeys FSDG rules which are more strict. <Blackbeard>civodul: the snow is terrible news :( I hope you are safe <civodul>yup, i'm fine, it's just surprising :-) <rekado>Blackbeard: it’s a cool project. The shepherd deserves some concentrated hacking. <Blackbeard>rekado: I can tell, I tried searching for bug reports and I didn't see many on issues.guix.gnu.org <Veera>guix-vits: saw your patch submission for neverball and neverput game in mail <Veera>submitted Xplanet patch for review and addition <guix-vits>Veera: cool (i was guided by many people there) *raghavgururajan have successfully built Linphone \o/ <mbakke>vagrantc: the target is 'oldconfig', not 'olddefconfig' <guix-vits>rekado: btw, in gnu/services/networking.scm: %default-nftables-ruleset is almost the same. Probably i need to make <a>%default-nftables-ruleset</a> to Guix git, to the lines where it's defined. <rekado>guix-vits: the link will become stale as the code changes <Blackbeard>rekado: so you think I should list the systemd API things to implement, things like type, execstart, restart, and so on <Blackbeard>so I have a more detailed timeline with all the things I have to implement <Blackbeard>and that way you can track exactly what I am doing <guix-vits>rekado: then make a copy and inser it in footnotes? <Blackbeard>or rather, supposed to do hahaha I am not accepted yet <Blackbeard>the description of the gsoc project on libreplanet mentions that it is not necessary to implement dbus related things <Blackbeard>maybe I should mention testing for everything that should be implemented <rekado>Blackbeard: I think having a little bit of extra detail can’t hurt. Don’t go overboard with it. <rekado>but a little more detail would be appreciated. <rekado>because right now it’s not clear to me how we would do the mid-term evaluation. <civodul>oh, "guix upgrade -n" behaves poorly now <mbakke>civodul: how so? seems fine on 'core-updates', though there are currently no grafts there. <mbakke>janneke: whoops ;-) good catch, LGTM. <janneke>yeah, my fault: this patch was rewritten at least once; the documentation is correct btw (ugh!) <janneke>also, working on "coreutils" on the "core-updates" branch is terribly confusing <rekado>guix-vits: I don’t know. I think the manual should contain a literal example, not a vague pointer to code. <guix-vits>rekado: i fear to draw the attention from %default-ruleset, as it's looks good already :( <guix-vits>(as me copy-pasted Arch's one, before realizing what the default is) *guix-vits reading Arch Wiki, nftables. "Probably the default config can be compressed somehow to show it as an example." <civodul>mbakke: it depends on your store and your profile, but in my case it just says that it'll download a couple of things, and that's it <civodul>it stops before even printing which packages would be upgraded <civodul>that's actually expected, now that i think about it <civodul>perhaps we'll just re-add --no-graft for guix package --dry-run <rekado>beautiful heavy snowfall in Berlin now <rekado>restrictions apply; I only watch the snow through the window. <rekado>we only leave the apartment once a week for groceries *janneke searches for berlin webcams <guix-vits>in Manual's example of 8.8.4, "Scheme Variable: iptables-service-type" -- aren't the example should also contain a rule that permit loopback incoming traffic? <janneke>one of the things i liked to not-so-secretly enjoy a, was the televised webcam stream on the flemish tv showing many snowy mountains in the alps. i only got to enjoy them at the start of the lockdown...apparently someone decided they needed "real television" now and the livestreams are mostly gone <jonsger>I heard rumours that a Swedish TV station will send 500 hours videos of elk's. Though it can hold some days until you will see the first one :P <mbakke>you'll have to click "sesong 1" to get the list of episodes <jonsger>can I just send a patch series to guix-patches@gnu.org or do I need to somehow get a bug id first? <rekado>jonsger: you need to get a bug id first :-/ <mbakke>so Rust seems to builds fine with Make 4.2 on core-updates, ref #39949 <jonsger>rekado: how can I do that. git send-email wants a patch but I only want a cover letter first <rekado>it might be possible to tell debbugs to put all the other messages as a follow-up to the initial report, but the debbugs docs are sparse <mbakke>I haven't quite reached IceCat yet, but not expecting any trouble, and that was the last thing missing from my packages <rekado>jonsger: you can generate a cover letter first <rekado>or you can send a single email, wait for the bug address, and then git send-email --to=12345@debbugs.gnu.rog <mbakke>can we