<nikita`>the probably unused but still existing pycanbera or whatever I named it back then in guix (py-pycanberra?) <nikita`>i think ftp.n0.is didn't exists for some time and it was just d.n0.is, so that must've been broken when I removed the redirect <nckx>‘No dependents other than itself: python-pycanberra@0.1.1’. <nckx>Would be nice to have it fixed of course but nothing urgent. <Blackbeard>nckx: git send-email is cool :) I just set it up <nikita`>old tar ball has been added to the release <nckx>Thanks. I'll push an update soon unless you're already on it. <nikita`>i still haven't gotten back to having enough computers to also do guix ***modula2 is now known as defaultxr
<nikita`>got this has embarassing little own work, it looks like I wanted to start making changes and dropped it at some point <crab1>Hey, I'm trying to write a package: 0x0.st/iQq9.scm and it fails on the 'configure step, but it should be disabled, can anyone help me figure why 'configure is still happening? <mbakke>crab1: I tried to run it and got no configure phase <mbakke>make: Makefile.linux: No such file or directory <crab1>oh silly me, I was re-reading an old log file <mbakke>crab1: you can use 'guix build -f the-file.scm' to get build output straight in your terminal <crab1>thanks, I have been lacking that <crab1>hm now it doesn't know what cc is, I should know how to fix that too, but alas, I do not <lfam>Either set CC=gcc in the make-flags, or patch the Makefile <Blackbeard>ice-9/eval.scm:293:34: In procedure abi-check: #<record-type <operating-system>>: record ABI mismatch; recompilation needed <nckx>Blackbeard: You need to recompile everything. Start with make clean-go to remove the old, stale files. <crab1>hm now what to do when a package wishes to populate /usr/local/share <nckx>crab1: Set PREFIX= in #:make-flags. <nckx>Probably to (assoc-refs %outputs "out"). <Blackbeard>nckx: when I open a guix .scm file inside emacs it gives a message about the variables containing values that may not be safe <Blackbeard>why do I keep getting the massage that source files are newer than compiled .go files, when I just cleaned and ran make <crab1>why am I getting "No rule to make target 'prefix-retro-latest'", why does guix try to specify a target? Aren't targets specified in the Makefile? So how does "prefix-name-version" make sense, since it's not going to be a target in the Makefile? <nckx>crab1: Guix does no such thing. <crab1>I'm trying not to use y'all as my rubber duck, but that seems to be how it's going <nckx>No problem. I'm curious what exactly happened (I'm guessing a stray name-version in make-flags). <crab1>basically, I put in the assoc-ref but forgot to place "PREFIX=" before it <nckx>I didn't mention the implicit (to us…) string-append, hope that didn't lead you astray. <crab1>no, I understood, this was oversight on my part <crab1>guix is more unfamiliar to me than scheme :) <Blackbeard>nckx: if everything went right, I just send my first patch with git send-email :) <crab1>if I have to define MANDIR=$PREFIX/man, should I repeat the (assoc-ref ...) or reference the rebound $PREFIX directly somehow? <nckx>crab1: Do you? The Makefile has MANDIR ?= $(PREFIX)/man/man1 already. <nckx>Blackbeard: Nice. I'll keep an eye on my mail box. <crab1>I am getting the error "cannot create regular file '/gnu/store/prefix-retro-latest/man/man1/retro.1': no such file or directory" and for some reason I figured that putting them in .../man/ instead of .../man/man1 would fix the issue but I'm unsure <nckx>If that's a literal error your PREFIX is wrong. Did you substitute ‘prefix’ for a hash yourself? <nckx>I'll continue assuming so and that your PREFIX is correct: <crab1>wdym by "substitute prefix for a hash yourself"? <nckx>The ‘install-manpages’ starts off with ‘install -c -m 644 man/retro.1 $(MANDIR)/retro.1’, assuming MANDIR exists. You need to add a short (add-before 'install 'create-output-directories (lambda* (#:key outputs #:allow-other-keys) (let* ((out (assoc-ref outputs "out")) (man (string-append out "/share/man/man1"))) (mkdir man) #t)…) phase. <nckx>I… should not have pasted that as an IRC message. Oh well. <Blackbeard>nckx: does it take longer to get a response in this case? I always got an acknowledgement response right away <nckx>crab1: cannot create regular file '/gnu/store/prefix-retro-latest/man/man1/retro.1' <nckx>Does it actually say ‘prefix’? <nckx>Blackbeard: No, it should be identical. Our bugtracker is sometimes slow and sometimes not, this is not unusual (yet). <pkill9>what benefits could the Hurd have for desktop systems? <crab1>so now I wonder why is this a necessary step to add when on a non-guix system I would not need to create output directories manuall? <nckx>Blackbeard: You used --to=guix-patches@gnu.org or have ‘to = guix-patches@gnu.org’ under [sendemail] in .git/config, right? <nckx>crab1: Because they'd already exist? <nckx>Every Guix installation starts with an empty target directory. <nckx>Some packages don't expect that, they assume they're never the first package to be installed somewhere. <xelxebar>Blargh. issues.guix.gnu.org is still broken. <xelxebar>The issue I'm encountering looks to be on there, but the server is still returning 500 when trying to open it. :( <xelxebar>Kind of dead in the water at the moment, trying to define my first package <nckx>crab1: Since (I *think*) this means that this package's ‘make DESTIR=/foo’ behaviour is also broken, I don't think it's worth working around in Guix. <nckx>xelxebar: Is not so much broken as on hiatus. <nckx>Mumi (issues.) is just a front-end. The connection with debbugs (where issues gets all its data) has been cut. I'm not sure why, but I think the debbugs team thought we were causing too much load. <nckx>rekado_ thought this shouldn't happen so I guess it's down for investigation. <nckx>Don't quote me on any of that though, it's just a half-remembered glance at the backlog. <nckx>Many people (ahem) still use that because issues. just can't be relied upon when it counts. <xelxebar>nckx: You're Mana from Guix Heaven. Thank you. <nckx>?package=guix-patches for patches. They're not unified like on issues. <nckx>Don't get me wrong: issues is much nicer, when it's up. It just has… issues. <nckx>Blackbeard: Your patch made me notice a typo in the preceding package (‘previos’). Thank you. <nckx>Blackbeard: I'm a bit confused. What's the last package in your emacs-xyz.scm, above emacs-typing? I don't have it but it's not in my mailbox as a patch either. <Blackbeard>nckx: I've sent emacs-org-roam, emacs-uml-mode, emacs-4clojure, emacs-typing <nckx>One of them has a typo (‘previos’) in the description. Searching for that in my mailbox returns only emacs-typing because of the patch context, but maybe it's my MUA's fault. <Blackbeard>nckx: but now that you mention it, I think I made a mistake and did not send 4clojure <nckx>Blackbeard: It's not a rule but I prefer adding most packages in somewhat alphabetical order in the .scm file, not always to the end. It makes merging them much easier. Just something to keep in mind for your next patches. These are fine. <nckx>Blackbeard: It's not in my mail box. <Blackbeard>nckx: yes, sorry I should have sent emacs-4clojure yesterday night, I was too tired and just wrote the patch and forgot to send it <crab1>does (assoc-ref %outputs "out") include the "/local" portion of the path? <Blackbeard>ohh i made the mistake of sending emacs-typing patch to bug-guix not to guix-patches <nckx>Blackbeard: That's all right 🙂 <nckx>I mean, don't do that next time, but it didn't break anything. <Blackbeard>no I corrected it. I have an org-file where I document all the workflow to send a patch. I just wrote the correct email <nckx>I've sent a follow-up to your 4 patches with the correct order. That's my fault: when you asked about sending them separately I didn't think about patch context, since I never add my packages to the end like that. <nckx>No need to be sorry with that attitude 🙂 *nckx goes to make some tea, because coffee shops are a non-essential business now 😒 <Blackbeard>nckx: