<brendyyn>anyone else find chromium really laggy? it feels like maybe some gpu rendering isnt working and its fallen back to cpu. even moving the mouse around in the chromium window is laggy <vagrantc>the emacs-realgud description is hard to fix... *vagrantc moves on to easier targets <mbakke>brendyyn: is that with version 86? <xelxebar>Aw... I have a fat12 partition with an 8-char uuid like "abcd-ef01" and guix system tells me that it's invalid :/ <xelxebar>Aaaand... it even fails to find it with file-system-label <xelxebar>Just to be clear, using "/dev/disk/by-uuid/" or "/dev/disk/by-label" works as expected. I guess fat12 is confusing guix? <brendyyn>xelxebar: did you put the type in the (uuid ...) field? <xelxebar>brendyyn: Yes, (uuid ...) and (file-system-label ...). <lfam>sneek: later tell raghavgururajan: I was able to build that linux-libre 5.4.72 derivation successfully for x86_64. It's unfortunate that the build farm isn't building it, but I don't think the package is broken <sneek>raghavgururajan, you have 1 message! <sneek>raghavgururajan, lfam says: I was able to build that linux-libre 5.4.72 derivation successfully for x86_64. It's unfortunate that the build farm isn't building it, but I don't think the package is broken <nly>time emacs-27.1 (guix-all-packages) (quit-emacs) -> 48 seconds <nly>time emacs-28.0.50 (guix-all-packages) (quit-emacs) -> 34 seconds <nly>yeah i'm switching to emacs-next <apteryx>nly: wow, impressive performance gain! <apteryx>NetworkManager does seem to really want a reboot to get network-manager-openconnect working... bah. <roptat>how do you suspend a computer from the CLI? for some reason the internet is telling me to use systemctl suspend (which won't work for obvious reasons) or pm-suspend which I can't find a package for <roptat>and closing the lid doesn't work anymore for some reason... <apteryx>I think there was something called 'pm-suspend' <apteryx>brendyyn: eh, perhaps things are better for Intel, but at least with nVIDIA (nouveau) I wouldn't expect much video acceleration; this is often only supported by the gpu drivers when their binary firmware blob is loaded, helas. Modern AMD is even worse in my experience (even the OpenGL/3D support is missing if you're lucky it doesn't blackscreen at boot without the firmware blobs). <ryanprior>Does Guix support tags or categories for packages? <ryanprior>I hadn't seen any so I assumed no, but I'd assumed that about maintainers too but then Guix actually does support a maintainer field and we don't use it <apteryx>I had never heard of such a field. It doesn't seem to be in the public package interface (guix packages). <apteryx>I don't see a 'maintainer' field on the package record. <apteryx>ryanprior: interesting. It's probably a relic. <apteryx>especially given your last finding :-) <ryanprior>gnu-package-descriptor records aren't used anywhere though <xelxebar>Say I have gnu/store on a usb and would like to copy items from /gnu/store onto the usb. Is there a guix command to do this? guix copy assumes you want to use ssh <xelxebar>Would it make sense to just cp -r the directories over? <xelxebar>apteryx: Ah, I considered that, but was unsure whether it makes sense in this case. <apteryx>You want the guix database /var/guix/db/db.sqlite to know about the entries under the /gnu/store, so you cannot simply 'cp -r' them; they wouldn't be considered. <apteryx>'guix import' on an archive produced with 'guix archive' should register them correctly <apteryx>my guix database file weighs 487M :-D <xelxebar>Blah. Having issues setting up a guix system with luks. <xelxebar>guix system init works. Grub decrypts and loads the kernel and initrd just fine, but for some reason the initrd doesn't ask for a password to decrypt the luks device and fails to find the root partition... <apteryx>are you sure it's not obscured by log messages? It's not always super clean. <apteryx>You basically have to guess there's an input waiting sometimes. Perhaps try blindly entering your passphrase. <xelxebar>Hrm. It complains about not finding the root filesytem a bunch of times and then drops into a recovery repl <xelxebar>Will that happen even if it's waiting for input? <xelxebar>Yeah, it doesn't seem like there's any