train sneek to return issues.guix.gnu.org URLs upon seeing a #12345 bug ID? <rekado>perhaps we should add our own bot to #guix <bricewge>jonsger: First tell git to write a patch set to file <jonsger>bricewge: yes. thats the question. how to do that or let git open an editor <bricewge>Then `git send-email 0000-*` that will send the cover letter <bricewge>jonsger: Try `--cover-letter`, I'm using magit so I don't know all the arguments this command takes <janneke>rekado: ooh, i've been playing with 8sync's irc demo; guile-irc sounds interesting! <mbakke>jonsger: if you have a set of patches on 'master', use 'git format-patch -n --cover-letter ..master'. <bricewge>jonsger: So you should have some file starting with 0000-* wich is the cover letter and the patch series would start at 0001-* onward. <mbakke>then edit 0000-cover-letter.patch with your favourite editor and send away <jonsger>ah got it you can't seperate the patch generating and the writing of a cover letter :( <mbakke>jonsger: derp, the format-patch commit specification above should be 'origin/master..' <wxie>mbakke: When will 1.1.0 be released? Will the coreboot bug be fixed as well? ***CcxWrk_ is now known as CcxWrk
<rekado>jonsger: you don’t have to generate a cover letter. You can also just write a free-form email to guix-patches@gnu.org and wait for a response. <rekado>think of it as an announcement of your upcoming patches <mbakke>wxie: the coreboot bug will only be fixed if we find a way to fix it! I encourage you to raise the issue on bug-guix@gnu.org so that it gets more eyeballs, hopefully someone have enough coreboot experience to know what the problem is <jonsger>hm. 30 minutes gone. All mails arrived but thunderbird makes 3 threads out of it :( <mbakke>wxie: 1.1.0 is due for release in the upcoming weeks, no definitive date yet :) <nckx>Good morning apparently snow-covered Guix. <nckx>Is ‘the coreboot bug’ new since the last release? <rekado>jonsger: do you think it would better not to propagate glib:bin and to patch invocations of gio-launch-desktop instead? <jonsger>rekado: I don't mbakke proposed it and we have to only touch one place <jonsger>rekado: and we can remove it when core-updates merges <mbakke>rekado: it's a temporary workaround since the issue has been fixed on core-updates <jonsger>ah in this cover-letter I forgot to mention the fix in core-updates :P <nckx>Happy holidays all, you lucky bastards. *jonsger found out that his mail provider allows only 20 mails per "session" :( <mbakke>my condolances to everyone snow-covered, here things are finally melting away :P <wxie>mbakke: Thank you. Is there a workaround? ***wxie1 is now known as wxie
<mbakke>wxie: I don't know enough about it to say, which is why I encourage you to file a bug. <mbakke>wxie: there are many people using libreboot, so I don't understand why coreboot does not work. <mbakke>hopefully there is some easy workaround we're overlooking that the fine bug-squashers on bug-guix@ can identify :-) <mbakke>guix-vits: it would be nice to include a more complete example with e.g. SSH and a web server instead of just the default ruleset <guix-vits>mbakke: can i use them as-is (with regard to the fact that shepherd cares about "flush" and invoking nft)? <rekado>guix-vits: linking to the Arch wiki from the official Guix manual is not a good idea. <brendyyn>this implies your a root, and then you are using sudo <guix-vits>and also i'll change "# The default ruleset" to "# My ruleset", as default isn't empty. <mbakke>guix-vits: the flush command implemented by the service does not need any special rules. <guix-vits>mbakke: yes, i meant: strip the "#! xyz/nft -f" and "flush" from the examples on Arch Wiki. <mbakke>guix-vits: I think 'flush ruleset' is necessary so the new rules won't be appended, not sure what to do about the shebang, hmm <mbakke>guix-vits: the shebang can be removed <mbakke>guix-vits: also, restarting the nftables service will flush the ruleset, so you are right they are not needed :) <jas4711>hi. do anyone use guix with pxe booting? i'm setting up a couple of machines and want to pxe boot them, and guix seems ideal for this but i can't find any documentation on how to do it <rekado>jas4711: doesn’t work yet, but it is a project for Google Summer of Code. <apteryx>rekado: on your machines, does the NFS service take a long time to start, doing lots of IO? <apteryx>on a desktop it is quite visible, because it delays the start of the graphical session by about 30 s <rekado>apteryx: I’m not currently using it. I don’t see this in the nfs-server test, though. <pkill9>my only guess is that the package definition replaces the 'check phase, which maybe overrides the "#:tests? #f" argument <pkill9>I'll try removing the phase instead ***alicemaz is now known as Guest89630
