IRC channel logs

2019-12-19.log

back to list of logs

<mbakke>dejanr: the Purism Librem 13 is reported to work well: https://puri.sm/products/librem-13/
<anadon>Got an error when making a container, not sure what it means: https://paste.debian.net/1121604/
<alextee[m]>leoprikler: thanks
<leoprikler>I'm pretty sure, that --pure is implied by --container
<leoprikler>anadon: Perhaps you want guix environment python singularity --container --ad-hoc python singularity
<dongcarl>leoprikler: is `--container` relatively new?
<leoprikler>I'm not old enough to know that.
<nckx>dongcarl: No, it's very 0.x.
<gnutec>Upgrade system is take a to long time to be done.
<raingloom>is wacom support broken? i have xf86-input-wacom installed, pressure sensitivity works, but the eraser is not detected as a separate device.
<raingloom>xsetwacom also doesn't list the device.
<raingloom>wacom kernel module is also loaded. not sure how to proceed.
<gnutec>raingloom: What is wacom?
<raingloom>gnutec: drawing tablet. i'm using a cth-480
<gnutec>With printer, we have to edit config.scm and use guix system reconfigure config.scm to install. Maybe wacom is the same procedure.
<gnutec>raingloom: ...if was the case, look to guix.pdf.
<leoprikler>I never had Wacom erasers work ever, so it might just be an issue with the driver.
<raingloom>leoprikler: i'm pretty sure this one worked on Arch. but i can do a reboot and check.
<raingloom>i haven't used it in a while though, so my memory is a bit foggy.
<raingloom>gnutec: xf86-input-wacom is installed that way, through the system config
<gnutec>My guix system reconfigure /etc/config.scm still building linux-libre-5.3.18.drv ('build' phase). :/ Is any problem if I reboot the system?
<leoprikler>apart of the additional build time if you decide to reconfigure later and substitutes are still not there… no
<raingloom>yup, it works. xsetwacom list devices shows four devices.
<leoprikler>and on guix it's how many?
<raingloom>zero
<raingloom>i think it doesn't use the wacom driver for it and just falls back to something else
<leoprikler>but the tablet still works to some extent?
<gnutec>leoprikler: I just want to stop and restart upgrade after reboot.
<raingloom>yes, pressure sensitivity works, touch mode works, multitouch works
<leoprikler>gnutec: Yep, that should be possible. If you want to be really sure, you kill the process with ^C before rebooting though.
<leoprikler>raingloom: "something else" is probably libinput
<leoprikler>does your arch have any wacom-specific config?
<gnutec>leoprikler: Ok! I will upgrade system tomorrow.
<raingloom>none as far as i know, i just installed xf86-input-wacom, restarted Xorg, and bam, it recognized it
<raingloom>is this something udev might be responsible for? (i don't know much about udev)
<leoprikler>What version of GNOME are you using (Guix vs. Arch)?
<raingloom>i don't use GNOME on Arch, on Guix, uugh, whatever is current, i just upgraded a few hours ago
<leoprikler>3.32 then
<leoprikler>what do you use on Arch?
<raingloom>i3
<raingloom>with lightdm, if that makes any difference
<gnutec>Reboot and... guix pull work normally. Only guix system... stop.
<leoprikler>Building the kernel can take some time, how long does it usually take for you?
<gnutec>leoprikler: 3 hours
<leoprikler>raingloom: what does `libwacom-list-local-devices` tell you? (you might want to `guix build libwacom` to get the store path for it
<leoprikler>and this build was exceeding those 3 hours?
<raingloom>leoprikler: it lists 3 devices. /dev/input/even{6,7,8}
<raingloom>s/even/event/
<gnutec>leoprikler: Yes! I start guix system... than the system build libre-linux-...guix.drv two times with libre-linux, I have 3 hours without conclusion.
<raingloom> https://paste.debian.net/1121688/
<gnutec>leoprikler: I am upgrade right now guix-emacs and everything is normal.
<leoprikler>Could be an issue with the kernel itself then; I must admit I haven't built one for a while now.
<gnutec>leoprikler: wait...
<leoprikler>you could try building it verbosely, but that will make the build take even longer
<leoprikler>raingloom: can you try installing i3 in Guix and check whether you experience the same behaviour there?
<ScaredySquirrel>ok when connecting to chat.freenode.net with the SSL thingys set
<raingloom>i'm already in i3, that's what i've been writing from this whole time
<leoprikler>okay, so i3 on both systems?
<ScaredySquirrel>SSL_CERT_DIR=/home/tails.prower/.guix-profile/etc/ssl/certs
<ScaredySquirrel>SSL_CERT_FILE=/home/tails.prower/.guix-profile/etc/ssl/certs/ca-certificates.crt
<raingloom>leoprikler: yep
<ScaredySquirrel>i have the nss-certs installed
<ScaredySquirrel> irssi says connecting to chat.freenode.net port 6697 error unable to get local issuer certificate for TLS auth
<leoprikler>ScaredySquirrel, perhaps you want nss-certs system-wide?