prompt waiting. Blindly typing my password in while it complains about not finding the root fs doesn't do anything, and eventually all those keypresses get sent to the recovery repl. <apteryx>I see. Perhaps you've made an error in your system config with the root partition UUID? <xelxebar>Well, I'm using (file-system-label "root") and have successfully mounted the same filesystem with mount LABEL=root /mnt <xelxebar>Also, since grub successfully decrypts the luks device, doesn't that mean that the guix config is correctly resolving the partition? <apteryx>I don't think GRUB needs any UUID to detect the whole drive is LUKS encrypted. <apteryx>GRUB could successfully decrypt the drive then the initrd would go and try to locate the partition on it using a wrong UUID and fail <xelxebar>Hrm. Looking in grub.cfg, it has `search --label --set root`. And we're successfully loading bzImage and initrd.img from there... <xelxebar>apteryx: Are you saying that my *luks* uuid might be incorrect because grub is autodetecting the luks volume without a need for said uuid? <apteryx>I'm saying that the source field of the mapped-device corresponding to yoru root file system may be invalid <xelxebar>Sorry, yeah. I was originally misunderstanding you. <xelxebar>I just copied the output of `cryptsetup luksUUID /dev/foo` into my config.scm and compared the UUIDs side-by-side. Unless my eyes are broken, the look the exact same... <xelxebar>Ideally, I'd like to debug this from the recovery repl, but my guile skills are way too meagre for that... <apteryx>eh, it's not easy for anyone, I'd guess, with the lack of guile-readline there. <apteryx>I don't think so; the init is Guile :-) <xelxebar>Yeah, from a purist standpoint, that makes the most sense ;) I have used `,L bournish` to drop into a semi-shell before, but for some reason it's saying that language doesn't exist <apteryx>perhaps you could start bash by changing the init from GRUB but I doubt that'll be easier from there. <apteryx>It has guile-static-stripped and the raw-initrd itself. <apteryx>I'm off to bed, I hope you'll find the problem! <xelxebar>apteryx: Thanks for you input! Good night :) <xelxebar>Oh... blah. `,L bournish` doesn't work, but `,bournish` does. This is documented in guix/build/bournish.scm... <xelxebar>Ah! I forgot to include (dependencies mapped-devices) in my file-systems! <xelxebar>Does guix system reconfigure let you specify a root directory? Or should I just do a full guix system init again? <sneek>I think I remember zimoun in #guix 10 hours ago, saying: oui oui tout le débat… pour moi, il faut décider pre- ou post- et s’y tenir. C’est à dire corriger quand une incohérence est trouvée, comme une typo en fait. (-:. <xelxebar>Yay! My new laptop has now successfully been christened with a guix system! <abcdw_>Guys, can you share your configs please? Need some source of inspiration) ***apteryx_ is now known as apteryx
<xelxebar>Still need to configure networking, setup services, install software, etc. <abcdw_>xelxebar: cool, thank you. Also looking at btrfs setup with disk encryption, seems like a good choice. Especially because Guix doesn't support LVM yet. <xelxebar>leoprikler: ibus-anthy... do you happen to live in Japan?? <xelxebar>I never quite understood the function of the name and comment metadata on user accounts. Do they get used in userland tools anywhere? <leoprikler>xelxebar: name is your user name, e.g. xelxebar@hostname <xelxebar>g_bor[m]: That makes a lot of sense. I have just used startx and bspwm for ages, so that usage definitely has escaped my notice. <leoprikler>comment is used as the display name in login screens, like g_bor said <leoprikler>Few people know it, but Europe/Vienna is actually part of Japan ;) <xelxebar>Oh, lol. Yeah, my brain did unfunction. Of course. <xelxebar>Hey, it could have been the exact opposite. You live in Japan but refuse to conform to be colonized by JST :p <xelxebar>leoprikler: You should come here. We can hold a Guix convention for all the ~4 people in Hokkaido that know about Guix. <leoprikler>It wouldn't be "colonized" though. I'd actually embrace colonization through Europe by conforming to Vienna. ***iyzsong- is now known as iyzsong