here almost everything is still working :( <nckx>Blackbeard: For new packages please do. It's not possible to do it well: most .scm files aren't in alphabetical order (there are exceptions like perl.scm). So I add a package in the first spot it fits. <nckx>If two new packages still end up right under each other, please do send them as one bug. <nckx>But at least it should happen less frequently. <nckx>crab1: What's the problem? <nckx>Most of us here have seen many ways in which packages can fail incomprehensibly. <crab1>I've added in the create-output-directories phase and it is failing with "error in procedure mkdir: No such file or directory", no other information is really given, it is apparent where it happened and that's about it <nckx>Guile is no innocent where horrible error messages are concerned, but ‘error: No.’ is all Unix. <nckx>Returning the actual subject of an error would have wasted valuable bytes. <xelxebar>nckx: What? "error: No" is way too verbose. They way of enlightenment is ed: ? <xelxebar>Is anyone able to build master (ab3ad88ded)? doc/guix.de.texi is throwing lots of "@ref reference to nonexistent node" errors for me <nckx>xelxebar: I've been building Guix all day. Let me re-bootstrap. <nckx>Still running. Creating the XDG mime cache took this long. Unbelievable. <apteryx>nckx: I think we need to tackle optmization of the profile hooks at some point. It's getting unwieldy. <nckx>This machine is a special pathological case, not a data point, but yes, these things need optimisin'. <nckx>I'm so glad you volunteered. <apteryx>I'd say that 'guix install something' running on my 160 ish packages profile is probably slow to the point of scaring onlookers away from Guix. <nckx>apteryx: Do you have an SSD? On my X230+SSD it's OK, but I hate to be ‘that guy’ who expects everyone to switch. <xelxebar>I'm confused. Is this directly related to the ./bootstrap time or instead running guix environment? <nckx>The guix environment guix required to bootstrap, so… both? <nckx>This machine has random HDD reads in the low single-digit MiBs, not normal, it should not take an hour on any sane box. <nckx>Wow. I obviously don't have enough packages. <nckx>xelxebar: I'm pretty sure they sound the same. <apteryx>guix install ed --> 1m4.736s To be fair, most time is spent the manual database, which was a hook I added in the first place... eh. <nckx>xelxebar: I don't know if it's still relevant at this point, but Guix master builds fine. Sorry about that. Might you have some stale files hanging around? <xelxebar>nckx: Still very relevant. I am not sure. I've done make distclean and git clean -df on the repo. Not sure what I'm screwing up, especially since I'm running in a --pure environment. <nckx>xelxebar: I'm not actually familiar with git clean. Is it like ‘git reset --hard’? <nckx>Reads like it but this is git so who knows. <xelxebar>Pretty much. It removes everything not under version control. <nckx>Seems like it might not touch changed but tracked files. <nckx>Does git diff report anything? <nckx>My guess is it will report something in doc/. <xelxebar>Thanks. I'll try cloning a fresh repo and see if it builds there <nckx>Let's do that. It'll at least reveal which one of us has the weird (non-)building repo. <Blackbeard>nckx: is it ok if I send responses patches for everyone of my bugs, putting them in the correct alphabetical order? or should I just wait <nckx>Blackbeard: Don't bother for these. <Blackbeard>ok, I will send the new one now in proper position <nckx>Your previous patches were not ‘incorrect’; this isn'ta rule in Guix. I just think it has advantages. <Blackbeard>nckx: I can make a patch sorting by alphabetic order everything in emacs-xyz <nckx>Thanks, but not worth it 🙂 <xelxebar>nckx: Well... make got past the *.texi translations <xelxebar>head: invalid number of lines: 'po/doc/guix-manual.zh_CN.po' <xelxebar>This is bothering me, so I'll see if I can figure out how my repo which looks clean actually isn't <nckx>xelxebar: Can you ‘git reset --hard’ and run make again (inside the guix environment), just to make sure? <xelxebar>For clarity, this should be equivalent to what I did: git pull <guix-url>; cd guix; guix environment --pure guix; ./bootstrap; ./configure; make <xelxebar>I feel like those "head: invalid number of lines" and "line++: not found" errors are indicating that I'm doing it wrong <Blackbeard>despite the fact that I screwed it with my earlier patches <nckx>You didn't screw anything. <drainful>Would a FOSS video game client which is indended to connect to proprietary server software be compatible with Guix? <brendyyn>drainful: I've used that before, but I wasn't sure if it downloaded any proprietary java code or not, or is it completely stand-alone? <nckx>drainful: Probably. Any free software that doesn't recommend non-free software to users for use on their own machines, and otherwise complies with the FSDG, can be included in Guix. That doesn't mean it will be; that would depend on the software and discussion. <nckx>Talking (exclusively) to non-free server software alone doesn't bar it from inclusion. <drainful>brendyyn: All of the software in the git repository is freely licensed. I don't believe there are any other components <usney>is there a way to install guix package manager on mac os x? <usney>I mean a really old mac os x from the powerpc days <nckx>usney: …and everyone else, so we generally discourage ‘guys’ here 🙂 No, there isn't. <nckx>Guix System can run on those machines, but not with OS X in between. <nckx>Hm, no, I shouldn't say that. I *think* it can. <usney>In the states guys can refer to both genders I am not sure about other dialects of English. Even in the States they have slightly different usage from region to region here. <usney>but I will try to remember to say "hi everybody!" <nckx>usney: Thanks! So many better words to choose from, and that won't lead us into dull discussion 😛 (as a semi-USian myself, boy would I have opinions to bring to that tedious table…) <nckx>I worked in Italy until the virus came. <usney>cool are you still there now? <usney>I heard they aren't doing good there with covid19 <nckx>Well, no, hence the flag & past tense & general sense of sadness. ☹ <nckx>You're not doing too shabby yourselves. <Blackbeard>Let's not talk about covid please. It is depressing :( <brendyyn>drainful: probably its ok then. i never expected it to be though ;). I know technically its against their rules to use it but everyone does anyway i think <usney>oh I got my guix fonts to work again! <nckx>Blackbeard: It has a huge effect on people's lives and if they need to get that off their chest I think that's fine in moderation (see what I did there) but you're certainly right that Guix is more fun. <nckx>brendyyn: Against the rules of what? <usney>I think it was because I switch desktop environments or because I didn't update and upgrade in a while. <Blackbeard>nckx: I know sorry. I don't want to be like that. <nckx>brendyyn: I don't know what that means so it's not about GNU rules so that's fine! <brendyyn>nckx: yeah its fine. basically it was a proprietary game, then users made their own free client for it regardless of the game developers rules and a large % of users started using it, so its not like they would ban them. democracy wins :) <usney>I no longer live in an urban environment it is a rural college town so because of the college it has all the basic stuff you would need. :) <usney>So we have very few cases here. <drainful>Blackbeard: It fetches non free assets from the server, but the distributed jar only contains some freely licensed icons. <usney>It would suck too if no college town satellite internet I heard sucks big time <usney>I really like nnn file manager it is a tui command line app with the ability to open files in graphical programs <nckx>brendyyn: That's great, that the success story we want (and that we're sometimes accused of using as a cover story to run, say, Wine). <nckx>I used to use ranger but switched to nnn for those tasks. <usney>I just started using it today or