*guix-vits wiki.nftables.org: "iptables... Thus, all chains are registered even if you only need one of them." -- so i can omit the output chain in the examples (as nothing is blocked, anyway). <PotentialUser-61>Hi I am getting `Gtk-Message: 17:38:47.078: Failed to load module "canberra-gtk-module"` error while running emacs, icecat from my terminal. I have installed these packages using guix over my ubuntu bionic distribution <kovind>Hi I am getting `Gtk-Message: 17:38:47.078: Failed to load module "canberra-gtk-module"` error while running emacs, icecat from my terminal. I have installed these packages using guix over my ubuntu bionic distribution <apteryx>nckx: I've migrated to a desktop machine at home, and my primay x11 clipboard is again fussy about pasting: I must middle-click exactly on the prompt character on the line I want to paste to... That's quite annoying. *apteryx wonders what setting/driver could cause this behavior <guix-vits>kovind: i'm not a developer, but you may try to `guix install libcanberra`. <kovind>guix-vits I have already tried it <apteryx>guix-vits: that's on my TODO, but I'd have to migrate from ratpoison as well, IIUC. Which I don't have the time to do ATM :-) <pkill9>instead it tries to rebuild minetest <apteryx>guix-vits: I'm mostly interested in the better DPI handling that I expect Wayland to be able to do. <pkill9>i've tried inheriting (package-replacement curl) in curl-graft instead *guix-vits *nfs -s list ruleset*, not "nft -s list". stupid i am. <kovind>I think due to this issue `"canberra-gtk-module"``, I am getting weird fonts in emacs and icecat. By the way icecat is completely unusable. *guix-vits probably, as a + to examples that will be, the nftables project's wiki should be included as a link to our manual. <guix-vits>kovind: did you'd installed the fonts mentioned in the manual? <ngz>kovind: FWIW, I think this is unrelated. I get the canberra-gtk-module error, but my fonts are fine in Emacs. <ngz>kovind: Don't forget the "fc-cache -rv" dance. <kovind>guix-vits can you refer to the link where I can check this <guix-vits>(replace "manual-devel" to "manual" if you need the "stable" documentation applicable to the instalation image state) <kovind>The issue persists even after installing all the fonts and that cache thing <kovind>Its pretty bad I can share screenshot, <kovind>The icecat is completely unusable, emacs is still usable. the emacs look like lucid version <apteryx>kovind: are you using Guix on top of a foreign distro? <apteryx>and you have run guix pull and updated your whole profile recently? <kovind>I am new user of guix, I think it is similar to `apt update` and `apt upgrade` <apteryx>yes, basically 'guix pull && guix upgrade'' as your user should update your Guix profile. <kovind>I am getting this `Gtk-Message: 18:08:55.507: Failed to load module "canberra-gtk-module"` error message in my terminal. May be this is main issue. I have already installed libcanberra-gtk via guix <guix-vits>kovind: make sure it's 'libcanberra', as guix has it and 'libcanberra-gtk2' only. <kovind>apteryx Icecat is completely ruined. Emacs is still better. Both have same error <kovind>guix-vits I have installed both `libcanberra` and `libcanberra-gtk2` <guix-vits>kovind: can you try this: `guix environment --pure --ad-hoc emacs -- emacs`? <demotri>kovind: I'm pretty sure it has to do with the fonts. I have a working installation of icecat (via Guix) on top of Ubuntu. Fonts are "nice". <demotri>To confirm, I started icecat from within a Guix container: guix environment -C --share=/tmp/.X11-unix --ad-hoc icecat <demotri>Then inside the container, you have to set env-variable for X: <demotri>If I do it like that, then I also have the "bad-fonts" problem. ***mroh_ is now known as mroh
<dongcarl>A source of non-reproducibility that may affect many packages for us: a lot of packages bootstrapped with libtool 2.2.7a will not produce reproducible archives, since the `find` invocations are not sorted. <dongcarl>I'm wondering if there should be some kind of detection and auto-rebootstrap in that case? Or do we do it case-by-case? <dongcarl>A prime example: gcc, for all released versions so far, is bootstrapped with libtool 2.2.7a <apteryx>dongcarl: perhaps for now we can do it case by case, and if it occurs that there are enough packages affected to make it work it, then automate it in the gnu build system. <dongcarl>Cool, I'll post on the ML just so people can watch out/consider it when making packages reproducible. *guix-vits remember that he forget to use --rounds with