<leoprikler>Where I live, that means "I'm still in lesson 1." <xelxebar>Oh, okay. I feel you on that one. Anyway, if you ever end up making plans coming here, I wouldn't mind helping out a fellow guixer :) *botsnack understands the deep meaning of that sentence. It means he does not have the proper fonts installed <xelxebar>botsnack: No, Japanese typography is actually just a bunch of boxes. What matters is the subpixel kerning between them ;) <xelxebar>Also, you must get highlighted all the time for irrelevant posts. lol ***botsnack is now known as divoplade
***amfl_ is now known as amfl
<xelxebar>What you all use for wireless configuration? <brendyyn>xelxebar: ive heard unicode includes every combination of hangul even though its a simple composable alphabet <brendyyn>do modern applications make use of that though or use the composable versions? <xelxebar>Looking at wpa-supplicant-configuration, apparently it wants an external config file. I guess this is because passwords are in the wpa_supplicant.conf? <leoprikler>IDK but I think compression might be better with the non-composable ones. <xelxebar>Speaking of... is there a way to get cjk in the linux console? <brendyyn>yes, with kmscons. thats what the installer uses <brendyyn>it also allows better multiseat support. i wish it was continued <brendyyn>google also made another one for their chrome book <brendyyn>Funny we have a commit "circumvent RIAA censorship" <leoprikler>As long as we conform to article 13, everything should be fine. <leoprikler>In other words, our CI needs an upload filter ;) <xelxebar>Crud. I forgot to install wpa-supplicant on my new, shiny guix system... <xelxebar>Rather, I forgot that I needed *some* internet connection in order to contiue configuring the system :/ <xelxebar>Is there really no way to do a reconfigure from the live os? <leoprikler>We have yet to find someone, who manages to run Guix from chroot <leoprikler>Well, I rather wanted to point out, that you're not the first one, who had the idea ._. <xelxebar>Yeah, I spent 20 minutes or so looking for options, but for some reason my brain never thought to search chroot. <xelxebar>Also, even reading the docs for --chroot-directory, I'm not confident that this does what I want, since there's also the build chroot in /tmp. <leoprikler>You'd have to mount your system, then manually `cd /mnt; chroot /mnt`, but how to continue from there is an open question <xelxebar>Um... now I can't tell if you're being sarcastic or straightforward :/ <brendyyn>nly i installed emacs-next and i get this Warning: Lisp directory '/home/b/.guix-profile/share/emacs/27.1/lisp': No such file or directory <leoprikler>It really is an open question, many have tried and failed. <brendyyn>oh i think i have to log out and back in again <leoprikler>ahh, but the old values may still be stuck in there <xelxebar>leoprikler: Ok. I read too much into jonsger's smiley. I'll revert to taking up the chroot challenge. :) <leoprikler>Not sure, but I think they might have been the last ones (chronologically, up until now) to publicly report their failure. <xelxebar>Uh... I got hello to build in a chroot... <jonsger>how do I change the keyboard layout in sway/wayland. does it goes over xorg-configuration <brendyyn>jonsger: it needs to be set in the config that sway reads <brendyyn>so, in the same way it would be done on any other distro, in ~/.config/sway/config <xelxebar>leoprikler: Hrm. As far as I can tell, guix is working just fine in this chroot. <sneek>wleslie was in #guix one minute and 26 seconds ago, saying: sneek: botsnack. <ani__>how shall I use git send mail inside guix environment. <ani__>guix environment guix --pure --ad-hoc help2man git strace vim nano send-email <mfg>ani__: you need git:send-email afaik <ani__>I run above command, but I am getting error <ani__>yes because previously I used git send-email after getting out of the environment <ani__>so now that I won't have to exit <ani__>mfg: Also, my git global configs which are stored in ~/.gitconfig do not work under guix environment. Why so, I mean I have to define editor inside guix/.git/config. <ani__>what shall I do to overcome this problem? <mfg>Hm, so you mean that your global gitconfig doesn't work when you ar einside an environment? <roptat>if