gchristensen changed the topic of #nixos-chat to: NixOS but much less topical || https://logs.nix.samueldr.com/nixos-chat
neeasade has quit [Remote host closed the connection]
<ashkitten> wtf
<ashkitten> $ mount | grep games
<ashkitten> tank/home/ash/games on /home/ash/games type zfs (rw,relatime,xattr,posixacl)
<ashkitten> $ sudo umount games
<ashkitten> umount: /home/ash/games: not mounted.
<pie_> i think zfs might handle those mounts separately?
<pie_> or does this normally work
<pie_> you seem to be running into pie_ style issues
<ashkitten> this is probably a systemd bug
<ashkitten> and i have no idea how to fix it
<ashkitten> because setting mountpoint=legacy on the dataset tries to unmount
<ashkitten> and when that fails because it's not mounted, it doesn't set the property
<pie_> .-.
<ashkitten> okay
<ashkitten> i fixed it by setting canmount=noauto on the dataset which doesn't try to immediately unmount it
<ashkitten> and then rebooting
<ashkitten> and setting mountpoint=legacy
<ashkitten> lesson learned, fuck systemd
<drakonis> yeee systemd
<drakonis> init in nix lang when?
<ashkitten> gimme gimme
<pie_> calling infinisil
<infinisil> On it
<pie_> lololo
<drakonis> huehuehue
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-chat
rajivr has joined #nixos-chat
neeasade has joined #nixos-chat
<abathur> nixit
<julm> date -d @1600000000
<infinisil> "MPV captures keystrokes, calls corresponding functions from Lua script. They in turn call Haskell functions through a C binding library. Haskell code creates a bash script and accumulates there pieces composed of input timestamps. Running the script produces all accumulated pieces of media file using FFmpeg in stream copy mode."
<infinisil> Wowzers.. from https://github.com/AleXoundOS/mpv-cut
monsieurp has quit [Ping timeout: 244 seconds]
monsieurp has joined #nixos-chat
makefu has quit [Ping timeout: 240 seconds]
makefu has joined #nixos-chat
andi- has quit [Ping timeout: 258 seconds]
cole-h has joined #nixos-chat
endformationage has quit [Quit: WeeChat 2.9]
waleee-cl has quit [Quit: Connection closed for inactivity]
<Mic92> convoluted design
cole-h has quit [Quit: Goodbye]
<drakonis> yikes.
CRTified[m] has quit [*.net *.split]
alaskacanyon[m] has quit [*.net *.split]
colemickens has quit [*.net *.split]
pasukon has quit [*.net *.split]
feepo has quit [*.net *.split]
tazjin has quit [*.net *.split]
leonardp has quit [*.net *.split]
crazazy[m] has quit [*.net *.split]
balsoft has quit [*.net *.split]
bbigras has quit [*.net *.split]
flokli has quit [*.net *.split]
ninjin_ has quit [*.net *.split]
colemickens has joined #nixos-chat
CRTified[m] has joined #nixos-chat
tazjin has joined #nixos-chat
pasukon has joined #nixos-chat
feepo has joined #nixos-chat
bbigras has joined #nixos-chat
flokli has joined #nixos-chat
crazazy[m] has joined #nixos-chat
alaskacanyon[m] has joined #nixos-chat
balsoft has joined #nixos-chat
leonardp has joined #nixos-chat
maxdevjs has quit [Ping timeout: 244 seconds]
ninjin_ has joined #nixos-chat
andi- has joined #nixos-chat
<sphalerite> why does my firefox sometimes just… fail to load websites?
<sphalerite> just a continuous load indicator and no progress
<sphalerite> while I can curl them just fine
<MichaelRaskin> What happens if you press escape?
<sphalerite> it stops trying to load the page, and the address bar goes blank
<MichaelRaskin> Hm
<sphalerite> this is for all tabs in my session, which I only just started
<MichaelRaskin> Network tab says nothing?
<sphalerite> though about:performance works
<sphalerite> nope, nothing
<MichaelRaskin> Do you have DoH enabled or something?
<sphalerite> not that I know of
<sphalerite> no, I don't
<MichaelRaskin> It is also called trr
<sphalerite> huh, now they've suddenly all loaded
<MichaelRaskin> Do you have stupid stuff like safe browsing explicitely blocked?