neverball... <pkill9>im getting this error when trying to `make` guix: automake-1.16: error: cannot open < ./doc/guix-cookbook.de.texi: No such file or directory *raghavgururajan also successfully packaged qt version of linphone \o/ <apteryx>is the Qt version what they call linphone-desktop? <apteryx>or is there actually multiple flavors of linphone-desktop? <raghavgururajan>apteryx Linphone is almost ready to use. All packages successfully builds. Just need to patch some run-time gtk/qt related errors. <apteryx>I'll be happy to test, or provide assistance with those last errors <raghavgururajan>apteryx Current "released" versions have packages named linphone (lib+gtkclient) and linphoneqt (qtclient). In upcoming versions, those packages will become liblinphone (lib) and linphone-desktop (qtclient). <apteryx>so, they are migrating from GTK to Qt, if I get it correctly? ***xkapastel is now known as Guest45017
<raghavgururajan>apteryx Yep, they are migrating to Qt. Also, they are separating library and client into two different packages. So, in future, there will be one unified linphone library (liblinphone), and different clients (linphone-desktop, linphone-android and linphone-ios). <raghavgururajan>For now, the run-time errors appear to trivial. gtk client needs some engine/theme/module and qt client needs qtquick module. <apteryx>np! I'll give it a spin after my work day ends ***xkapastel_ is now known as xkapastel
***slyfox_ is now known as slyfox
<Blackbeard>Just waking up, going to improve my gsoc application a bit <guix-vits>Blackbeard reading the nftables' wiki; started to listen to EDM-genre. <Blackbeard>guix-vits: is nftables in guix? I never used it. But setting up the firewall is something I should do <guix-vits>Blackbeard: (service nftables-service-type %my-nftables-ruleset) <mehlon>Blackbeard: it means electronic dance music, basically dance music that uses "electric" sounds, there's glitch, trap, dubstep etc <guix-vits>Blackbeard: default one in gnu/services/networking.scm (it's similar to Arch's default) <guix-vits>Blackbeard: (service nftables-service-type) gives nftables running with the defualt config. <mehlon>Blackbeard: probably? and Armin van Buuren and anything you hear in dance clubs is probably EDM <guix-vits>mehlon: are you sure that Blackbeard visiting that land-dwelling rats' dance clubs? On a peg and with the bottle of rum (and the crew, and the Davy Jones closet)? <Blackbeard>guix-vits: can it be used to filter hosts and ads? <guix-vits>Blackbeard: idk, as i'm not really know how it works. When i was on Arch, i'd used it for that, but found that hostname-based blockers are lot better. <apteryx>Blackbeard: a hosts-file can be used to filter out hosts. Search for %facebook-host-aliases in the manual. <guix-vits>Blackbeard: there is a chances that the above mentioned hosts-file manipulations will be needed, as browsers keep trying to connect to IPs and give up not that fast (so page-rendering can become quite slow). Hosts-file manipulation tell the browser: "you've me connected to X, so cheel off!" <mehlon>oh yeah, that's dubstep all right <Blackbeard>apteryx: yes, but I am not sure where I read that hosts files are slow and there are better ways to deal with adblock <mehlon>personally I prefer to connect to a DNS-over-TLS supported adblock server <mehlon>and then ublock origin on top, unless you use Bromite for android <Blackbeard>mehlon: but do you know any that is Freedoom respecting <guix-vits>Blackbeard: also consider this: if you're blocking all outgoing by default, then you'll need to check all the cases (maybe with tcpdump), like weather-apps, bug-reporting apps, IRC... <mehlon>they all have their problems yeah <Blackbeard>guix-vits: ahh 🤔 although my traffic comes mostly from browser, wget, git, and email but that's good to know <guix-vits>Blackbeard: but i'll try to block outgoing http from my machine (and allow httpS) <mehlon>though subgraph does it with application sandboxing, not iptables <mehlon>maybe you can try blocking all non-tor traffic and become tinhat <guix-vits>mehlon: it's not recommended afaik (to use only tor, for everything) <guix-vits>mehlon: ask on #tor to know for sure; i'm not remember when and where. Just all of that, starting from they aren't recommend to use tor outside the TorBrowser. But i'd seen that for sure. <mehlon>this may be because your applications tend to de-anonymize you unless they are specifically tuned for Tor <mehlon>but you can use tor for most things, such as accessing IRC <galaxie>mehlon: I mean, I'm using Tor for IRC.. <guix-vits>mehlon: idk, honestly. But logically: tuning of one app is easier than 100500. <mehlon>that's why one