you use -C or --container your .gitconfig will not be visible because it's not in the current directory <roptat>otherwise, it should be visible and git should be able to read it... <mfg>i don't use them that often, but depending on the directory in which you started this environment, you might need to expose your ~/.gitconfig <mfg>otherwise it's not there <roptat>if you're not in a container, then maybe try and use strace on git: strace -f git fetch 2>&1 | grep .gitconfig <roptat>that should let you see what syscalls are made related to .gitconfig, so you will see what git is trying to do <roptat>or maybe your $HOME is unset or set to an incorrect path? ***jonsger1 is now known as jonsger
<mfg>jonsger: Hi, does sway work with your monitor? ***jonsger1 is now known as jonsger
<xelxebar>sneek: later tell leoprikler: In case you really were serious about people having problems with chroot. Here is an outline of what I did to get it working: http://ix.io/2CAm <xelxebar>sneek: later tell leoprikler: With that, I was able to guix system reconfigure and get my wireless working :D <leoprikler>Nice, now write that as blogpost or cookbook entry. <sneek>Welcome back leoprikler, you have 2 messages! <sneek>leoprikler, xelxebar says: In case you really were serious about people having problems with chroot. Here is an outline of what I did to get it working: http://ix.io/2CAm <sneek>leoprikler, xelxebar says: With that, I was able to guix system reconfigure and get my wireless working :D <xelxebar>leoprikler: Oh, cool. I had no idea about guix-cookbook! <xelxebar>I guess adding a cookbook entry just amounts to submitting a patch to the repo? <xelxebar>How does one go about submitting a blog entry? <apteryx>xelxebar: you could contact maintainers at guix-maintainers@gnu.org with your blog post proposal. Note that there's a tradition of having the blog posts Guix-focused and technical rather than personal. <rekado>xelxebar: send a draft to guix-devel@gnu.org and we’ll review it <rekado>we used to use a special blog email alias, but we’re using guix-devel for that now <rekado>it’s fine not to write it as a patch to the repository if you’re not familiar with that. <rekado>just sending a markdown draft would be fine <xelxebar>Cheers! I'm throwing putting this in my todo list right now ***maav is now known as miguel`
***miguel` is now known as maav_
***maav_ is now known as maav
***chrislck1 is now known as chrislck
<roptat>mh... I'm trying to understand how package-provenance works. From the repl, when I do ,use (gnu packages gettext) and ,use (guix describe) then (package-provenance po4a), I get a result. When I do the same with a package from a different channel, I get #f <nly>hi rekado, where can i find your guix repository for guix user services <civodul>roptat: it looks at info found in the profile's manifest <civodul>(since that's something we discussed earlier) <roptat>civodul, I don't think so, I don't have po4a installed for instance <roptat>looking at the code, it seems to first try to find where the file comes from: (search-path %load-path file) (where file is (location-file (package-location po4a))) <roptat>that returns a path in /gnu/store for po4a, but in ~/.config/guix/current for any other channel <roptat>which makes it fail the (string-prefix? (%store-prefix) file) test, and return #f <roptat>even though readlink -f shows that that path is ultimately in the store <civodul>so it looks specifically at the manifest of the profile of the 'guix' command <civodul>(then "guix install" propagates that info to ~/.guix-profile/manifest) <roptat>there seems to be only one clause, and it will always return the main repository (guix) <civodul>the 'or' is a typo/leftover i think, because (or x) == x <roptat>ha nevermind, apart from the useless or, it looks good <roptat>is there something similar to readlink in guile/guix? <janneke>that's not similar, it's identical ;-/ <roptat>er... nope :/ In procedure readlink: Argument invalide <roptat>I'm passing the location of the file, as a string <roptat>but the file itself is not a symlink, one of its parent directories is <civodul>it's a direct binding of libc's 'readlink' <civodul>(use only when you have to, because it's costly!) <PotentialUser-83>Hi! Is