<ScaredySquirrel>leoprikler: that shant matter
<ScaredySquirrel>I don't want it system wide
<ScaredySquirrel>not really
<ScaredySquirrel>it'll bite my foot in hard if I need SSL for root GUI access
<gnutec>leoprikler: The system try to upgrade kernel to libre-linux-5.3.18. I'm using 5.3.11.
<ScaredySquirrel>wow what git revision are you using? your guix seems really old
<ScaredySquirrel>oh and why would youtube say that it needs HTML5 and IceCat debug in a terminal says something about WebGL not using backbuffers
<ScaredySquirrel>it says disabling antialiased backbuffers
<ScaredySquirrel>WebGL is currently disabled
<raingloom>it looks like videos are broken both in icecat and epiphany
<raingloom>ungoogled-chromium works
<raingloom>or just use youtube-dl / mpv / something mpv based
<ScaredySquirrel>lol celluloid crashes because mpv wasn't found
<ScaredySquirrel>celluloid is the new name for Gnome MPV
<ScaredySquirrel>let me see if celluloid works when I can launch mpv
<ScaredySquirrel>and play something with it
<ScaredySquirrel>why does updating the manual pages database when installing something take literally about 2 min 30 seconds?
<ScaredySquirrel>it's a hard drive not any SSD
<gnutec>We can use vlc to do what youtube-dl and ffmpeg do.
<leoprikler>probably because it's some very unoptimized disk r/w
<leoprikler>you have to walk over all your packages, collect the manpages, and process them
<ScaredySquirrel>ah
<leoprikler>and walking takes a long time with symlinks
<leoprikler>same issue on boot as well given the many symlinks to follow
<leoprikler>(IIUC at least)
<ScaredySquirrel>it's not the latest linux kernel with the newest ext4
<ScaredySquirrel>would that improve it?
<leoprikler>not that I'd know of any significant improvement
<leoprikler>splitting your packages over multiple profiles makes hooks faster
<ScaredySquirrel>mpv errors with unable to get local issuer certificate
<leoprikler>🤔️
<ScaredySquirrel>OK how odd I can't see the difference between _ and - very clearly
<ScaredySquirrel>Iosevka Term is a very bad font
<leoprikler>That's why I just use Monospace.
<ScaredySquirrel>Ubuntu Monospace is really nice
<ScaredySquirrel>ok yes it's SSL_CERT_DIR
<ScaredySquirrel>and it's in the mpv process' environment list
<ScaredySquirrel>I can't see why it's not working other than the SSL bundle is simply wrong
<ScaredySquirrel>at least that seems like the issue
<gnutec>I think is the internet connection but the download is OK.
<ScaredySquirrel>err ok mpv launches fine but celluloid still crashes from the same shell
<gnutec>It's take a long time to upgrade app in android.
<gnutec>Build blender-2.81 have the same problem, but icecat work fine.
***catonano_ is now known as catonano
***dftxbs3e_ is now known as dftxbs3e
<str1ngs>wdkrnls: brettgilio here's a diff if you need one. I also have substitutes if they are needed http://paste.debian.net/1121696
<brettgilio>str1ngs: can't check tonight. Will check in the morning. Thank you
<str1ngs>brettgilio: no problem, I could use testing systems using radeon and intel drivers.
<str1ngs>on systems*
<ScaredySquirrel>gah I can't watch Tubi because there is no webgl
<ScaredySquirrel>I tried tons of movies and they all say "Opps, your web browser is no longer supported. Please upgrade your browser"
<ScaredySquirrel>ok is there a fix for how you can't change font smoothing settings in Xfce or Gnome?
<ScaredySquirrel>guys why not try the mosh terminal emulator and remote shell thingy?
<ScaredySquirrel>it's really good and it says that an 8-bit vt220 and a UTF-8 vt220 are really different and incompatible because UTF-8 goes below the vt220 state machine
<ScaredySquirrel>I can't see why it's not working other than the SSL bundle is simply wrong
<Blackbeard[m]>ScaredySquirrel: youtube-dl
<ScaredySquirrel>Blackbeard[m]: it's Tubi how would i use that with Tubi?
<ScaredySquirrel>oh and wait I think my mpv is broken
<ScaredySquirrel>where's my AltGr key?
<ScaredySquirrel>the keyboard key literally says "Alt Gr"
<ScaredySquirrel>um all ssl fails with "self signed certificate in certificate chain"
<ScaredySquirrel>Ok huh I install nss-certs but it's not under ~/.guix-profile/etc/ssl like it should be
<ScaredySquirrel>madagest: I went to install nss-certs but afterwards ~/.guix-profile/etc/ssl isn't there like it should be
<ScaredySquirrel>fps: slow slow down magic fps into the mesa where unix lie far beyond with your wc -l lucicon into the wireless modem far beyond
***ScaredySquirrel is now known as parallelinfinite
***parallelinfinite is now known as ScaredySquirrel
<ScaredySquirrel>oh crap now it's going to first run on a router UGH
<ScaredySquirrel>I'm looking at squirrel boys
<roptat>hi guix!