<sphalerite> well the setting in the proxy settings for DoH is disabled
<sphalerite> so I'm faily sure it's off :)
<sphalerite> aah, not sure
<MichaelRaskin> I guess in principle you could also have hit something like CRL update?
<sphalerite> ah, browser.safebrowsing.{malware,phishing}.enable are true
<MichaelRaskin> But blocked on Pi-hole?
<sphalerite> does it block all requests while updating CRLs!?
<sphalerite> no
<MichaelRaskin> No idea
<sphalerite> I guess I'll deactivate them and see if it happens again
<sphalerite> it's doing it again!
<sphalerite> MichaelRaskin: any suggestions what else to try?
<sphalerite> hrm, and it's working again
<MichaelRaskin> I kind of want to tell you to do a tcpdump and look at it in wireshark already
<sphalerite> I'll do it!
<MichaelRaskin> The problem is that you possibly need to start it around when the problem starts
<sphalerite> hm, it doesn't even make the DNS requests until 100s after the start of the capture (and I started loading a page right after the beginning of the cap)
<MichaelRaskin> Any network activity at all?
<MichaelRaskin> Have both you an infinisil entered a competition who gets the most hilariously laggy system after an upgrade?
xd1le has quit [Remote host closed the connection]
<MichaelRaskin> (look in yesterday evening logs if you want relief that you are not alone in having weird unexplained lags)
<sphalerite> yeah some light HTTPS chatter until the actual load
<MichaelRaskin> Well, whois to help?
<sphalerite> oh wow
<MichaelRaskin> I guess you reached the part in the logs where there is a ton of system CPU consumption on ZSH startup even with warm filesystem caches
<sphalerite> yeah
<MichaelRaskin> Does whois tell anything interesting about that HTTPS chatter?
<sphalerite> one of them seems to be matrix updates
<sphalerite> which is interesting in that it's going through firefox
<MichaelRaskin> Wait how do you know it goes through Firefox?
<MichaelRaskin> What client do you use?
<sphalerite> because I don't have a standalone matrix client
<sphalerite> :)
<MichaelRaskin> Ah
<MichaelRaskin> So it's Element-Web
<sphalerite> yes
<sphalerite> the other seems to be slack, same as for matrix
<MichaelRaskin> So basically the workers for already-loaded tabs
<sphalerite> yes
<MichaelRaskin> I guess you could filter out these IPs and hope to see something more interesting
<sphalerite> yep that's in progress :)
<sphalerite> huh, some communication with one of my servers
<sphalerite> oh right, kanboard
<MichaelRaskin> Which is supposed to be one od the tabs?
<sphalerite> no
<MichaelRaskin> Ah, same pattern
<MichaelRaskin> No? Separate client?
<sphalerite> no, it's a previously-loaded firefox tab too, so same pattern
<MichaelRaskin> Ah fine
<sphalerite> I just noticed something interesting though — I don't see any IPv6 before the bit where it succeeds.
<MichaelRaskin> Looks like the issue is with new loads…
<MichaelRaskin> At that point I start wondering whether carefully looking at "ip address show" during the issue would be revealing
<sphalerite> maybe curl is better at happy-eyeballing than firefox
<sphalerite> I'll leave an mtr running in the background and see if it has anything interesting to tell me next time firefox is being funny
<MichaelRaskin> Maybe Firefox has IPv6 preference, _and_ it is long-running enough to remember you just had it, and it waits for IPv6 connectivity to recover?
<sphalerite> yes! my IPv6 is just going down completely
<sphalerite> as in I no longer have a default route for IPv6…
<sphalerite> wtf
<sphalerite> ok and it's not a problem with my internet connection, because my router can still get out just fine
<sphalerite> I wonder if using systemd-networkd was a bad idea :D
<MichaelRaskin> Let me check… strcmp returns 7, so maybe yes
<MichaelRaskin> Hm no, it's not strcmp with this pattern of return values
<sphalerite> aah I'd forgotten about the existence of strfry
<sphalerite> ha. `networkctl down wlp3s0 && dhcpcd -i wlp3s0` and I have IPv6 again.
<sphalerite> ugh.
<eyJhb> joepie91: which kind of donut?