it possible to have an encrypted boot partition with Guix System? <maav>PotentialUser-83: i'm not sure, but i think that an encrypted boot without encryption for root would work, but the use case for both encrypted in different partitions isn't implemented... yet <roptat>you can have an encrypted root partition, with no separate boot partition <maav>that too, thanks, roptat :) <maav>civodul: maybe yesterday it was lost on the logs... should i update po/packages/POTFILES.in? i used "find src/{packages,services} -name '*.scm' | sort" to populate it, and we have around 20k+ messages missing :) <civodul>maav: ah, well, you can add a few more files <civodul>initially the TP coordinator was not super enthusiastic about having all the descriptions translated there <civodul>so they halfheartedly consented to have "some" of that work at the TP <maav>yup, i can understand, tp robot can go full skynet with each update... <civodul>so i don't know, i guess you can add a few important files there <civodul>typically files that contain applications <maav>i think that at least the services would be helpful, as they have more "daily work" involved <PotentialUser-83>So I must have / and /boot on the same partition, can I have /home in another? Btw I'm not talking about LVM, just differents gpt partitions <maav>PotentialUser-83: yes, you can have /home on a separate one, but for now / and /boot should reside in the same one <maav>PotentialUser-83: i think that isn't implemented yet, as I don't see how to provide that option to mapped-devices... patches are welcome :) <maav>civodul: i'm going to be afk for some hours, later i'll check how big is with the only services to not open any scar at that front, even though i don't understand any reason other than the skynet hypothesis :) <maav>(we all know that the robot hates us, but i belive it has good bits inside) <civodul>i never interacted with the robot, only with its coordinator <divoplade>Hello, why does guile-rdf depend on guile 3.0.2 and not 3.0.4? <roptat>mh... good question, I might have made a mistake when adding it <roptat>actually no, I think every package depends on guile-3.0.2 <roptat>because that's guile-3.0, as opposed to guile-3.0-latest <divoplade>Is there a breaking change between 3.0.2 and 3.0.4? <divoplade>Because maybe it would be better to just use 3.0.4 everywhere <mfg>is this updated, next time core-updates or staging get merged? <ani>mfg, roptat: will take that into action. <ani>does GNU has some community specific questions for outreachy applicants? <ani>It has been asked in final application. <sneek>zimoun?, pretty sure was seen in #guix 19 hours ago, saying: oui oui tout le débat… pour moi, il faut décider pre- ou post- et s’y tenir. C’est à dire corriger quand une incohérence est trouvée, comme une typo en fait. (-:. <ani>g_bor: In my final application under contribution description, I have also described what skills I learned through each contribution. Is it okay? <g_bor[m]>I did not read it, but I assume it is ok. <g_bor[m]>I believe zimoun would have told it to you if there were any problems. <ani>can I include two URL's in contributing section one for V1 of patch and another for V2 of patch, or URL for V2 is enough ? <ani>g_bor: By this I finished revision of my final application. Thank you. ***jonsger1 is now known as jonsger
<ani> g_bor[m] : Thank you. <clone11>gnome-builder fails to install for me, every other package i've tried works fine. Is it the package or my system? <mfg>Do you have an error message? <mfg>Good afternoon nckx :'D <nckx>☝ if I'd have to guess I'd eye PHP suspiciously. <leoprikler>clone11: I think I sent a patch for that ages ago <davidl>Is there anyone here who has a *working* cuirass config with custom packages, and would like to share? <lfam>clone11, leoprikler: I just pushed the patch fixing compilation of gnome-builder. You should be able to build it now after doing `guix pull` <lfam>In a while, there will be substitutes <aadcg>I compiled a common lisp with guix package and I'm using slime to have a repl. but when I run M-. I get that it's looking for definition at /tmp. Any idea on how to solve this? <lfam>clone11: You're welcome! Ideally the patch would have been a while ago :/ <lfam>The patch would have been pushed, I mean <divoplade>davidl, the configuration is quite easy to get, but the database is a little stubborn <divoplade>I have an almost-working configuration, but depending on whether I deleted the database or its content, deleted the cache or not, or whatever, I get differently working results (ranging from: not working at all, to: some packages work) <divoplade>(restarted the cuirass service but not cuirass-web, or both, reconfiguring with those services stopped or disabled, reboot before or after restarting them...) <jonsger>divoplade: I had quite good experiences with deleting the cuirass db <sneek>jonsger, you have 1 message! <jonsger>I stop cuirass{-web} and do a `sudo rm /var/lib/cuirass/cuirass.{db,db-shm,db-wal}` <divoplade>I still have hope to get back to my best score^^ <jonsger>I think I never saw that error and the error message is just cryptic for me :P <civodul>maybe i shouldn't say that, but if you just want to build packages of interest off "master" frequently, you can just run this periodically: guix time-machine -- build -m my-manifest.scm <pkill9>cool, so it gets the latest guix revision without instlaling it to your profile <pkill9>i didn't think of using time-machine to go into the future, lol <civodul>now if you want to be able to browse evaluation results and all, Cuirass is a better fit <ryanprior>When I try to `make` or `make clean` in guix, I get: <ryanprior>automake-1.16: error: cannot open < ./doc/guix.de.texi: No such file or directory <ryanprior>Anybody know what the deal is with that? Those texi files are indeed missing in master. <ryanprior>But they're called for by local.mk, also in master. Did somebody remove some files accidentally, or do it on purpose but forget to update the local.mk file? <divoplade>ryanprior, when you clone guix, run ./bootstrap and not autoreconf <ryanprior>Oh I ran ./configure, didn't think to try ./bootstrap <davidl>divoplade: Im having similar inconsistent results... Im always deleting all files in /var/lib/cuirass/cuirass* and reconfiguring, currently it seems to build stuff but report them as failures :-S appears to be a never ending set of problems with cuirass. <divoplade>davidl, I'm back at the "mostly working" state, and there's a lot of lines in the log file like: <divoplade>2020-10-31T19:30:07 build started: '/gnu/store/ddnc4r1n4cmn8vf1dw01rnh4w8hv4z2n-alsa-lib-1.2.2.drv' <divoplade>2020-10-31T19:30:07 build failed: '/gnu/store/ddnc4r1n4cmn8vf1dw01rnh4w8hv4z2n-alsa-lib-1.2.2.drv' <davidl>I haven't had build started and failed at all. <divoplade>I have cloned guix and my collection of packages under /srv/git <davidl>If I were you I would change: (#:load-path . ".") to "divoplade/packages", and (#:proc-args (subset "pomdappi" "pissenlid" "index-page" "guile-mkdir-p")) to (#:proc-args ((subset . ("pomdappi" "pissenlid" "index-page" "guile-mkdir-p")))) <divoplade>Maybe, but for now I'm the one with builds started :P <davidl>divoplade: true.. :P however, I had successful builds before. <davidl>but I couldn't add a fifth package, so then I guixed pulled and reconfigured and now it doesn't work again. <ryanprior>divoplade: I ran bootstrap, configure, clean, and make again, and it's not the same error but I'm still getting doc errors <ryanprior>eg doc/contributing.de.texi:1430: @ref reference to nonexistent node `Channels' <ryanprior>And there's like 500 errors like that, not just a few. <davidl>you may want to do (#:proc-args . ((subset . ("pkg1" "pkg2" ..)) (systems . ("x86_64")))) if you don't want to build for the other archs. <divoplade>ryanprior, maybe the repository is fully messed up. Try to clone from scratch, and then re-run these commands <divoplade>davidl, in my custom packages, divoplade/packages/thing.scm is a module with name (divoplade packages thing), so the load path is really "." <davidl>divoplade: yeah, that probably makes a difference. <divoplade>And I want to know if my packages build for the hurd, so I want to keep all the architectures! <ryanprior>divoplade: re-cloned and re-ran all the commands, same dumb thing. <divoplade>ryanprior, do you already have a guix installed? <divoplade>So after ./bootstrap, you could run guix environment guix <divoplade>You can