<roptat>for webgl, you can enable it by going to about:config and settings security.sandbox.content.read_path_whitelist to /gnu/store/ (with a trailing / to allow read to it and its subdirectories), then restart the browser
<jayspeer>I'm back to building packages and I
<sneek>Welcome back jayspeer, you have 1 message.
<sneek>jayspeer, efraim says: I don't know of a package definition for python 2.6 but I have one for 2.4 if you need
<jayspeer>sneek: Thanks but I don't need it :)
<jayspeer>...and I have problem with guix daemon building stuff.
<roptat>hi jayspeer :)
<jayspeer>hello :D
<roptat>please describe your issue so we can help you :)
<jayspeer>yeah right :) - from build log In unknown file: 0 (open "/gnu/store/0icm4b0z3vvv9rjn6wi9ngvjnmkrgl0a-ter…" …) ERROR: In procedure open: In procedure open-fdes: Permission denied
<jayspeer>
<jayspeer>and I end up with a single directory in store with wrong owner (guixbuilder1 with rwx) and lock on this directory
<roptat>how do you run the daemon? as root?
<jayspeer>yeah, systemd service
<roptat>ok, I'm trying to think of what could be the issue...
<jayspeer> https://paste.debian.net/1121722/
<jayspeer>here's the ls of problematic files
<roptat>could you give me more lines of your log, before the error you get?
<jayspeer>I can post the whole log somewhere
<roptat>I'm thinking, maybe this is expected, but I don't know enough about the build process to be sure
<roptat>in that case, it could be that there is something failing in your recipe, simply
<roptat>are you running with -K?
<roptat>(--keep-failed)
<jayspeer>I run: guix package -f terraform_0.11.7
<jayspeer>I'll post my package def and build log to pase
<jayspeer>paste*, give me a sec
<roptat>sure
<jayspeer>package: https://paste.debian.net/1121724/
<jayspeer>tail of build log: https://paste.debian.net/1121725/
<jayspeer>whole log is too big
<jayspeer>Same package definition for terraform 0.12.10 finished just fine. So I'm not sure what exactly is the problem
<roptat>let me try building it locally
<roptat>there doesn't seem to be anything wrong with the package definition
<jayspeer>like I said - same package definition for newer terraform passed
<jayspeer>let me paste it
<jayspeer> https://paste.debian.net/1121726/
<roptat>I get the same issue though, so that's probably not a configuration issue on your side
<roptat>maybe something wrong in "reset-gzip-timestamp" then
<roptat>there seem to be three gzip files with read-only permission
<roptat>reset-gzip-timestamp tries to open them with write permission, but gets denied, causing a build failure
<roptat>you could try adding a phase to make them read-write
<jayspeer>how would I go about that? It's my first package, I'm not really sure what I'm doing :)
<roptat>ok, let me try it and I'll tell you if it worked
<roptat>yep, it worked :)
<roptat> https://paste.debian.net/1121728/
<roptat>see I added a #:phases argument
<jayspeer>hurray! can you explain what and why?
<roptat>so as I said, reset-gzip-timestamps tries to overwrite something in any gzip file (that end with gz (.tgz or .gz))
<roptat>but these files are read-only, so it fails to open them
<roptat>hence your error
<roptat>so I've added a new phase before that one (that's the meaning of add-before) to make them read-write instead
<jayspeer>so basically it was an issue with non standard build process
<roptat>yes, but the reset-gzip-timestamps phase could be smarter...
<roptat>however I can't change it, because it would invalidate the whole package graph of guix, and the build farm would have to rebuild the world for nothing...
<civodul>Hello Guix!
<civodul>looks like berlin is down, no?
<jayspeer>doesn't seem like a big issue to me - if it's only my package causing the problem. However if there are several problematic packages it would be better to fix it now, rather than keep pushing with annoying phase. Future will only bring in more packages
<jayspeer>will overshadowing reset-gzip-timestamps with updated one still cause package rebuild?
<roptat>civodul, ah, maybe that's why I suddenly can't run a guix command anymore... it's waiting for a timeout
<roptat>jayspeer, it will, for any package that has the updated phase
<roptat>the hash in the store is combination of inputs, sources and recipes. That phase is part of the recipes, so if we change it, we change the hash of any package that uses it, causing a rebuild
<jayspeer>wasn't the big thing for lisp like languages changing function definition during runtime? :)
<jayspeer>is the definition of reset-gzip-timestamps include in that?
<jayspeer>because that would be weird
<roptat>what do you mean?
<jayspeer>if you change reset-gzip-timestamps to improved version it should not cause changes to package definitions
<jayspeer>it just would be smarter
<jayspeer>so future will packages can use improved function without adding additional step
<roptat>I don't see how one could do that
<jayspeer>where can I see the source code of reset-gzip-timestamps?
<roptat>in guix/build/gnu-build-system.scm
<roptat>it uses reset-gzip-timestamp from guix/build/utils.scm
<jayspeer>So the function seems pretty straightforward... Why not just add chmod +w before the "meat" of the function?