yesterday <usney>so still learning how to use it <usney>right now I want to see how I can change it to mount disc images instead of opening a disc burner app. Since I don't really burn discs anymore. <nckx>Blackbeard: I understand, and I hope you stay safe. <usney>yes I hope you stay safe also Blackbeard <nckx>usney: I don't actually use the graphical openings that much, but I assume it just uses the XDG MIME database. You can change those associations (just don't ask me how), it's how desktop environments manage their ‘Open with…’ preference. <Blackbeard>drainful: are the non free assets fetched automatically? <nckx>You pay enough for the damn thing every time you generate a new profile. Might as well use it. <usney>yes I didn't ask you how because I have been tinkering with my system and its config files I just said that in case you did know or someone else did to solve that issue. :) <Blackbeard>thanks nckx and usney :) I hope you are safe too. I am glad nckx is better :) <usney>I am going to figure it out on my own most likely <usney>I have solved some of my issues I have having with openbox and running guix programs on a foreign distro. <usney>I can't get it in the menu but I can run it from terminal though. <usney>I just created a custom launcher in the openbox menu in the config file. <usney>then I use rofi -show run to open a run dialog to run apps I don't use very often that I installed with guix. <usney>qbittorrent was crashing a lot on pureos but the one from guix package manager works a lot better plus it can understand my config folder which sometimes newer versions of software can have problems with. <usney>I mean the config folder for any software when you install really new software over it <usney>I've had this issue in the past but not lately <nckx>To be honest, ‘the Guix version works better and crashes less than the native package’ isn't something we hear often but it feels good to hear it. <nckx>Hm, it's not Opposite Day. <usney>well the guix qbittorrent package hasn't crashed yet <brendyyn>i should add transmission qt so i can throw out the ugly gtk version <usney>how come you don't like gtk? <usney>I don't like all the gtk themes <brendyyn>with a drop down list on qt, you can click and hold the mouse in, move down and release to select an option. on gtk that doesnt seem to work <usney>they even have some gtk themes that look like for a computer from 1995 <nckx>brendyyn: Really? It works here in GIMP. <drainful>Blackbeard: I'm going to do a bit more research before continuing, but at leask I know that in principle software with this function could be packaged in Guix. <usney>maybe it is a transmission app bug? <nckx>Not doubting you, just wondering if this is a theme or desktop thing. <brendyyn>gtk2 or 3 thing? developers can set the option? who knows <usney>I am not having issues like that with icecat on pureos with guix installed as a foreign package manager. <nckx>Hm, IceCat indeed not, but they have their own custom(-looking) menu so I'm less surprised (not less disappointed, but still). <usney>I'll have to download and install that <nckx>Me? Sorry, for reasons too tedious to explain. <usney>I have it for apt but not for guix <usney>I am not having any issues with it not working right with apt so I didn't try the guix package <nckx>usney: You'll need to install transmission:guix to get the GUI. <nckx>Er, transmission:gui. Force of habit. <usney>but now since I can use qbittorrent again I no longer need transmission <usney>so I will uninstall it and install transmission via guix <brendyyn>do we have a client that supports webtorrent? <usney>it will take a few minutes maybe because I am using almost a decade old laptop. But I am using a very light DE and it isn't xfce or lxde or lxqt its even lighter. <nckx>xelxebar: I've cloned Guix from scratch into test-guix, then ran guix environment --pure guix, ./bootstrap, ./configure --localstatedir=/var, make -j4. Still can't make it fail… <brendyyn>I get this warning on almost every gtk application) <usney>guix search peertube | less | grep webtorrent <nckx>xelxebar: I do get the bogus ‘head’ warnings, and they need to be fixed, but they weren't fatal. <usney>I used that didn't find anything <nckx>I'm honestly out of hypotheses. <brendyyn>that search isnt exactly going to find it <usney>when I search for other software <Blackbeard>there are two games I want to package, the first one is "Blob Wars: Metal Blob Solid" and the second "Blob Wars: Blob And Conquer" <Blackbeard>what should the names be? blob-wars-metal-blob-solid ?? <nckx>It should work. There's just no mention of ‘peertub’ in Guix. <nckx>That's something else. ‘Peertube.’ <usney>maybe search for what network it uses? *brendyyn holds nckx back <nckx>Must… not… /kick… … friend… ***xMopx- is now known as xMopx
<brendyyn>nckx: i made all my money off bitcoin. kick me :) <nckx>brendyyn: I'll mainly kick myself for throwing away my bitcoins when I had any. Yes, I'm one of them. Scoured old hard drives but no luck. <brendyyn>In other news i just discovered the `sync' command. I was using cat to copy an iso to usb. turns out when the cat command returns it still hasnt finished yet <usney>I am going to install transmission now to see if I had that same issue you have brendyyn <nckx>Eh, ‘several’, at least :-/ <usney>I am using an old laptop so it might take a while <nckx>Enough to buy a sweet set of wheels. <usney>but I am using a really light DE even lighter than lxde/lxqt and xfce. <nckx>Blackbeard: I guess that's fine. I don't see any better options, either upstream or in my imagination. <brendyyn>its alright, i never realised how destructive crypocurrencies were at the time. I'm much more excited about GNU Taler <usney>oh yes I read about gnu taler <usney>what is the status about that software? <brendyyn>Im not sure but i think it all works, just hasn't been adopted by banks and companies <usney>what's the best way to use proprietary software when you have to for work and doctor visits? I am thinking about using a usb gnu/linux with slax for that. <usney>I don't want to install it on pureos <usney>since that would defeat the purpose of pureos <nckx>I see there was an ‘fsdg’ distribution first, then it got combined/removed/…? Probably nothing, but might be worth checking what that means. <nckx>Eh, dfsg, I will never not confuse the two. Still, same thing. <nckx>Blackbeard: Chillax, friend. It's fine if that happens once or twice. <usney>I am just verifying my linux isos in transmission before I purge transmission in apt and install transmission with guix. <usney>they have a transmission emacs edition! <usney>I am installing transmission now brendyyn <usney>brendyyn my transmission that I installed for transmission doesn't seem to be having any problems <usney>I am able to use the menu without any problems <Blackbeard>should I put the license list in alphabetical order, the game is mostly gpl2 and music has different license according to the song <usney>you mean click the menu and go down the list and click on the item you want? <usney>I can click and hold and go down the list to open the item I want or I can click once move down the list and click again to open the item I want. <brendyyn>ok well it doesnt work for me that is strange <usney>I am using a foreign distro and probably different DE than you are using. <usney>What Desktop environment are you using? <nckx>Blackbeard: That's entirely up to you, as long as it makes some kind of sense. I write them so my comments are readable, e.g. ‘expat ; foo.c, bar.h, and everything in zurp/ | gpl2 ; everything else’. <usney>yes I know what tiling window manager is <brendyyn>yep its basically an i3 clone for wayland <nckx>Blackbeard: Just write your licence comments for maximum clarity to humans, and the order will follow. <usney>I have never used a tiling manager extensively as my main desktop brendyyn so I don't know if it is just an issue with sway or the program its self. <usney>I have used it before though I think long time ago. <brendyyn>Ok so transmission works fine when i launch it in --pure. I guess it really is a theme thing <usney>I use a light weight graphical environment though just not a tiling one. <usney>yes I am not using the themes you are using brendyyn <usney>I really like the themes I am using <usney>I am not sure if guix has these themes or if they are called the same let me get the names. <usney>I am using the themes from apt <brendyyn>Ok we dont have that but we could add it most likely <usney>I use spacefm and nnn file managers <usney>but since they are moving to qt I made the switch <usney>I may switch to just using nnn since I just started using it I really like it so far. <xelxebar>nckx: :( ... Well, thanks for going out of your way to check. <xelxebar>One odd thing. It looks like make built *more* things in the freshly pulled repo than in the original after a git reset --hard <nckx>xelxebar: Want to file a bug? This isn't the standard ‘clean up your repo’ problem so go ahead; there's only so much I can say without seeing the full make output & errors. <nckx>I wanted to ask you: which mail client do you use? Because the copy that landed in my mail box was full of ^M line endings, and it seems to confuse the Web archive as well. <usney>where can I upload an image to show others with a gnu/fsf approved service? <usney>brendyyn can you accept files in your irc client? <nckx>I honestly don't know, usney. <usney>okay I'll see if I cant send what the theme looks like to you. <xelxebar>nckx: Yeah, that message went badly. You can see the User-Agent, but the text itself was written in ed. lol <usney>I am sending it now but looks like you didn't accept it yet <usney>i'll see if there is a mediagloblin instance for images <xelxebar>nckx: I assumed the extra line breaks, shown with a line-final =, were just some lits.gnu.org post-processing or something. <nckx>It's definitely Free though, and fine to promote here. <usney>oh I have heard about there is a recommend free software alternative to flickr but I forget its name <xelxebar>Should I try sending it to bug-guix, including the bit of extra info about the issue you helped me clarify? <brendyyn>guix environment --ad-hoc magic-wormhole; wormhole send screenshot.png <nckx>There's an XKCD about this somewhere. <nckx>xelxebar: TBH I read your message in the HTML moderation interface, didn't check the headers 😛 <usney>it has not approved gnu/linux distros in the download list. I am just trying out lightweight gnu/linux distros to get some ideas and possibly take the config files too to make my light desktop work better or look nicer! :) <usney>that is an image of transmission with the theme I talked about <nckx>xelxebar: Just send a follow-up to your previous mail. ‘Cross-posting’ a bug won't look good. <usney>the icons are for adapta-nokto too but they are by papirus <Blackbeard>nckx: I've decided to name it just "blobwars" that's how it runs on the terminal and that is the name on the URL and makes the package more readable overall <nckx>Blackbeard: That's what I would suggest if there was only one unambiguous blob(-)wars, but you said there were 2? <brendyyn>drainful: are you planning to package runelite <brendyyn>i was wondering what about nonfunctional data? the game appears the same so wouldnt all the textures etc also be proprietary? <usney>transmission has a window border that scrot screenshot app didn't seem to capture brendyyn <brendyyn>also as a part of playing the game i believe it downloads the map data, so i suspect its not truely free software <Blackbeard>nckx: the other is named blobandconquer by the project <nckx>Yes, those are preferred. <nckx>We prefer the ‘project name’, which is (deliberately) vague but usually means the tarball name, or sf.net/project/foo, or github.com/bob/foo, … I really thought there was something in the manual about naming but I don't find much. <usney>i have to be up early tomorrow <Blackbeard>I wanted to start coding a game, a simple one either with chickadee or with sfml <Blackbeard>I was considering sfml or love2d because I can compile it to android <usney>there is a sney in debian we are not the same person just fyi <usney>we have similar nicknames on irc <usney>I registered this nick years ago <efraim>how do I pass enable networking to qemu? something about -net or -nic? <janneke>efraim: something like that; what kind of vm are you running? <efraim>janneke: I'm trying the hurd vm again <janneke>efraim: i'm using this right now: -device rtl8139,netdev=net0 -netdev user,id=net0,hostfwd=tcp:127.0.0.1:10022-:2222,hostfwd=tcp::27528-:17528 <janneke>efraim: note that between the dash and the colon here 10022-:2222, there is an implicit guest ip address <janneke>if the guest uses the qemu default: `10.0.2.15', then you can leave it off <efraim>i'm tempted to add -curses, the qemu window doesn't always work for me <janneke>efraim: i would love it if we could get -nographic going <janneke>i gave it two serious tries, but failed until now <janneke>efraim: if you're on latest wip-hurd-vm, there's a console bug that was reported <janneke>it looks like simply removing the second console (`fancy?') service fixes that -- testing right now <janneke>i developed a blind spot for that ("need to look into that some time...") and always pass --no-offload <efraim>it might be time for me to modify the image. I default to less and which and neither are there <efraim>then I can also add a custom .inputrc and the like :) <janneke>efraim: i think it's OK to add those, and please feel free to push to wip-hurd-vm <janneke>efraim: i even made a fix for less, a couple of weeks ago <g_bor[m]>Could you have a look, and see if it could make it to the release? <Blackbeard>nckx: thank you a lot for helping me so much today <efraim>I was going to see about bumping volk but it seems to pass -march=native in 2.2.1 <rekado_>janneke: I had been working on netdde and the Linux drivers during a few sleepless minutes. The build system is a bit weird, but I’m getting closer to preparing another patch. <janneke>rekado_: ooh...very nice; making the best of what happens, i like that! <efraim>oh, thats cool. a second look at it, it builds the library multiple times, with different optimizations each time <smithras>civodul: I wasn't on IRC last night, but yes the installer gets stuck on the connection screen <civodul>ah ok, but it doesn't actually crash, only nothing happens, right? <smithras>right, it just says 'connecting' forever <civodul>so it could be the issue that Florian reported <smithras>civodul: I have the installer on my thinkpad right now, frozen in the same spot. `ps -aux` shows a running dbus-daemon process <smithras>I don't know much about debugging so I'm not sure what to do next! ***apteryx_ is now known as apteryx
<civodul>smithras: could you check /var/log/messsages? <smithras>civodul: the only two messages I found were the ones I sent in the email, where dbus-daemon activates the net.connman.vpn service <civodul>so connmand is running, just not doing anything? <smithras>civodul: yes, connmand, connmanctl (2 separate processes) and connman-vpn are running <efraim>apparently vim doesn't like being cross-compiled to the hurd <smithras>PotentialUser-70: last I checked it wasn't packaged <smithras>You can still use it on guix via flatpak if you want <smithras>you have to install flatpak, add the flathub repo, and then install eclipse <g_bor[m]>civodul: could you review the latest patchset in this? <g_bor[m]>One part is the git config interference, and the other is to put the test stuff into gitconfig. <rekado_>I wonder if anyone has tried packaging the Eclipse IDE yet <rekado_>we probably miss a few Java packages, but we’ve come a long way <PotentialUser-70>smithras: thanks, but I'm on a debian based distro and eclipse did some error now I fix it <alextee[m]>is there a link somewhere to download an uncompressed ISO of guixsd? <civodul>there are other questions left unanswered :-) <g_bor[m]>civodul: I tried to answer those in the cover letter. <g_bor[m]>Git only respects HOME, so with an unmodified git we can't avoid it. <g_bor[m]>Regading the libraries, the C