would use Whonix or Tails, where all the applications are pre-tuned for Tor <guix-vits>mehlon: and watch out to not say "100500" in every of the apps (as this is an local meme)? <guix-vits>or that "hello from Kemerovo!" -- anonimous one of the 100 Linuxoids in Region X <mehlon>I suppose it's not smart to say "hello, it's me, mehlon, real name mehlon mehlonson from mehlonia on mehlonstreet 1337" everywhere <rekado>this seems to be drifting off-topic for #guix <mehlon>especially since only I can say that *guix-vits if i use "device enp0s29u1u2" in my nftables.conf, i can't start the service. mistery... need to think. <pkill9>does anyone maintain their own version of guix? <lfam>What do you mean pkill9? I think most people involved in Guix have some private patches but I wouldn't call that a separate version <pkill9>lfam: private patches that they use for their own machine <pkill9>is there a good workflow for that? <pkill9>or just good tips for making it as smooth as possible i mean <lfam>It depends on what kind of conflict you mean <apteryx>pkill9: for regular Guix dev I just use Emacs + Magit, which I'm most comfortable with. I resolve tricky conflicts with ediff (accessible with the 'e' shortcut on a conflict in Magit). <lfam>I think the best way is to use channels, but I am still using GUIX_PACKAGE_PATH <lfam>Or just rebasing Git commits <pkill9>apteryx: do you point guix to your local repository in your channels.scm, or do something else? <lfam>But it depends on if you mean code conflicts, like in Git, or conflicts between variables exported in different places <apteryx>pkill9: when you're in your Guix tree you can use ./pre-inst-env to run it from there, and use git to navigate between any branch/tag/commit, so I'm not sure what you mean by "pointing Guix to your local repository" <apteryx>I don't attempt to make my user installei Guix (with guix pull) interact with the packages defined in my ~/src/guix Guix checkout. I just use ./pre-inst-env guix to tests local changes. <joshuaBPMan>lfam: I would agree that channels are pretty awesome! <joshuaBPMan>That's a service that a wrote that will ONLY work for me. It hard codes the vpn set up. <raghavgururajan>sneek later tell apteryx: Gonna need your help to fix gtk/qt run-time errors. 😅 <pkill9_>apteryx: i mean with `guix pull`, i want to change a package used by many other packages, and have that change affect them all, specificlaly i want to build curl with a ca-path set so applications using it find the SSL certs <pkill9_>and i want to avoid using the environment variable for build option for two reason: 1) i think it's slower to have it replace inputs each time, and 2) it removes any issues of environment variable being set <apteryx>pkill9_: that's not possible using channels unfortunately. Someone corrects me if I'm wrong, but channels can only *extend* Guix, not modify it in any way. <sneek>apteryx, you have 1 message! <sneek>apteryx, raghavgururajan says: Gonna need your help to fix gtk/qt run-time errors. 😅 <pkill9_>apteryx: you can replace the guix channel <pkill9_>also, i find dynamically applying graft replacements seems kind of temperamental, for some packages like mesa it doesn't seem to work, it seems to try to build stuff it shouldn't <apteryx>now I understand where the needs of maintaining a custom version of Guix comes from <joshuaBPMan>so I'm reading via phoronix that the intel developers are considering removing the classic mesa drivers from the tree. That includes the T400 graphics driver. That probably means that other libreboot laptops may have issues in a year or so. <pkill9_>why are they considering doing that? <joshuaBPMan>It sounds like they may keep the classic drivers in a fork. <joshuaBPMan>It sounds like they are trying to avoid end of lifing old devices. But essenitally it is a big maintence burden updating all the old drivers. <joshuaBPMan>so maybe it a year we'll have to have a guix package for mesa-classic-drivers. <vagrantc>much easier issue for guix to handle than other distros ... <vagrantc>e.g. possible to have multiple libraries installed in parallel... <lfam>Easy to handle the packages but no capacity to maintain them <lfam>I mean, to maintain the development of the software <lfam>apfel: Thanks, I think it's related to a recent commit that changes how we handle configuration files for nano <lfam>Commit bd71818df31364bbd04f3f13b3bbb251636748b4 <lfam>apefl: Does your config.scm do anything special regarding nano? <lfam>I'm going to try reproducing the bug now <apfel>Ifam: no, my config does not do anything with nano <efraim>I'm on my phone right now, looking at the error it looks like I should've created nanorc and then