<hwpplayer1>I am working on a motto and a logo and wallpaper
<mekeor[m]>if anyone is interested, here's my guix-system-declaration for my server, including certbot/letsencrypt, nginx, gitlab (inside a docker container), an mcron-service running domain-connect-dyndns from my guix-channel https://gitlab.com/mekeor/guix-channel ) and a self-made ssh-firewall based upon nftables: http://ix.io/3nfS
<mekeor[m]>mstrom: maybe like this? (let* ((port (open-input-pipe "wg show wg0 fwmark")) (str (read-line port))) ; from (ice-9 rdelim) (close-pipe port) str)
<mekeor[m]>mstrom: but i'm not sure if this is actually a good idea. i mean, the system-declaration gets somewhat confusing with this. i wonder when exactly the wg-command will be executed when you run guix-system-reconfigure
<mekeor[m]>terpri: which container-based privacy extensions are you referring to?
<terpri>mekeor[m], if i try to install Multi-Account Containers or Temporary Containers through addons.mozilla.org, i get a message that the extension is "corrupt". i didn't investigate past that
<terpri>plain firefox is good enough for me tbh, as long as EME is disabled (mozilla's telemetry is also one of the few telemetry systems i trust, along with, e.g., debian popcon, though i understand why many people may not want them)
<terpri>and i check licenses (and sometimes read source code) for the extensions i use
<terpri>(i worked closely with mozillians on bignums-for-JS, otherwise i'd probably be more cautious about telemetry, addons, etc.)
<terpri>actually, packaging libre extensions as sideloaded ones might be a neat project; debian used to do something like that iirc
<terpri>maybe i'll investigate the "corrupt extension" issue more; i'm not sure what could be causing it, *maybe* some error in version/variant negotiation because icecat has a unique User-Agent string, or something like that
<terpri>it seems unlikely that they'd allow something like Multi-Account Containers to break for firefox esr
<apteryx>I'm a bit puzzled by the pid argument that fork+exec-command/container wants
<apteryx>The only usage of it in Guix is in gnu/services/base.scm, and it seems if I don't have any specific PID to pass I can pass that of the parent
<iis>Hello, How to modify permanently /etc/hosts file in guix? In linux Mint I do that as root user and change it with nano or gedit. In gnu guix modifications to /etc/hosts file are no longer after reboot.
<apteryx>the operating-system record has a 'host-file' field you can use to define the content of /etc/hosts
<apteryx>terpri: eh, I just realized even though my Shepherd start slot needs to be a custom lambda because I need to do more than just calling the lambda produced by make-forkexec-constructor/container, I can still use make-forkexec-constructor/container, bind its result to some variable, and call that proc at my convenience :-)
<apteryx>so I shouldn't need to figure out exactly how it works (yet)
<meo>hmmm, should there be a chapter in the manual about hardware security keys
***iyzsong-- is now known as iyzsong-w
Quits:jess (jess@freenode/staff/jess) (Killed (grumble (My fellow staff so-called 'friends' are about to hand over account data to a non-staff member. If you care about your data, drop your NickServ account NOW before that happens.)))
Quits:ChanServ (ChanServ@services.) (Killed (grumble (My fellow staff so-called 'friends' are about to hand over account data to a non-staff member. If you care about your data, drop your NickServ account NOW before that happens.)))
***services. sets mode: +o ChanServ
Quits:grumble (~Thunderbi@freenode/staff/grumble) (Killed (grumble (My fellow staff so-called 'friends' are about to hand over account data to a non-staff member. I'm leaving freenode forever and I recommend you drop your NickServ account to keep your data safe.)))
<tissevert>no, for various reasons: first I think everyone should get credit for their contribution, second, I personally like the icon the way it is and third : this was a genuine idea, I have no idea how this works and if the site is open to contribution like the rest of guix is, I don't even know the corresponding repository if there is such a thing, and, like you I'm new around here and still discovering
<meo>this is the new network being set up by people who resigned
<tissevert>so a mere migration and we're good ? the malicious «lawful» people who tricked freenode into giving their data will bite into nothingness and «freenode» will continue as it is under a different name somewhere else ?
<meo>there is a list of links circulated in #freenode right now explaining the situation
<apteryx>currently it's 16 hours on master and 8 hours on core-updates
<apteryx>gdk-pixbuf should depend on rust for SVG support; I've yet to fix this on staging, there's a cycle to fix, but the regular 'gdk-pixbuf' should really be 'gdk-pixbuf+svg'
<apteryx>we could keep 'gdk-pixbuf-sans-svg' for special cases, but that one should never be propagated as that too easily leads to erroneously linking against it and then wondering why GTK apps don't render SVG icons.
<mbakke>civodul: the big remaining blockers for GCC 10 are texlive and nss -- I am almost done with the former (TeX 2020 upgrade), but still ironing out issues further in the graph in order to actually test things
<civodul>mbakke: oooh, i did see "fix GCC 10 builds" commit of yours, nice!