<jayspeer>That should not cause packages to rebuild
<civodul>rekado_: hello! you around? berlin seems to be down or very slow
<jayspeer>well - that has it's own issues... like changing timestamps of file we don't want to
<roptat>jayspeer, it wouldn't change the result, but it would change the recipe to get there, which causes a another hash (/gnu/store/<hash>-something) to be computed for the package
<janneke>jayspeer: you may want to read about functional package management, e.g. here: https://guix.gnu.org/manual/en/html_node/Managing-Software-the-Guix-Way.html
<jayspeer>so hash depend on every single definition in the tree. So my idea goes to trash :/
<snape-through-th>Hello! Where is the documentation for Cuirass' search bar?
<snape-through-th>i've no idea what I can put in there
***snape-through-th is now known as snape-web
***snape-web is now known as severus_snape
***severus_snape is now known as snape-web
<civodul>snape-web: rekado_ posted examples on guix-devel a while back :-)
<civodul>there's "spec:", "system:", and maybe "jobset:"
<snape-web>civodul: great, I found it on guix-devel indeed
<janneke>jayspeer: right; what we gain by being functional is something rather unique and valuable, though
<civodul>hmm berlin seems to be back, but i'm not sure what happened
<civodul>the load was reasonable AFAICS
<civodul>might have been a transient networking error on site
<civodul>rekado_: ↑
<g_bor[m]>civodul: any suspicious in the logs?
<civodul>g_bor[m]: lots of SSH connection attempts, perhaps more than usual
<g_bor[m]>noted.
<g_bor[m]>civodul: nginx people came back to me. You found a genuine bug :)
<g_bor[m]>They are working on a fix, and told us that disabling fionread support in the meanwhile would work.
<g_bor[m]>Should we wait for the proper fix to land, or update the package with the new config?
<efraim>I should probably add #:substitutable to the linux-module-build-system
<valignatev>Is issues.gnu.guix.org dead? It gives me 502
<jayspeer>how can I create guix pack with packages I just created definitions for?
<leoprikler>If you're using a local build of guix: ./pre-inst-env guix pack ...
<kirisime>Today I almost joined #:guix
<leoprikler>If you have just a file or directory, guix pack -L path ...
<kirisime>I'm building guix from source so I can finally send my packages in but `make test' seems to get stuck after passing tests/packages.scm.
<kirisime>It was stuck there when I checked my computer this morning, and now it's stuck again after I cancelled it and ran make test again. Is it just expected to take hours?
<kirisime>make check*
<leoprikler>Not sure, but I think a simulation of guix pull is included there.
<valignatev>savannah seems to be alive unlike issues.gnu.guix, so it shouldn't stuck on downloading the sources
<dejanr>hello, is there are a community page regarding hardware support for guix
<leoprikler>h-node.org is probably the best utility
<leoprikler>It is not exclusive to Guix, but it does list how certain pieces of hardware function with linux-libre in various FSF-endorsed distros.
<civodul>g_bor[m]: woow, thanks! i guess we can wait for the proper fix
<civodul>if you want you can add a link to the bug fix as a comment in sysadmin/services.scm
<civodul>er, in hydra/nginx/berlin.scm
<civodul>and/or you can send the link to 38411@debbugs.gnu.org
<jayspeer>what should I do when package from repos fails to build?
<roptat>check if after guix pull, the package is still broken. If so, you can report to bug-guix@gnu.org or try to fix it yourself and send a patch to guix-patches@gnu.org
<roptat>I have some time to look at it, so I can be of any help... you could tell me which package fails and I'll see if I can fix it :)
<roptat>(well, currently I'm trying to build icecat
<roptat>so my computer is a bit busy, but I'll see)
<roptat>I think I found a fix for the mp3/4 playing issue
<jayspeer>I'll try pulling first, but the problematic package is awscli
<zap`>Hi guix!
<jayspeer>o/
<zap`>I've seen mentions of "GuixOps" on the mailing list
<zap`>What is this thing?
<roptat>not sure, maybe the concept of using guix for devops?
<roptat>like guix deploy and other tools
<zap`>no people discussing it like it's a real thing. Or I just can't get it right :)
<roptat>jayspeer, I got a failure in a dependency, behave, is that what you get too?
<jayspeer>roptat: yeah
<janneke>zap`: aiui, there's a `guix partial deploy' effort for when you have a heterogenous setup, e.g. with debian boxes
<zap`>janneke: that would be really cool. Is it intended to use shepherd?
<janneke>zap`: http://git.genenetwork.org/pjotrp/guix-notes/src/branch/master/DEPLOY.org
<zap`>janneke: Thanks for the link!
<valignatev>bandali: I just sent an email to 38662@debbugs.gnu.org with a patch that can be applied to your patch. It's not visible in the issue yet and I hope I didn't messed anything, this is really the first time I'm doing it (yeah, I'm github era child, don't judge :D)
<leoprikler>Mach ein Ziehbegehren, wenn du mit der Vereinigung fertig bist.
<valignatev>Lmao, I honestly don't know how, but my single message from gmail web messed up the message in the issue: https://issues.guix.gnu.org/issue/38662#3
<leoprikler>wouldn't a better fix to be fixing the search path instead of the version?