libraries do not use the environment variables, but I don't now hat the guile ones do. <g_bor[m]>If they aer a thin wrapper they should not be affected either. <g_bor[m]>setting XDG_CONFIG_HOME was not required to make this, work, so I did not do that finally. <civodul>g_bor[m]: oh sorry, i had overlooked the 2nd cover letter <civodul>setting HOME isn't great because it also modifies what we're testing <civodul>does git have a --ignore-config or similar flag? <civodul>we could change (guix tests git) to do that <civodul>or actually, we could simply set the variable(s) there, no? <g_bor[m]>I tried to find an ignore-config option, but could not. <g_bor[m]>completely ignoring confgi would not work by the way, as author name and email are required, and git does not provide any other option to set those. <civodul>i think we could set GIT_CONFIG_NOSYSTEM and HOME and whatnot right above 'invoke' in (guix tests git) <civodul>with dynamic-wind to restore the environment afterwards <civodul>sorry for the delay and for the back-and-forth :-/ <ecbrown>hi all, i'd like to "insert" a simple script (written in GNU R) into a Guix artifact (e.g. a docker image) so that i could call the script at runtime: e.g. ... -- Rscript the-script.R wondering if there's a way to do this in Guix (i confess mostly i've written recipes for building .tar.gz <ecbrown>(i have a channel available for the source code, in case this helps) <civodul>ecbrown: i think you could do that with a "manifest" that you'd pass to "guix pack -m" <civodul>in your manifest, you'd have a (local-file ...) referring to the-script.R <ecbrown>civodul: (hello to you! some time away from guix and dang you all are running hurd.) <ecbrown>civodul: thanks, i think this is exactly what i was looking for <efraim>what's the git reference for a local directory? <civodul>efraim: you can use (git-reference (url "file://...") ...) <efraim>I was thinking more from the command line, like 'git clone file:///path/to/repo' <civodul>--with-git-url clones the thing for you <civodul>you mean you want to build a package from a local repo, right? <efraim>I'm trying to figure out Julia and I'm trying outside of Guix first to even figure out how it works without network access and I want to clone a repo from a directory on my filesystem into another directory <rekado_>yay, Glenn offered us rsync access to debbugs! Unfortunately, I can’t access yet because of the firewall… <civodul>efraim: so you can clone julia and build within "guix environment julia", no? <efraim>I want to use 'guix environment --ad-hoc julia' (with -C? no network) and use local clones of julia packages <efraim>I managed to modify the 'precompile phase to actually precompile and build a package, but I need to figure out anyway how to "install" it for reuse and use it for the next one <efraim>Currently our julia-build-system works for dynamically loading libraries, it doesn't actually compile anything ***rekado_ is now known as rekado
<smithras>civodul: Let me know if you need more info on the wifi bug, I'm going to make lunch now but I'll be checking irc thoroughout the day <civodul>smithras: thanks, i'm going for lunch as well but i'm not sure how to debug this <civodul>maybe that'll have to wait for next time :-/ <efraim>I'm trying to avoid making %out/share/julia a pseudo homedir during the process <rekado>civodul: the rsync access is tied to the IP of issues.guix.gnu.org, so no SSH tunnels this time. <efraim>it's a bit more manual than using julia "as intended", even with no global repository set it still reaches out to the global respository on the internet, making all builds fail <rekado>I already asked IT to make the necessary changes. <efraim>If I can unvendor again the julia zlib package I think I can figure it out nicely *efraim once again dives deeply into languages <efraim>oh, and if anyone is curious, zile cross-compiles to Hurd while nvi, vim and busybox don't <efraim>and nano does. I'm not sure (gnu system hurd) has a text editor currently <janneke>efraim: no...but it has sshd, so => emacs+tramp! :-D <kraai>When I run `cargo build`, it produces the following error: SSL peer certificate or SSH remote key was not OK (server certificate verification failed. CAfile: none CRLfile: none). <kraai>strace shows it reading /etc/ssl/certs/ca-certificates.crt. <efraim>kraai: you need to export SSL_CERT_FILE=$HOME/.guix-profile/etc/ssl/certs/ca-certificates.crt; if you have nss-certs installed <janneke>efraim: it might make sense to document or hint at these possibities somewhere <efraim>I was about to say 'source ~/.profile', but then I remembered I just added it there <efraim>janneke: the GIT ones are added to $GUIX_PROFILE/etc/profile <janneke>i had this 'THE-HURD' readme, but i'm not sure if that was a good idea, and it was about creating bootstrap binaries and the wole debian vm setup which is all eh, not the most important thing right now <kraai>efraim: That didn't work, though. :( <efraim>I have a half-written write-up about hand compiling and using stow to build Guix from source but then Debian got the dependant packages <efraim>janneke: I need to go back to before we got the hurd changes into core-updates to get guile-2.0 working again on ppc, it was working for a while and now it's back to segfaulting <efraim>kraai: do you have nss-certs installed? <kraai>strace now shows it reading /home/kraai/.guix-profile/etc/ssl/certs/ca-certificates.crt. <efraim>did you export the variable in the same terminal window as your cargo command? <efraim>I assume there is actually a file at /home/kraai/.../ca-certificates.crt <efraim>maybe you also need SSL_CERT_DIR? export SSL_CERT_DIR=$HOME/.guix-profile/etc/ssl/certs <kraai>I tried `export SSL_CERT_DIR=$HOME/.guix-profile/etc/ssl/certs`, but that didn't make a difference. <alextee2>having internet problems, not sure if my messages go through, can someone read this? <efraim>i'm running low on ideas, anyone else? <kraai>efraim: Thank you for the suggestions. <alextee2>it looks like it tries to load a kernel module twice or something. any ideas how to fix it? <efraim>tcsh wants 'cc' set when cross-compiling <kraai>efraim: Running `export CARGO_HTTP_CAINFO=$SSL_CERT_FILE` fixes it. Found the answer in the help-guix archives. <kraai>efraim: Should this be changed in the rust:cargo package somehow? <efraim>perhaps. I think most people don't notice it if they have nss-certs installed globally <kraai>efraim: I think I have nss-certs installed globally. <efraim>actually file isn't installed by default <kraai>efraim: Yeah, I just discovered that. <kraai>efraim: `ls -d /etc/ssl/certs` outputs `/etc/ssl/certs/`. <efraim>It might be a good idea to set the searchpath for CARGO_HTTP_CAINFO in rust *efraim can't remember which its vs it's is possessive <reepca`>"its" is possessive, "it's" is contraction ***reepca` is now known as reepca
<efraim>not a good candidate for 'the exception makes the rule' but it is the outlier <dutchie>efraim: "its" is a possesive pronoun, so no apostrophe just like in "his" or "hers" <efraim>dutchie: that's a good way to remember it! never thought of it that way <dutchie>well, i suppose "hers" is not quite the same. but i remember it by analogy to "his" <rekado>divansantana: issues.guix.gnu.org has not been updating for a few days because of unresolved problems with connecting to debbugs.gnu.org. A path forward has been found and issues.guix.gnu.org will soon be updated again. <ecbrown>my rule is, "it is" is it's, anything else is its <efraim>I can't think of a good way to slot it into the conversation, but apostraphe-s is possessive, not "Look out! Here comes the 's'!" <dutchie>ecbrown: "it's" can also be "it has" *ecbrown hangs head in shame <dutchie>but right, it's (heh) always a contraction <dutchie>even native speakers make mistakes with its/it's all the time <ecbrown>its been a long time since i've made that mistake *rekado thinks about starting a new project called “nacre”, which will be a subset of Perl implemented in Guile… <dutchie>professional writers even: just this morning i noticed errors in the