<eyJhb> Ohh nvm. I was waaay back
<MichaelRaskin> sphalerite: mwahaha
<sphalerite> MichaelRaskin: well, thank you for your help.
<sphalerite> MichaelRaskin: is there something like systemd-networkd but better? :p
<MichaelRaskin> You are welcome
<MichaelRaskin> I dunno what «like systemd-networkd» means, to be honest
<MichaelRaskin> Because I just know at each location which IFs should be up, and, well, set them up and pass to dhclient
<sphalerite> you tell it how you want your interfaces to be configured, and it watches for link changes and configures them
<MichaelRaskin> (all scripted, of course)
<MichaelRaskin> In my experience, on a laptop, just running dhclient or ip a a on arrival and never touching the IF again is good enough for me
<MichaelRaskin> If something goes wrong, well, it goes wrong into a static and pretty transparent state, so I can just look what went wrong and invoke the «set things up for ${location}» again
xd1le has joined #nixos-chat
<colemickens> something very bad interesting to my pbp temporarily. The screen went completely wonky, and the wonky image was restored after a reboot (with bits of garbled screen content). I took it back to my office and plugged it into barrel power and started it a third time and it seems to be back to normal.
parsley936 has joined #nixos-chat
xd1le has quit [Read error: Connection reset by peer]
xd1le has joined #nixos-chat
lassulus_ has joined #nixos-chat
lassulus has quit [Ping timeout: 240 seconds]
lassulus_ is now known as lassulus
andi- has quit [Ping timeout: 244 seconds]
andi- has joined #nixos-chat
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-chat
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-chat
Plien has joined #nixos-chat
kroh has joined #nixos-chat
[krow] has quit [Ping timeout: 240 seconds]
<MichaelRaskin> What is PBP?
<hexa-> pinebook pro
<MichaelRaskin> Thanks, hexa-
<MichaelRaskin> hexa- ++
waleee-cl has joined #nixos-chat
xd1le has quit [Read error: Connection reset by peer]
xd1le has joined #nixos-chat
cole-h has joined #nixos-chat
rajivr has quit [Quit: Connection closed for inactivity]
Plien has quit [Quit: Connection closed for inactivity]
[krow] has joined #nixos-chat
kroh has quit [Ping timeout: 265 seconds]
maxdevjs has joined #nixos-chat
__monty__ has joined #nixos-chat
<drakonis> i feel we need a package count ticker on the main page
<drakonis> every time i bump my nixpkgs clone, there's a dozen new packages
<samueldr> let me get philosophical for a minute, but how do you even count a package in Nixpkgs? :)
<samueldr> all derivations available through `pkgs.`?
<samueldr> then you get pkgs.pkgsCross!
<drakonis> hmm
<samueldr> what about python3XPackages?
<samueldr> should they be counted multiple times?
<drakonis> i usually check out what's in top-level/all-packages
<drakonis> that's a fine question
<samueldr> what I meant to say is that the standard idea of "packages" is kind of blurry here
<samueldr> we probably do need to find a satisfactory enough answer to that though
<drakonis> certainly.
<drakonis> i'm partial to the notion of adding tagging metadata to packages
<drakonis> to derivation definitions tht is
<drakonis> that is
<drakonis> it'd be entirely avoidable if it was possible to mark which derivations are in fact packages
<__monty__> I think the only sane thing is to not include packages from language infras.
<__monty__> If there are sensible packages in those hierarchies they can be aliased at the top level explicitly.
<samueldr> __monty__: but they are important, they need to be introspectable for package search, and for comparative reasons with other package managers
<samueldr> it really is not a trivial issue
<__monty__> Introspectable sure. But not count towards the nixpkgs package count imo. Too many libraries.
<__monty__> A count per infra might be useful.
<__monty__> Though with something like the haskell infra that's basically "all of them" so it's not that useful.
<samueldr> yeah, and some are hand-crafter, like the python ones
<samueldr> hand-crafted*
<samueldr> there is no universal answer to the problem
<__monty__> Maybe not but when people ask "How many packages does NixOS support?" The answer they're expecting is for "packages with binaries I could run."
<__monty__> That's a start at least.
<__monty__> Counts for how complete the language ecosystem coverage is are useful too but less generally so and they should definitely not be included in the top-level figure.
<samueldr> though when they ask, why do they ask? to compare with other package managers?