<leoprikler>that way, you can `guix package -u emacs-next`
<valignatev>leoprikler: guix package -u emacs-next won't work with hardcoded version?
<leoprikler>not as far as I know, since the versions would be the same
<valignatev>Hm, if it doesn't take commit and sha hash into account to figure out if there was a change then yes, hardcoded version is a problem. I didn't know that. Then yeah, hardcoding the search path or making it into some kind of a regexp is a better way
<brettgilio>Good morning, Guix.
<civodul>hey!
<civodul>new post! https://hpc.guix.info/blog/2019/12/optimized-and-portable-open-mpi-packaging/
<mjw>ah interesting
<mjw>we have a wrapper for openmpi in valgrind, it is a bit of a pain because it wraps the actual openmpi implementation
<mjw>so shipping a precompiled one is often not actually helpful for the user, because they might be using some other version
<leoprikler>The situation described there reminds me of https://imgs.xkcd.com/comics/standards.png
<janneke>what could mpi be, /me wonders...
<mjw>from valgrind.scm I cannot figure out how guix handles the openmpi wrapper
<janneke>hmm, message passing interface
<janneke>civodul: interesting stuff
<jonsger>hm https://hpc.guix.info/javascript gives a 404
<civodul>leoprikler: very much applicable here!
<civodul>jonsger: where did you get that URL?
<jonsger>civodul: e.g. here https://hpc.guix.info/package/intel-mpi-benchmarks at the end of the page
<mjw>OK, my guix package skills are somewhat bad, but it looks it simply doesn't build the valgrind openmpi wrapper.
<mjw>that is one way to deal with it :)
<kirisime>On a second look, Geary has multiple COPYING.* files. Am I required to add all the licenses mentioned in them to the license field of the package?
<jonsger>kirisime: yes
<leoprikler>find out what meson/README.md/the source files say
<leoprikler>some packages may include licenses of their dependencies, which may or may not affect said dependencies
<kirisime>leoprikler: COPYING says LGPL 2.1, COPYING.icons says creative commons stuff and public domain stuff, and COPYING.snowball looks like the expat license to me but it's not like I'm an expert...
<leoprikler>In that case yes
<kirisime>Licensing is complicated...
<bandali>valignatev, hehe no problem :) and yeah, thinking about it, i think it'll be better to fix EMACSLOADPATH as leoprikler suggested
<bandali>valignatev, also, i would highly recommend using a proper mail client :) the gmail interface is notoriously bad
***ng0_ is now known as ng0
<valignatev>bandali: I know! I plan to quit gmail for good, but it'll take some time
<bandali>valignatev, nice :) it seems like they recently proposed to kill standard imap/exim authentication, forcing people to use oauth2 or whatever they call it
<bandali>i'd expect that'd break many "traditional" mail clients or mail-syncing programs
<bandali>also, i meant imap/smtp
<efraim>the initial sending of patches isn't bad. git format-patch -N; git send-email --to=bug-number@debbugs.gnu.org <list patches here>
<efraim>although on the mail front my fetchmail was failing to pass mail to the MDA on my server so I was missing about a month worth of mail from two older email addresses so I probably shouldn't talk about mail being "easy" :)
<bandali>haha
<bandali>well yeah if people used git-send-email for sending patches that'd be great
<bandali>but those that prefer to use their MUA and attach their patch, that's when it gets a bit tricky, especially if the MUA is as awful as gmail
<bandali>Gnus is one of the better/saner ones, ime
<leoprikler>Evolution is thankfully also sane.
<sneek>I'll keep that in mind.
<leoprikler>sneek, what is Evolution?
<sneek>I've heard Evolution is thankfully also sane.
<roptat>sneek, later tell jayspeer I fixed behave, you should be able to use awscli after a guix pull
<sneek>Will do.
<valignatev>I'm not a big fan of "emacs all the thing", but I'm looking at mutt
<bandali>any reason you dismiss emacs for mail? there are some really good emacs-based MUAs like Gnus and notmuch-emacs
<civodul>mjw: heh :-)
<civodul>jonsger: oh i see
<kirisime>Mail in emacs is a mistake and the first reason I'm packaging geary. I hope it's sane enough for whoever receives the patches.
<bandali>mistake? lol
<bandali>for me it's been the best decision i've ever made w.r.t. doing mail
<valignatev>bandali: No any particular reason. Just me not wanting to use Emacs for that kind of stuff :) It might change in the future though, so who knows
<bandali>valignatev, sure :) you may be missing out on some good stuff tho ;) i'd be happy to answer questions if you decide to give it a shot
<jonsger>kirisime: geary seems to be more focused on "modern" mails e.g. HTMLish
<valignatev>bandali: Maybe a part of "why" is that I've tried emacs for IRC via circe and it was bad. I understand that you can't compare two different packages doing two completely different jobs, but I realized that I'd better try something off-emacs if the purpose is not related to my programming/writing stuff
<valignatev>Totally not biased lol
<bandali>valignatev, i kinda understand how that can leave a not-so-great impression :) but indeed, one can't really generalize based on that
<bandali>many folks use circe and quite like it. me? not so much. i do love ERC though
<kirisime>bandali: I constructed a thing from offlineimap, notmuch, some sendmail thing and some elisp and python to get things working as I'd wanted, it was nice until it no longer downloaded my mail. The real reason why I don't like it is that there isn't a completely self-contained mail client available which is all I care about.