Guardian live blog and in the proposal document i'm reading for work <smithras>civodul: Have others been able to successfully connect to wifi? <str1ngs>hello janneke did you ever recover your pinbook pro? <str1ngs>janneke: what was the symptoms again? <janneke>in 8 weeks i managed to exchange 1 email with a zoho support compliance drone and 1 email with a somewhat technical person who asked a question that was answered in my original report; i also tried a tweet the past week <janneke>str1ngs: it's dead; the only sign of life is the red led next to the power plug <str1ngs>janneke: there is an out of band mode. called maskrom. there is a way to put it into maskrom I'll link the URI <janneke>str1ngs: are you suggesting that a (re)flash of the firmware may fix/repair it? <janneke>i was afraid it must be some hardware component that broke <janneke>...but i don't know much if anything about such devices <str1ngs>janneke: I'm suggesting you check if you can get into mask mode. since that could me a recoverable state <janneke>rekado: ...and get perl hackers on board on guile-emacs ;-) <str1ngs>janneke: the reason I brought it up. is I had similar symptoms. no power light. but I was able to get into mask mode and reset. <str1ngs>janneke: no power on light to be specific. <janneke>str1ngs: oooh, that would be something... ***vup2 is now known as vup
<apteryx>civodul: on 1.1.0 branch, avahi-daemon doesn't start anymore, due to a DBus error. You don't have this issue by chance? <janneke>str1ngs: right, and also nothing over the serial console? <apteryx>civodul: hm, good question, let me see if I get the same result in a guix VM with my config *rekado updates R packages again <str1ngs>janneke: it might be in mask mode already. try connecting the usb-c to another computer <str1ngs>janneke: rkdeveloptool is pretty handy rkdeveloptool ld might list the pinebook <janneke>str1ngs: hmm, it's not in debian -- is it free software? <str1ngs>it's kinda esoteric maybe why it's not been packaged <str1ngs>janneke: seems I have brought mine back from a similar state. <janneke>str1ngs: well, who knows -- packaging rkdeveloptool right now <janneke>while reconfiguring me system to 1.1.0 <str1ngs>janneke: once you have that packaged. connect the usb to your computer. then do rkdevloptool ld . and if you can see the device then it's in a maskrom mode. which would explain the no power on light *janneke did configure a worktree when "it" branched but hasn't got around testing it (boo! boo!) <apteryx>civodul: bah, I can't reproduce at the moment in a VM :-( <bricewge>When building a package, does the time is set to the unix time? <civodul>apteryx: i reconfigured my laptop yesterday on master and avahi-daemon is running fine, FWIW <bricewge>I have some tests hanging that are using certificates with validity dates between 2019-2020 <apteryx>civodul: OK! Perhaps some interaction with the services I have (I have nfs-service-type which slows the boot a lot, perhaps it's playing a role) <apteryx>and it didn't run in the test VM (I haven't investigated why yet). <bricewge>civodul: Just a package, trying to update libtorrent-rasterbar <civodul>bah, "guix build hello -d --no-grafts" takes 9s on my olinuxino <civodul>quite a lot of cacheflush(2) calls, dunno where they come from <mbakke>bricewge: you can use "faketime" or "datefudge" to ensure the tests run with a specific date <bricewge>mbakke: Thanks. Does guix really set the time to the epoch time when building a package? <mbakke>bricewge: Guix does nothing special when it comes to time apart from setting the SOURCE_DATE_EPOCH variable. <bricewge>mbakke: Thanks for clarifying it. So the issue should come from elsewhere... <janneke>str1ngs: well, the good news is: we have a new package :-) <janneke>...=> no such devices...but i'll try again in a while *janneke was paraphrasing: "not found any devices!" <bricewge>After rereading the service extension part of the manual again, it's still not clear to me if I need to instantiate each extended service a service use. Or are they pulled in automatically? <civodul>perfect trick for those who found downloads were too fast <civodul>NieDzejkob: can you run "guix repl", and then ",use(gnu packages toys)" <joshuaBPMan>hey #guix, does the getmail-service do synchnization of the maildir and remote mail server? <rekado>NieDzejkob: could it be that the hash is invalid? <rekado>joshuaBPMan: I don’t know getmail but the website indicates that it does support IMAP to local Maildir <joshuaBPMan>rekado: But will it syncronize that local Maildir to the remote IMAP server? <rekado>it just fetches from a remote server <NieDzejkob>rekado: I don't think so, I just checked that I've copied the output of guix hash correctly <NieDzejkob>weirdly enough, the text "Value out of range" doesn't appear in the source code of guix and guile <civodul>NieDzejkob: actually, (use-modules (gnu packages toys)) may work better <NieDzejkob>ouch, I grepped through guile-git instead of guile <NieDzejkob>civodul: use-modules also doesn't give a backtrace :/ <NieDzejkob>Hmm, the error seems to come from either scm_i_range_error or check_array_index_bounds <civodul>NieDzejkob: i have another idea: try (resolve-module '(gnu packages toys)) <civodul>i swear you'll get a backtrace this time :-) <NieDzejkob>Indeed, after a ,bt the problem became clear: I swapped commit and revision in the arguments to git-version, which passed an out-of-bounds index to string-take ***fr33domlover1 is now known as fr33domlover
<nckx>Blackbeard: My pleasure. <alextee2>how do you keep debug symbols in gnu-build-system? set #:strip-binaries? to #f? <NieDzejkob>so, why does (raise (condition ...)) error out with "Wrong type (expecting exact integer)"? <NieDzejkob>alextee2: Sounds right, did you try it and it didn't work, or what? <str1ngs> NieDzejkob keep be its float not an integer? <str1ngs>NieDzejkob: or something is out of bounds <rekado>NieDzejkob: hmm, I have seen something like that before! <rekado>when I first saw that I thought to myself that this may perhaps be a consequence of unifying the exception types in Guile. <NieDzejkob>`genv --pure --ad-hoc guile -- guile`, then ,use(srfi srfi-35) and (raise (condition (&message (message "test)))) <civodul>NieDzejkob: beware of backtraces in the presence of tail calls <civodul>perhaps the last frame results from a series of tail calls <nckx>Blackbeard: I see your patches have been merged already. Congrat. <efraim>busybox-1.31.1 also doesn't cross compile on core-updates <raghavgururajan>Folks! So I have added udisks-service-type to my config .scm. How do I make it automount devices? <bricewge>I can share what I tried few months ago if you want. <mbakke>hm, LC_TIME does not seem to affect `date` output; LC_ALL works however. <mbakke>I guess LC_ALL overrides the more specific LC_TIME <bricewge>raghavgururajan: I use udiskie, a cli frontend udisks2, for auto-mounting devices <raghavgururajan>bricewge Ah I see. Do you have to open udiskie, everytime you insert usb stick? <rekado>raghavgururajan: for me it was enough to add gvfs to the global list of packages. <rekado>(this may no longer be necessary) <bricewge>Nope I just get a notification that something has been mounted <raghavgururajan>bricewge Intresting, udiskie is cli front-end right? Did you initialy change any settings? <bricewge>I just run udiskie at startup (using a shepherd user instance) and that's it <bricewge>No settings. The one thing I'm missing from it is MTP support <lfam>I'm having trouble on Guix System with the /run/user/NNN directory not getting created again <lfam>I've had this issue before <raghavgururajan>bricewge Do you know how to create service definitions? I can package udevil. Would you be able to define udevil-service-type? <lfam>I was adding and removing users yesterday but my own user did not get removed <lfam>It seemed to break the user setup stuff <nckx>lfam: I don't have that directory. Is it a dbus thing? <lfam>The main symptom is that pulseaudio can not start <nckx>mbakke: Yeah, LC_ALL is the ‘do what I want, now’ hammer, LC_TIME is