<__monty__> Having to go "Yeah it's among the biggest on repology *but*..." is terrible.
<samueldr> what do other package managers use as a number?
<samueldr> that's really why it's not a trivial issue
<__monty__> I think they ask because they fear they might not be able to run half the software they like.
<samueldr> there are even conflicting ideas, all valid, of what the number should represent
<samueldr> while currently I "defend" the "all the packages" count, it's only because I'm bouncing off of your "some packages" count
<samueldr> I literally don't know which is the right answer, and we probably need all of the answers laid out all the time
<drakonis> repology counts all derivations in pkgs yeah?
<drakonis> i recall checking out how it does it once
<drakonis> it wasnt very pretty
<__monty__> I think so but I'm not sure. Since the infra packages are somewhat hidden from search now maybe that affects repology?
cole-h has quit [Quit: Goodbye]
<drakonis> the parsing script isnt included
<drakonis> this is questionable
<drakonis> holy crap
<drakonis> those stats are wrong
<drakonis> extremely questionable value
<aleph-> Huh
<drakonis> at this point i'd prefer to expand package metadata to include tags
<drakonis> because repology is unreliable
<drakonis> it includes everything
<drakonis> well, nearly everything
<drakonis> it seems to skip only a few sets
<drakonis> it also drops haskell's set
<samueldr> repology is lacking some data
<drakonis> i dont think some data describes it correctly
<samueldr> it uses the (previous) dump of data used by the previous search engine
<samueldr> which was artificially limited due to being too big
<drakonis> https://channels.nixos.org/nixos-unstable/packages.json.br this is the source right now
<samueldr> yes
<samueldr> that is it
<samueldr> this uses a limited subset of nixpkgs
<drakonis> hmmm
<drakonis> i see
<samueldr> this uses all recurseIntoAttrs from all-packages by default, then those additional ones
* samueldr was wrong
<samueldr> I thought those weren't merged, but they were
<samueldr> but we still might be missing some ecosystems
<sphalerite> __monty__: really? because a whole lot of apt packages in ubuntu are libraries without binaries you could run.
<drakonis> i mean
<sphalerite> Same for homebrew
<drakonis> debian has 24 thousand source packages right now
<samueldr> we could be better than them and provide richer information
<drakonis> yes that'd be perfect
<samueldr> with a sum total that's what other distros basically present
<drakonis> https://debtags.debian.org/ this is a pretty good starting point for such a thing
<__monty__> Yes, other distros do include libraries but not entire language ecosystems the way nixpkgs does (maybe modulo C).
<__monty__> Right?
<drakonis> no
<drakonis> they do
<drakonis> they typically include the most common language ecosystem libraries
<drakonis> but not the full archive like nixpkgs does, as they can use the language package manager for that part
<drakonis> i believe it would be beneficial to include metadata such as "available as a service/module"
<drakonis> you can get a subset of haskell's packages in fedora's repos
<__monty__> Sounds like useful metadata yah.
<drakonis> hmm, language ecosystem packaging varies in quality and availability between distros
<drakonis> languages that are highly used tend to have more packages in it
<drakonis> ie: go
<drakonis> there's far more python packages in the archive than that
<drakonis> turns out they're limiting the results
<sphalerite> yes, as it says there
<sphalerite> root@debian:~# apt-cache search python | wc -l
<sphalerite> 7439
<drakonis> hot diggity
<sphalerite> we only have 3545 :p
<drakonis> that's because it might be counting python 2
<sphalerite> oh, true
<sphalerite> root@debian:~# apt-cache search python3 | wc -l
<sphalerite> 2948
<sphalerite> ok, actually we win :D
<drakonis> they only axed it on testing
<drakonis> having a proper search engine with filters would be the bee's knees
<drakonis> then using that to redesign looking up modules
<drakonis> ho ho
<drakonis> there's a lot of room for applying that
__monty__ has quit [Quit: leaving]
<colemickens> Cool. Just gave my dad and brother a intro to Rust lesson with Gitpod.io. :)
<drakonis> nice
MichaelRaskin has quit [Quit: MichaelRaskin]
parsley936 has quit [Remote host closed the connection]
andi- has quit [Quit: WeeChat 2.8]
andi- has joined #nixos-chat