<valignatev>I also don't like IRC in general, but hey, here I am :)
<bandali>for me, the most compelling reason in doing more/most things in emacs is, emacs is where i feel home when using a computer. and naturally i'd love to maximize that comfort
<bandali>ha
<kirisime>Mail for emacs seems about as fragmented as email itself to me. I'm using ERC, though, maybe because IRC is simply a better protocol.
<bandali>my mail setup is a bit complex, but it's one that has worked best for me:
<nckx>Good evening, #guix o/
<bandali>i fetch/sync mail using getmail and mbsync (found it much more reliable than offlineimap) to local maildirs
<nckx>Ooh, holy war, nice.
<bandali>;)
<bandali>i then have a local dovecot running in front of those maildirs
<bandali>and i point Gnus's nnimap to the local dovecot
<bandali>(i've found nnimap to be *much* faster than nnmaildir)
<bandali>and i use msmtp for sending mail
<jonsger>sounds super convenient and simple :P
<bandali>yeah it's a bit elaborate, but it suits my needs best, which are blazing fast and offline access to my mail at all times
<bandali>if you're always online and offline access isn't a requirement, you could point nnimap directly to your provider's imap
<bandali>though that's slower
<bandali>also, morning nckx :p
<valignatev>nckx: a holy war? :) I though this phrase is only used by Russians when we talk about tech tools
<valignatev>s/phrase/term
<kirisime>And all I've done with geary is punch my login details in, it subscribes to fancy IMAP events rather than polling and has a touch-friendly UI.
<nckx>valignatev: I fear it is universal.
<valignatev>nckx: I've used that casually in one slack group and then had to explain what that means exactly
<valignatev>So I got an impression that "holy war" isn't used widely
<nckx>valignatev: Strange. I'd expect anyone in the Free Software community to be familiar with the tongue-in-cheek term (and the rich history behind it https://en.wikipedia.org/wiki/Editor_war). Anyway, my hatin' on Geary days are over (packaged it once, hated it, went back to Emacs) so I'll leave this for the young'uns.
<bandali>kirisime, not dissing geary, but as i said, a Gnus setup doesn't have to be this elaborate either
<bandali>and one *could* implement imap idle there too
<bandali>on the plus side, i've found Gnus and to some extent notmuch-emacs are the only two MUAs that let me work my way through a couple hundred emails everyday without getting in my way
<nckx>bandali: I don't follow Emacs development, but doesn't it support something vaguely like threads now? I assumed that was the main reason for missing Emacs-native IMAP (& IDLE) in the past.
<bandali>nckx, yeah emacs a limited form of concurrency with threads, but it's mostly co-operative, and afaik requires explicit yields
<bandali>emacs *gained a
<bandali>so yeah it can resolve some of those issues to some extent
<bandali> https://www.gnu.org/software/emacs/manual/html_node/elisp/Threads.html
<nckx>bandali: Thanks, will read.
<bandali>nckx, cheers
<zap`>I usually suggest non-techies to stick with protonmail or tutanota
<zap`>Very user friendly and privacy oriented
<Blackbeard[m]>zap`: I don't do that
<Blackbeard[m]>They are only private if you send an email to the same company
<Blackbeard[m]>Which, in most cases, does not happen
<Blackbeard[m]>And they have a terrible interface and can't be used with an email client
<gnutec>Now the guix system is downloading a kernel 5.3.18. 100MB to 48.7MB. It is working now?
***ajames` is now known as ajames
<nckx>Blackbeard[m]: Do either of them even support IMAP? Last I looked at Protonmail (‘for a friend’) they only supported it over some NIH ‘bridge’.
<gnutec>Yes! guix system upgrade. o/ reboot
<bandali>nckx, yeah i'd recommend staying away from them, esp. protonmail
<bandali>its "imap support" is you having to download and run their proprietary local mail server
<nckx>Oh wow it's proprietary‽
<leoprikler>very imap, much support
<bandali>yeah
<nckx>Their home page proudly proclaims ‘Open Source’ (with the GitHub logo because that is the company that invented sources and opening them).
<bandali>lol
<nckx>That's straight-out mendacious.
<nckx>Thanks bandali, next time someone asks I'll just scream instead of shrugging.
<leoprikler>FSF? Never heard about that.
<nckx>gnutec: \o high five.
<gnutec>Wow! New gnome 3.32 is nice. :)
<bandali>nckx, good :D cheers
<bandali>someone wrote a free alternative to their proprietary "bridge"
<bandali>but idk why you'd want to use such a mail service to begin with
<gnutec>New icons, and gdm theme.
<nckx>Free Stuff Forever.
<gnutec>With kernel 5.3.18 working fine.