the nicer way to ask. <nckx>raghavgururajan: Did you have questions for me about git? <lfam>Actually, I think it should be created elogind <bricewge>raghavgururajan: I think udevil is ought to be a user service. There is no user service support on guix ATM <apfel>hi there, i want to use virt-manager to manage libvirtd on a remote guix installation. This works via SSH and requires netcat-openbsd to be in PATH. But the remote command is not run with a login shell. I now source /etc/profile in .bashrc. Is there another to do this in guix? Maybe a global symlink to /bin? I did find something like this for guix pack, but not guix system <bricewge>raghavgururajan: So you should probably just use a shepherd user instance. Just replace udiskie by udevil in the file I shared <raghavgururajan>nckx Oh shoot! Yes, I will have to email them to you. I just dived into packaging udevil. <lfam>Is there a way to see derivations of services related to the current system? *lfam roots around in `ps` <nckx>apfel: You can probably hack around it with (extra-special-file "/bin/nc" (file-append netcat-openbsd) "/bin/nc")). <apteryx>lfam: guix gc -R /run/current-system <apteryx>it's not derivations though... but is this what you were looking for? <nckx>☝ is also what I use, no idea how you'd easily get the .drvs. <bricewge>raghavgururajan: Then use devmon instead, defining a naked shepherd service is easy. <apteryx>lfam: perhaps 'guix system build --derivation your-config.scm' would give you a starting point. Then open that file in Emacs (with emacs-guix it should look nice). <lfam>Urgh, I see myself complaining about this issue in the IRC logs from 2018 <lfam>Is anyone else using user Shepherd services on core-updates? I'd just like some confirmation that it's only me <lfam>That it's only me having this problem <bricewge>lfam: Did you managed to get the derivation? A mix of strings, ps and grep should get you there. <lfam>The code to create /run/user is there, and succeeds <lfam>But elogind is not working right :/ <lfam>I'm not sure if this is something controlled by Guix or if elogind is supposed to do it automatically <lfam>And yes, elogind is running <apfel>nckx: did not work, but PATH=/run/current-system/profile/bin is the default, so adding the package to the system profile helped <nckx>I didn't know it was already patched. That's probably as good as it gets. <lfam>When booting and logging into an old system generation from April 6, the /run/user/NNN directory does get created as expected <lfam>But then I have another issue :( Starting Shepherd for my user, I get "In procedure stat: No such file or directory: "/run/user/1000/shepherd"", and then the /run/user/1000 directory is deleted <Blackbeard>I don't understand why the indentation is off in my patches <Blackbeard>I always use etc/indent-code.el before the commit ***drakonis is now known as drakonis_
***drakonis1 is now known as drakonis
<lfam>I feel like I've had all these issues before and they were resolved as mysteriously as they arrived... <nckx>Blackbeard: Can you give an example of the bad indentation that came straight from your emacs (or indent-code.el)? <nckx>Blackbeard: Does C-M-q (indent-sexp) do better? 's What I use. <Blackbeard>I'll try to do that one instead, seems easier to press too <nckx>Blackbeard: Here, C-M-q correctly moves ‘(name’ from ‘( under p’ to ‘( under a’ in that link. <nckx>If you accepted Guix's indentation rules (!, as we discussed yesterday) I don't see why yours would do any different. <Blackbeard>nckx: I don't understand :( but I'll try to be more careful <TZander>just throwing a silly idea in there ;) <nckx>Blackbeard: Yesterday, you asked me about ‘unsafe variables’. If you accepted them, emacs ought to be configured to follow Guix's indentation rules. <Blackbeard>Ohh licenses should be determined by the header file not the license file <guix-vits>TZander: "indent-tabs-mode: nil" in games.scm <nckx>In the mail you linked above, having the ( in (name directly under the ‘a’ in (package is correct, and that's what C-M-q (with point at ‘(define-public emacs-org-roam’ does here. Does it do that for you? <Blackbeard>nckx: I wrote most of my patches before accepting it <Blackbeard>Anyway I'll be more careful. Seems like I still have a lot to learn <Blackbeard>nckx: oh I am on my phone. Give me 5 minutes to turn on my PC. <nckx>Blackbeard: Yes, you should always check licence headers. Unfortunately. If one file in the source tree lacks the ‘or at your option, any later version’ wording it doesn't matter what the rest (or COPYING) say, it's GPLvN-only. <nckx>Blackbeard: Life hack: whenever patching anything, check your work against Debian's ‘Copyright File’ (it's in the right-hand column of pages like these <https://packages.debian.org/sid/sl>). They generally do their homework (but do your own too). <nckx>If you use DuckDuckGo it's as easy as ‘!deb <package name>’. <nckx>Blackbeard: Sure, but they are just the ‘licence text’, not the actual ‘licence statement’. Confusing, I know, and there's a grey area since many projects don't have explicit headers. <nckx>Licencing is a mess and sadly things like GitHub aren't encouraging better practices. <Blackbeard>In México (I think) the LICENSE file would be the one that the judge uses. Because here when you give permission to use authorship rights you gotta put it in a contract <Blackbeard>But it is a gray area too. Legally you should have the signature of the author to be able to use the work <crab1>I feel like this is probably a oommon problem, but I'm trying to use sway via sddm, and it opens to a blank screen with the standard X shaped cursor and no keyboard shortcuts do anything, I do have the default sway config in ~/.config/sway/config, I'm not sure what's going on here and I'm not sure how to troubleshoot, anyone have any direction? <Blackbeard>The only thing you could argue is that it is a common practice in the software world to use this type of LICENSE files <crab1>cwm is working fine, I have not tried anything that uses wayland so I'm not certain whether it's a wayland issue or a sway issue <crab1>but it should not be a sddm issue <guix-vits>crab1: do you've an old i3wm config in your ~/ ? <rekado>issue #40629 lets you install packages from JSON definitions. Sadly it’s too late for April 1 to announce that Guix drops Scheme and will be using JSON going forward. <nckx>Blackbeard: Ideally, the author says ‘this is my software, and you can distribute it under <terms>, for details see <COPYING>’. That's what a licence header does. If it's missing, the mere presence of a generic licence file is certainly a sign of intent, but it's unpleasantly implicit. ‘I opened the box and there was a copy of the GPL, I guess the contents of the box are all GPL?’. But you don't have that on paper! <nckx>Luckily most judges are humans, but still. <guix-vits>crab1: and it's a blank screen, no "wallpaper"? <lfam>Sounds like it's working! ;) <crab1>nckx: isn't the preferred method to put a license header in each file you wish it to apply to? <Blackbeard>nckx: yes, I agree with that. The problem is obsolete copyright laws. <lfam>Does sway require i3status and dmenu, too, crab1? <lfam>On Guix, you have to install i3-wm, i3status, and dmenu to get something that works <lfam>You may also need to tell i3 / sway which terminal to use <nckx>crab1: Licences like the GPL are too long to include in each file, hence COPYING, but each file should have the blurb. <lfam>And install that terminal <crab1>I have not tried to start sway from tty yet <crab1>lfam: the default terminal it uses is urxvt, which I have installed, but clearly it is not using the config I have in ~/.config/sway/config, it seems no keybindings are configured whatsoever <usney>qbittorrent still works great! <lfam>crab1: It may not be using the modifier key that you expect <lfam>(I've never used sway but these are all common pitfalls with i3) <crab1>I have tried the modifier in the default config, and then every other modifier or modifier I can muster, and played jazz piano over the keyboard while holding