stuck it in place and now it tries to place it in a directory <lfam>It's lfam with a lowercase 'L' :) <efraim>Probably best to revert it and I'll fix it for real later <lfam>No worries, just want to make sure I get notified when you mesage me, apfel :) <lfam>Okay efraim, I'll do that now if I can reproduce it <lfam>apfel: I reverted the change that caused the bug. You should be able to reconfigure after doing `guix pull` <efraim>I mostly use GUIX_PACKAGE_PATH for my private packages, most of which are my-foo. I haven't played with special packages in my os-config <efraim>Also I realized I'd not been using pre-inst-env on my slow machine and the openssh service is really fixed <lfam>efraim: You mean the problem where the sshd would fail to start is fixed? <efraim>I basically tested using my kids' computer, an old clunker with spinning rust, almost 2 minutes to get to the slim login screen <lfam>For me it would work for months at a time before failing again <lfam>I figure it's a consequence of data layout of slow disks *vagrantc is bit by the nanorc in skel issue too <lfam>vagrantc: Fixed in `guix pull` <efraim>Not as bad as when I broke openssh <vagrantc>hrm. plugging in an ath9k-htc usb adapter kills my usb bus <lfam>I mean, I think it's clear by now that it's related to slow IPv6 interface set-up. Something about that is not being captured by our networking-service and the service is reported as "done" too early <vagrantc>good thing the built-in keyboard has dead keys... really improves my creativity when typing <lfam>Since I disabled IPv6 in my SSH service the problem seems reliably solved <apfel>lfam: it works, thanks for the quick help! <pkill9>i have a weird issue, running `guix package -n -u` only seems to show one package that will be upgraded <pkill9>i'll see if it's due to my change <janneke>lfam: thanks, i looked what went wrong and your revert was already there! <lfam>Just '-u' as the last argument should be the same as '-u .' when followed by another argument <pkill9>it's doing the same thing in vanilla guix, but i'm using it from a separate profile <pkill9>i'll see what happens if i pull it into the default one <Blackbeard>pkill9: I am doing a guix pull to try to reproduce <civodul>mbakke: i have a simple fix for the "guix upgrade" issue i mentioned this morning *civodul is in an enbug/fix loop <Blackbeard>pkill9: but I tried without the "-n" and it is upgrading other stuff, like youtube-dl or next <pkill9>Blackbeard: that's the issue i'm getting <Blackbeard>pkill9: mpv, gnupg, sbcl, and more stuff is being downloaded now <pkill9>Blackbeard: is it saying each time 'the following will be downloaded' on every package? <pkill9>civodul: is this the issue you say you have a simple fix for? <vagrantc>hah... updated linux-libre to 5.5.13 only to see 5.6 released <vagrantc>well, at least the linux-libre patches are on top of it! *vagrantc builds linux-libre 5.6 ... <vagrantc>maybe that contains fixes for pinebook pro, too :) <civodul>this is from a recent Guix, like from yesterday, right? <Blackbeard>civodul: I did a guix pull and upgrade, yesterday and today <civodul>Blackbeard: ok, hopefully i'll push a fix real soon <anadon>nckx: civodul: I have a draft paper for re-thinking files I'm putting together and I think it can now see the first lights of day. It was originally conceived by trying to solve GWL's need for tracking input computations and resulting output. Is here, the GWL maling list, or somewhere else a good place to send it out for critiquing? <civodul>anadon: nice! probably the GWL mailing lists is the best place <civodul>or if you think it's more general, you can email guix-devel <nckx>anadon: I was going to suggest guix-devel just for the larger audience. <nckx>By complete coincidence, I just approved a ‘call for papers’ mail to that mailing list 🙂 <civodul>i'm happy to be Cc'd anyhow because that sounds interesting :-) *civodul realizes being unsure who the nickname maps to <anadon>OK. Just a heads up, it is incomplete. I'm trying to keep an absolutely clear and complete concept and have the specification flow from that. I've re-written it more than a few times. <anadon>Subscription needs to be acted upon for "joshua.r.marshall.1991 at gmail--" <nckx>Annoyingly, I don't think we can help with that. Have you received and acted upon your confirmation message? <nckx>You can already send mail without being subscribed tho'. <anadon>OK, email sent. I hope this goes well. <anadon>And also, so this work now enters public domain. I should have added a license. <nckx>You can always do that later. <Blackbeard>anadon: that seems interesting, when can I read it <nckx>Licencing is Deny *; Allow 🙂