<bandali>lmao nckx
<bandali>anyone serious about security for email should gpg-sign and -encrypt it
<bandali>rather than trusting these for profit tech companies
<vagrantc>there's admittedly a space for "not serious, but would like to get out of the habit of plain-text everywhere"
<nckx>Phone owners: are there ‘apps’ for that?
<nckx>That seems to be the bar now.
<vagrantc>gpg routinely surprises very capable people, in my experience
<bandali>vagrantc, right. i think that's where things like https://autocrypt.org could be useful
<leoprikler>bandali: And probably not use Thunderbird
<vagrantc>yes, autocrypt is great
<bandali>:)
<bandali>leoprikler, i mean thunderbird is one of the good ones, if you use enigmail
<bandali>unless there's something awful going on with it that i'm unaware of?
*vagrantc loves using deltachat on marginally trusted devices
<bandali>^ that's a good one too
<nckx>vagrantc: See: this very channel; yesterday.
<vagrantc>is someone working on deltachat for guix? :)
<leoprikler>Well, I can't talk about current Thunderbird, but Efail worked on all of Thunderbird's crypto.
<leoprikler>whereas relatively smaller mail clients did not suffer from it
<bandali>i thought efail had to (in part) do with gpg's somewhat clunky interfaces?
<vagrantc>the problem with encrypted email is you have no idea what the other end of the connection is using ... you can do everything right, but if they use some questionable client, all your base are belong to us
<bandali>(i may be misremembering)
<leoprikler>perhaps, but only thunderbird got that one wrong
<bandali>ha
<bandali>but i wouldn't discredit thunderbird itself
<bandali>perhaps enigmail
<bandali>but even then
<bandali>i think they're doing good work
<leoprikler>even other clients that had s/mime exfiltration channels did it right
<bandali>gotta drop off for a few hours, will be happy to take this up again if y'all are around and wanna chat
<bandali>o/
<nckx>\o
*vagrantc is working on pinebook-pro support for guix and debian a bit between other projects
<leoprikler>I think Efail specifically targeted the combination of HTML rendering + Encryption
<nckx>That was my layman's understanding while I stood by eating text/plain popcorn.
<zap`>Blackbeard[m]: yeah I know but it is better to use them then gmail. You can generate keys and encrypt stuff using their web interface. They opensource clients. They respect GDPR. What else would you suggest as a user friendly alternative for absolute non-techies?
<zap`>oh I had network down. Just red the logs
<gnutec>How I activate SELinux in my guix system? Guix have all tools of SELinux instaled, but the MAC is not enable.
<gnutec>Just did it! I activate printer service just edit the config.scm. "shepherd: Service cups has been started."
<grillon>Hi there
<zap`>grillon: Hi!
<grillon>Hi zap`
***MinceR_ is now known as MinceR
<grillon>hi MinceR
<MinceR>hi
<grillon>did not know you were using guix
*zap` just red about autocrypt. Looks very cool
<gnutec>Printer in guix system [pt-br]. https://unixpablo.blogspot.com/2019/12/impressora-no-guix-system.html
<bavier>sneek: later tell civodul nice post on openmpi; glad you could put so much work into it :)
<sneek>Will do.
<bavier>sneek: botsnack
<sneek>:)
<kirisime>When submitting patches, can I just add the file `git format-patch' creates as an attachment? Can I still use a screen name?
<leoprikler>Yes, but for the sake of copyright lines, make sure to indicate which name + email are the ones you want to be used.
<leoprikler>Normally, it is assumed to configure your git in a way that the name + email in the patch are the correct one.
<leoprikler>Which creates problems if you have stuff like "Kirisime <kirisime@kirisime.local>" in there.
<leoprikler>Did that answer your question or did I talk about something unrelated?
<kirisime>Partly it did, yes. `git format-patch' seems to have some mail header-likes in it though. I guess they're fine.
<leoprikler>Yep
<leoprikler>Specifically, the "From:" line will be used for the sake of attribution.
<kirisime>I guess it's time to stop being afraid of having my given name online...
<leoprikler>I'm not sure, whether it has to be your real name, but at least it has to be some name + email that belong to you
<leoprikler>Putting a mail, that is only valid inside your home (if even), is not really a good idea.
<talope>hi, i'm wondering how to make suspend-to-disk work on my laptop. I've added swap space the guix way, and loginctrl hibernate works (as well as direct tee in /sys/power/stat) however the initrd does not resume properly, even with resume=/dev/sdaX set on the kernel commandline. Anybody knows where to look / how is initrd handled in Guix ?
<leoprikler>it is not (yet?)
<nckx>kirisime: I agree (at the least) with leoprikler, *please* use a working address, it doesn't need to be your primary. We already grant you anonymity; don't be an unreachable ghost 😉
<talope>ok, sorry for the noise, searching through google pointed me in the right direction.
<leoprikler>having hibernate in Guix would be quite a useful feature though, is anyone working on that yet?
<nckx>talope: Not noise, it's probably a reasonable expectation to have. leoprikler: Not that I know of, do you know what's required from the initrd?
*nckx is only familiar with TuxOnIce and even that's broken now, not with vanilla STD.