those modifier combinations <guix-vits>crab1: one can start sway from tty with single command: `sway`. the Default modifier is "WinKey" (Win + Enter = terminal) <crab1>I prefer to have a display manager though <crab1>I can try tty and see. Would I have to remove sddm from the config before trying it from tty? sddm seems to occupy several of the ttys <guix-vits>btw, one can open multiple sway-sessions from different tty... <usney>is sway easy to use if one has never used a tiling manager? <usney>is it lighter than openbox plus tint2? <Blackbeard>usney: I never felt as comfortable with I3 as I feel now with StumpWM <Blackbeard>Sway never worked for me. And I've not tried it in a long time. <Blackbeard>usney: it is a window manager written in common lisp. <crab1>It seems that sway was launched, I got the same visual as before except my mouse cursor was no longer the "X" and the background was the same as whatever tty I was last in. I did more key mashing and nothing happened, but I went to another tty and a couple instances of urxvt appeared in top, I'm not sure what's going on here <crab1>I even saw the mouse switch to the "I" that indicates you can click here to type stuff, but nothing else showed up <Blackbeard>usney: there is exwm that uses emacs as a window manager <guix-vits>crab1: pity. did you tried to move .config/sway/config out? <crab1>also I did not try passing config as an argument, that could be done <crab1>sway -c .config/sway/config could be worth a try <crab1>what would be the value in mving it? <crab1>it is possible, but unlikely since I have the default config <lfam>raghavgururajan: Please share your config.scm and the `guix describe` from where you ran the failing command <lfam>raghavgururajan: Never mind about that... what command did you run? <lfam>Whatever package you are building has some broken build scripts <lfam>It's Makefile (or equivalent) is not portable (you might call it broken) <lfam>There's no root user in the build environment <lfam>Nor should it try to install things for a particular user <lfam>Yeah, that will probably fix it <lfam>Weird that it tries to install as root, however. It seems to use a full Autotools setup, and it should "just work" <hacker>now sway works perfectly if I remove sddm ***hacker is now known as Guest49747
***Guest49747 is now known as crab1
<nckx>raghavgururajan: You can just substitute* " -o root -g root " to " ". It's not as uncommon as it should be. <crab1>thanks for the wayland assistance <lfam>Weird that dmesg includes "shepherd[1]: Service ssh-daemon could not be started." and then a few seconds later "shepherd[1]: Service ssh-daemon has been started." <lfam>I wonder if we can remove the advertising messages from ntpd's output <lfam>Can somebody try logging in to Guix System over SSH and tell me if there is anything in '/var/log/messages' from elogind or dbus? <lfam>On my system, the only messages are from sshd <lfam>I figured out why things broke for me. I had removed my custom '~/.profile', and that file set a custom value for XDG_RUNTIME_DIR. I still think this should work out of the box without hacks like that <lfam>It would be great if somebody with a working Guix System could tell me if elogind does anything when they log in and if they have a directory at /run/user/NNN <Blackbeard>lfam: I do have messages from both elogind and dbus <lfam>Thanks Blackbeard. Are you using %desktop-services in your config.scm? <lfam>I am using %base-services but I added elogind-service and dbus is pulled in by some other service I have <lfam>I guess that is not working anymore <raghavgururajan>nckx For the above package ^, is it possible to create service-type? <reepca>huh, so I've discovered that our configure script will consider only the highest-version guile "development files" installed, and will only consider the first guile binary it finds in $PATH. So if you have guile 3 installed and you run 'guix environment guix', it will put guile-2.2 at the front of $PATH but still find guile-3.0 development files. It will then complain about the mismatch. Weird. <cbaines>guix search mingw has some results for me <alextee[m]>otherwise it means i have to build every dependency manually when building something <cbaines>Ah, right, so normal Guix packages built for mingw as a system/target <cbaines>I don't know much about that, I remember dongcarl doing some stuff with mingw around bitcion <alextee[m]>oh hmm maybe i can just pass --target=x86_64-w64-mingw32. i will dig around, thanks <lispmacs[work]>so, was just wondering... if there was a hash collision one day, would the whole Guix project vanish suddenly in a cloud of smoke? <alextee[m]>anyone know where the <linux/errno.h> header is? <alextee[m]>maybe i need to have linux-libre in my profile for it to be found <reepca>well, if two store items end up with the same hash-part, it's not automatically terrible, since /gnu/store/aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-foo-1.0 and /gnu/store/aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa-bar-1.0 would still be distinct. They'd need to have the same basename ("package name"). In which case, assuming the collision was random and not malicious, the behavior likely wouldn't change drastically - it'd be like some package being <reepca>randomly overwritten with a newer/older version <reepca>well, any danger caused by it would be of the "subtle bugs are terrifying" variety, not the "dividing by zero causes automatic explosions, right?" variety *civodul is very much into GIFs these days <bandali>i like animated things, but would personally prefer an actual video that i can control (pause, rewind, etc) over a long GIF <reepca>I think it features a very proficient typist who can be a bit difficult to follow <bandali>also, the "arabic" language in the first frame i think is messed up <civodul>ah there's a bug about RTL languages <bandali>it's rendered left-to-right rather than right-to-left, and the letters are all separated <Blackbeard>I think it serves a purpose to have a quick idea of how the installer works <Blackbeard>Without having to watch the full video as bandali said <mbakke>civodul: maybe webm instead of gif? <civodul>i've uploaded a new GIF with doubled delay <bandali>but yeah directly using ffmpeg should be possible too <mbakke>civodul:how did you record it? perhaps there is an export option for webm. <civodul>mbakke: i hacked something with QEMU :-) <civodul>so QEMU gives me ppm files (bitmaps) <civodul>icecat doesn't want to play it directly tho <bandali>is it on a different origin than the page itself? <bandali>one of icecat's extensions (third party request blocker) blocks such requests by default <civodul>no, i think it's just the server no announcing the right MIME type or something <raghavgururajan>bricewge Is it possible to add an option like `enable-udiskie? boolean` to udisks-service-type? So, if value is `#t`, udiskie will installed ran along with udisks. <civodul>it's slightly smaller than the GIF but also blurier <civodul>thing is, we probably can't include that in Texinfo <civodul>that's also why i was going for a GIF <mbakke>the slower GIF seems fine, I mentioned webm because it is typically smaller and gets the browser "video controls" so it can be paused <civodul>then the GIF is 2.8M and we probably don't want that in our repo <civodul>maybe we can link to the GIF or webm from the announcement? <reepca>fwiw attempting to open the webm manually in icecat after it's downloaded causes it to show "Video can't be played because the file is corrupt." <mbakke>could we get away with an <origin> or external texinfo reference? <mbakke>I mean like a Guix origin record :P <civodul>i was thinking of using Texinfo's @image <mbakke>linking it in the blog post and release announcement seems like a no-brainer though :-) <civodul>the goal was to give people an overview of the whole process <civodul>so they can really get a feel of what it's like <mbakke>yes, I think many will appreciate that <civodul>i know i would appreciate such a thing <mbakke>so I guess tomorrow is the big day? :-) 