<apteryx>civodul: not using the containerized version of make-forkexec-constructor resolves the boot hang problem seen when using the opendht service: https://paste.debian.net/1198127/. Could forgotten mappings lead to this?
<apteryx>the service runs fine on a 'booted' system
<apteryx>katco: I don't really have a good experience with matrix outside of their browser client/android app (the linux desktop client requires as much resources as a web browser, so there's not much point to it in my opinion) I'd prefer keeping our channel on an IRC server; matrix allows their users to bridge to it
<civodul>apteryx: could be forgotten mappings, you'd have to hunt for debugging clues (possibly stracing the daemon)
<apteryx>not sure how to do that while the system boots
<katco>apteryx: there are many native clients outside of element (the element "desktop" client is just the web client repackaged). there are even some CLI clients. have a look here: https://matrix.org/clients/
<apteryx>yeah, I tried the weechat one in the past (also an Emacs one IIRC, althoug I think it had been abandoned), but they were both buggy at the time.
<terpri>yes, the base protocol is pretty straightforward (probably less so with e2ee, etc., but it's a good-enough starting point)
<jackhill>it will be a difficult decision, I think. We're cursed with many options. There's also XMPP, which is what I prefer, but I don't know what would be best in general
<terpri>not quite as simple as IRC ;), but not massively overcomplicated either, imho
<terpri>i wonder if that's a limitation of android (istr android is pretty hostile to having apps running persistently in the background, and one is expected to use google services for push notifications iirc -- i could have the details wrong though)
<rekado>jackhill: I also like XMPP. Has lots of clients, too.
<drakonis>time to jump ship to liberachat because freenode has fallen
<rekado>I know that former freenode admins have quit, but why should *users* move to a new network?
<apteryx>rekado: it seems the legal entity (associated to PIA) to which was sold the freenode domain name a couple years ago and promised not interfering with how freenode is managed has broken that promises, and is undergoing a hostile takeover
<rekado>pushcx: is this a good reason for communities to move to another network?
<tissevert>well yeah, depends who owns the backup I guess ^^' ?
<avalenn>"Contrary to public and private statements, the sale included the network as a whole, something christel should not have had the ability to relinquish, as most of the infrastructure is not owned by her or any of us» This is the part I don't understand. How can someone take over the whole network.
<apteryx>dongcarl: are you attempting without substitutes? I have a built output corresponding to the failed derivation in your log
<the-catster2>I have been trying to get WebGL to work in Icecat for quite some time now and can't figure it out. I have it enabled in settings, and it works on ungoogled-chromium and Nyxt, but not on Firefox. I have mesa and drivers installed. What could be a culprit?
<zimoun>I see, indeed, /etc/ssl/certs does not have the certificate on this foreign distro. “guix install nss-certs” fixes it. Somehow, does it make to have the expected certificates at ~/.config/guix/current/etc/ssl/certs? nckx, is it the pathc you are talking about?
<apteryx>not sure how that could fix it for GnuTLS. It's configured with the "--with-default-trust-store-dir=/etc/ssl/certs" option in Guix.
<apteryx>I know other tools such as OpenSSL can make use of the SSL_CERT_* environment variables to find the certs
<nckx>Hm, I wonder how it will handle duplicate channels names.
<apteryx>zimoun: that's not really the case currently; many parts of guix are 'optional'; one recent example is the go importer, which autoloads the guile-lib library. If your system doesn't have a recent guile-lib library, it'll won't be unusable.
<zimoun>Well, I do not know what “guix gc --requisites $(readlink -f ~/.config/guix/current) lists.
<zimoun> And for instance using --references, you get /gnu/store/j3g8krwvz530r0jdnq1r8i95ha3wxzma-guix-f8acd1aee then again on this, /gnu/store/a2ipch77lwgxbhyrv3iyjwf3nq4xv8li-guix-command and again on this, /gnu/store/a6b1s696w04gdvlffjk7mvv94sdfsbwq-guix-module-union and again on this, you finally get /gnu/store/0by44hplsmf03g28m2sdz72y85n3p4hn-guile-lib-0.2.7. (Maybe there is a way to do in one go, but I do know it ;-))
<apteryx>perhaps -R? guix gc -R $(readlink -f ~/.config/guix/current) | grep guile-lib has a match
<civodul>"staying in bed" sounds like an option until this sorts itself out ;-)
<jonsger>rekado: hm, I think thats the concept of caravans :) or night-trains ^^
<rekado>I miss the concept of affordable night trains.
<pushcx>If you need a hand, I can intro you to kline. He helped us get #lobsters set back up. Mostly it was just sorting out who got the 'founder' bit with chanserv. Then I set the old channel to +m with an updated topic and repeating message to tell people to update. Not too bad.
<leoprikler>raghavgururajan: I haven't looked at it too closely, please leave ample time also for others to review it
<katco>i was discussing this elsewhere; i know there are other concerns when considering a protocol, but moving to a protocol which is resilient against these kinds of events would be nice. i.e. no one having to move where they read guix chat in the event that a server becomes untenable
<vagrantc>katco: i missed some context ... is this regarding concerns about freenode?