*nckx chose that acronym poorly.
<leoprikler>Not quite, I just recently found a comment about initrd not handling this case somewhere.
<leoprikler>STD = SysTemD?
<nckx> https://wiki.gentoo.org/wiki/Custom_Initramfs/Hibernation#Kernel_configuration
<nckx>Assuming we consider mdraid out-of-scope for now (and we don't support LVM anyway), seems like most of the ‘work’ would be FFS DON'T TOUCH ANYTHING.
<talope>nckx you're mentionned in there:
<talope> https://www.mail-archive.com/bug-guix@gnu.org/msg14466.html
*nckx duckduckgo'd ‘STD’ for a suitable link to send and it shows images first. Thanks… so much.
<talope>and the patch lives in pantherx there:
<talope> https://git.pantherx.org/mirror/guix/commit/af8d58efa05927b24694c87379cccc378f3fdde7
<nckx>‘It needs a bit more work before it's ready to include in Guix, and I'm sorry to say that I lost motivation to work on it because it seems that almost no one cares about hibernation today.’
<nckx>That's sad.
<nckx>talope: My patch was older, uglier, and me-specific, Mark's is the one you want. And much less likely to have bitrotten in ‘interesting’ ways.
<talope>well, next step is: is there a way to apply patches to guix itself during the guix system build, or do I need to fork off the guix git and somehow have my installation point to the forked repository? I'll test the patch and report on https://debbugs.gnu.org/cgi/bugreport.cgi?bug=37290
<lispmacs>hi, does anybody know how to remove the privacy settings disable in new tabs in icecat. The settings btn does not give an option for it
<nckx>You'd do the latter: git clone guix, guix environment guix → ./bootstrap && configure --localstatedir=/var && make, then invoke your custom guix as ~/the-git-repo/pre-inst-env guix system reconfigure …
<lispmacs>s/disable/display
<nckx>talope: ☝ bootstrapping is described in (a bit) more detail in the manual.
<nckx>Once you've done that you don't need to rebuild guix for every change, just save your changes and run ./pre-inst-env guix.
<nckx>lispmacs: Could you rephrase ‘remove the privacy settings disable’?
<nckx>lispmacs: Have you searched about:config for ‘newtab’?
<talope>nckx: thanks for the help
<lispmacs>nckx: search for newtab brings up about 50 items. I'm not noticing one clearly related to the privacy options display, but I could be missing it
<lispmacs>what I really want is to disable anything displaying in new tab, but don't know how to make privacy otions display go away
<lispmacs>nckx: wait, I just found an option that disables the newtab page display, that worked
<brettgilio>civodul: Mind if I private message you about fencepost and gnus?
<brettgilio>Have some questions you might be able to help with
<brettgilio>Or nckx
<nckx>lispmacs: I was going to recommend that option specifically, but then toggling it didn't make a difference here (blank in both cases). Glad it works there.
<nckx>brettgilio: Not a gnus user.
<brettgilio>nckx: traitor. :)
<nckx>mu4e4lyfe 🖖
<brettgilio>Basically my issue is that fencepost is POP, and for whatever reason built in gnus fetcher doesn't seem to want to fetch from Fencepost. I can fetch using getmail, but not from gnus directly. It's odd
<brettgilio>Was wondering if somebody would want to share their setup.
<brettgilio>Asked in #gnus too, but it is silent
<brettgilio>I'm AFK anyways now though
<brettgilio>But if you have suggestions, private message me, or send me an email brettg@gnu.org
<kmicu>I’m still using Gnus but with nnmaildir backend and local Dovecot. Need to go back to mu though. 😺
<civodul>brettgilio: sure, go ahead
<sneek>civodul, you have 1 message.
<sneek>civodul, bavier says: nice post on openmpi; glad you could put so much work into it :)
<civodul>heh, i'd have happily done... anything else ;-)
<dejanr>what do you use to connect to wifi network?
<nckx>dejanr: network-manager (through nmtui).
<bandali>it'd be great if one of y'all could sort out polkit rules for it to be able to add connections etc as a non-root user member of netdev
<dejanr>nckx: thanks, and how do i configure wireless? just noticed there is no wireless card in ifconfig
<dejanr>sorry first time guix install
<dejanr>maybe i should have gonne trough reference manual
<dejanr>is the reference manual only manual? i would be interested in some user guide as well
<leoprikler>It's available on Alt-F2. There's also a cookbook, but that's not on the installation DVD afaik.
<dejanr>leoprikler: thanks
<ng0>dejanr: tty2, strg-alt-f2
<ng0>or is alt-f2 a more modern thing to reach it?
<leoprikler>oops, forgot about the Ctrl
<leoprikler>ng0: Fellow German Keyboard user :)
<ng0>strg/ctrl/whatever :)
<nckx>dejanr: What kind of wireless card do you have (e.g. $ lspci)? Guix only supports cards without separate non-free firmware.
<nckx>Eh.
<nckx>dejanr: What kind of wireless card do you have (e.g. $ lspci)? Guix only supports cards without separate non-free firmware.