run ./configure, make and make check and everything in that prompt <divoplade>(well, most certainly the "new prompt" will look exactly like the previous one) <ryanprior>2) guix environment --container guix; ./bootstrap; ./configure --localstatedir=/var <davidl>divoplade: I got some interesting results: I ran guix pull back to early october and now the otherwise same output is saying success instead of failure for the builds. I think something between /gnu/store/7ysr4ffm04yy29ckfk991n1gzdkpdzrd-cuirass-0.0.1-50.71aac24.drv and 0.0.1-53 has caused a problem. <UnderSampled>I just did a new install, selecting Enlightenment because KDE is not on the list <lfam>We don't have a full KDE desktop yet <lfam>We do have some of the flagship packages, and we have a lot of the KDE libraries, but we are still waiting for a KDE champion to put it all together <lfam>We definitely want a KDE desktop, but so far nobody has done that stage of the work <UnderSampled>So what do the "desktop-service-type"s do, beyond packages? <davidl>UnderSampled: a lot of services are added. ***jmarcian` is now known as jmarciano`
***jonsger1 is now known as jonsger
*vagrantc forgot guix was in freeze the last few days ... <vagrantc>though typo fixes in descriptions are hopefully not too disruptive <mfg>cbaines: is this a frontend which should make patch reviewing easier? (i don't know patchwork, just went to that site) <cbaines>mfg, sort of, at least make it easier to be more thorough <mfg>and it's the first time i actually see a service that uses laminar :D <joshuaBPMan>Trying to set up opensmtpd-filter-dkimsign, and now the service refuses to start. <joshuaBPMan>Man, trying to debug a failed service starting is awkward on guix. I have to go about finding out the latest *.conf file via find /gnu/store/ -name '*smtpd.conf' and then just cat-ing all the files, until I find the most recent one...actually there's probably a way to sort by most recent... <joshuaBPMan>then again maybe not, I think all files in the store have the same "creation" date... <mfg>you mean something like this? find . -printf "%T@ %Tc %p\n" | sort -n <cbaines>Sometimes you can spot the files/derivations at the reconfigure time <cbaines>Also, if you follow down through the shepherd configuration, you might be able to figure out what files are involved that way <joshuaBPMan>cbains, it's mainly one configuration file. There are a couple of other files, but I don't think they are relevant here. <joshuaBPMan>mfg: that may not work...It looks like all files in the store have the same creation time... <nckx>joshuaBPMan: But not birthday (stat -c %w). <nckx>So I guess, no, all files in the store do not have the same creation time. <nckx>(find's %T is modification, not creation time.) <mfg>joshuaBPMan: Hm, yes they have the same mtime, but this snippet sorts with regard to last modification time, no? Otherwise i misunderstood it... sry for that! <joshuaBPMan>ahh. Ok. Let me go look up how to sort by creation time. I'll look at find's manual page. <nckx>mfg: Yes, but in the store ‘last modification time’ is identical for each (valid) file. <nckx>The snippet is correct, but the store is ‘weird’ 🙂 <mfg>Ah i see, i didn't think about that... thx for pointing it out :-) <mfg>the store is the store *shrug* <nckx>Stores gonna store... stuff. <joshuaBPMan>hmmm, I was hoping that the "find" command could print out the date created time. that doesn't seem to be the case. <mfg>also reading the manual you seem to be right :/ <nckx>Can't you use -exec stat -c %W {} + (admittedly untested)? <joshuaBPMan>nckx: I didn't realize that was something find could do... <mfg>such wizardry... much wow <nckx>Yeah, ‘find /gnu/store -exec stat -c '%W %n' {} +’ works a charm. <nckx>joshuaBPMan, mfg: Learning about find -exec (and especially ‘+’) unlocks much superpower. <mfg>true! i will look into it :-) *nckx sleepy boy, good night. <nckx>(Not denying that Unix file systems storing up to 4 different ‘times’ for each file is... well, a bit much. The fact that Guix/Nix don't zero all of them is probably an oversight. But a handy one. Anyway, o/ ) <joshuaBPMan>what's weird is sudo smtpd -f /path/to/smtpd.conf seems to work. I don't get any errors, but sudo herd restart smtpd fails.