<drakonis>
the pull requests count keeps rising faster than they're merged
Arahael has left #nixos-chat ["WeeChat 2.0.1"]
<jasongrossman>
drakonis: I wish we had a way to measure important pull requests ... although I expect there would be the same pattern, but who knows?
<drakonis>
test new packages and push them to a staging repository
<drakonis>
that then gets collected every x amount of time
<drakonis>
boom you halved half the pull requests now
<drakonis>
205 new packages waiting to be merged
<drakonis>
with packages that go back way back
lassulus_ has joined #nixos-chat
<drakonis>
we have a year of packages that can get merged that need triaging
lassulus has quit [Ping timeout: 264 seconds]
lassulus_ is now known as lassulus
<drakonis>
ah there's a nix package for selinux policy
<jasongrossman>
drakonis: Good thoughts, IMO.
<drakonis>
i think that the updates could also be harvested
<drakonis>
put them together to ensure a workable state
<drakonis>
as sometimes there are new packages that depend on other pull requests to enter
<colemickens>
Is there a reason this isn't done now? Need more person power? More automation? More hydra jobs? More build capacity?
<colemickens>
I've been curious lately about the meta aspect of the project - how it is run, etc, to know if donated servers, etc would be useful. If there's a roadmap, major initiatives, etc?
Jackneill has quit [Ping timeout: 240 seconds]
<drakonis>
there's a bunch of those
<drakonis>
a issue to build a roadmap from, will show up soon
<drakonis>
having a way to list the meta aspects would be nice
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.1)]
<jD91mZM2>
sphalerite: Never mind about xidle, I saw the man page and thought "wow, this does exactly what I need, let's see how it works". And then it turns out it requires a small window in one corner, and your mouse over that window.
<jD91mZM2>
I'm not against using plugins that are installed by default on most systems, such as XScreenSaver. If one of these supply a configurable idle event that's awesome. If not, I don't want to fall back to using mouse/keyboard detection as it might not necessarily be the only ways to go active again
jasongrossman has joined #nixos-chat
jD91mZM2 has quit [Quit: WeeChat 2.2]
alienpirate5 has joined #nixos-chat
jackdk has quit [Ping timeout: 246 seconds]
<sphalerit>
jD91mZM2: are you sure you can't just use the currently focused window for xidle?
emily has quit [Ping timeout: 252 seconds]
<ekleog>
,locate xidle
<{^_^}>
Found in packages: xidlehook
<ekleog>
but… but… it's software you wrote oO
jD91mZM2 has joined #nixos-chat
<jD91mZM2>
ekleog: Yeah, I'm trying to make it work in a more efficient way. When I said "xidle" and not xidlehook I meant another thing, not in nixpkgs
<jD91mZM2>
sphalerit: What if there's no focused window?
<jD91mZM2>
sphalerit: Perhaps I could do what XScreenSaver supposedly does behind the scenes - creates an invisible window with passthrough events, but still captures them.
<ekleog>
jD91mZM2: oh ok :)
alienpirate5 has quit [Ping timeout: 264 seconds]
<jD91mZM2>
sphalerit: Actually, you sure xidlehook activating every single mouse/key/whatever event (especially if I do do the invisible window trick) is more efficient than just polling XScreenSaver?
__Sander__ has joined #nixos-chat
<sphalerite>
jD91mZM2: it doesn't need to, only after the first activation, and then you won't have many events
<gchristensen>
one thingis you have to manually start the unit the first time, since nixos won't start newly created user units
<sphalerite>
gchristensen++
<{^_^}>
gchristensen's karma got increased to 41
<sphalerite>
gchristensen: not a fan of chmod 777 though :p
<gchristensen>
lol there was some u+w or something that I needed... anyway, it gets normalized to 555 or whatever when its put in the store
<gchristensen>
patchShebangs silently doesn't do the thing if it can't write to the file, and `cp a-store-path ./` is not writable
<jD91mZM2>
sphalerite: I am not sure I follow
<sphalerite>
jD91mZM2: you only need to start getting events once the first timer has run
<sphalerite>
jD91mZM2: and at that point there won't be many events, since, well, it's idle
<andi->
gchristensen: why not just use xautolock for that case?
<jD91mZM2>
sphalerite: But the question I'm trying to solve is how we know when the first timer has run
<gchristensen>
andi-: it didn't seem to work?
<andi->
oh, works for me :/
<gchristensen>
probably pebcack tbh
<jD91mZM2>
sphalerite: Depending on how much you use your computer, I'm thinking xidlehook sitting through constantly detecting when you're active is worse than sometimes checking if you're inactive
<andi->
gchristensen: I like you choice of background image :) I looked up my old python script which was supposed to do almost the same in <10 lines of code.. I was targeting the PreSleep DBUS event from systemd. Which sometimes only arrived AFTER suspending. So thats why I am / was slightly negative towards that approach :)
<gchristensen>
ah... there might be a race there, I guess
<andi->
yeah, I wanted to do things like kill any potentially running ssh-agent's (that I only spawn amnually when needed, gpg-agents, ...))
jasongrossman has quit [Ping timeout: 240 seconds]
<gchristensen>
speaking of which, my gpg agent is broken. I can't sign anything. need to look in to why.
<tazjin>
I have a similar issue, but only with signing git commits
<tazjin>
it works fine for other stuff
<gchristensen>
hmm!
<andi->
works fine for both for me..
<andi->
Last signed mail was like ~10h ago on (latest, ancient ;)) 18.09 with minimal patches
<gchristensen>
ruh roh, gpg seems to think I don't have any secret keys.
<andi->
it expired?
pie_ has joined #nixos-chat
<gchristensen>
yep! bingo
<andi->
so you are compromised? :P Someone might distrust your key from now on because it isn't good practice :)
<gchristensen>
couldn't blame them
<andi->
even tho modifying system time, extending it, and only now pushing it to keyservers would be edgy
<andi->
it's broken.
<gchristensen>
gpg is definitely a good thing _sigh_
<ekleog>
next year sequoia gains hardware token support, or so it seems :)
<andi->
so there is the ref `pull/ID/head` on github to fetch pull requests (even merged ones). Is there a similar thing to figure out the base the PR was made upon? I am lazy while picking a few larger commit series..
<gchristensen>
andi-: good news :) my top level key hasn't expired, just a subkey.
<gchristensen>
(and I can't extend it, since that is a detached key)
<LnL>
that gives the last commit of the pr, I think you can also get the parent
__monty__ has joined #nixos-chat
<andi->
that i can massage with my local git client
<andi->
LnL: thanks
<LnL>
and if not can can do something like "$commit"^ with the first commit
<andi->
thats what I meant with massaging :)
<LnL>
about that, does anybody know if it's possible to get an archive tarball for the pull/$pr/merge refs?
<LnL>
that would make my thing much nicer
<andi->
great, now I have exceeded my api ratelimit ^^
<jD91mZM2>
sphalerite: While I'm typing this message, each keypress is logged to the console. Success!!
<jD91mZM2>
Keyloggers are scarily easy to implement. Apparently an unfocused window can grab the keyboard, and even forward events to the original window to make it hidden
<andi->
thats something wayland is also supposed to fix :/
<andi->
I feel like making improvements in regards to wayland (sway & kwin, …) would be a great thing for 19.03 but can't dedicate much time.. Huge kudos to whoever works on that :)
<jD91mZM2>
andi-: Nice! Does wayland provide a good way to still detect when a person is idle?
<andi->
jD91mZM2: no idea
<andi->
I used sway for a few days ~2y ago.. was painful coming from years of X11
<jD91mZM2>
andi-: How so?
<andi->
I had a touch screen, whenever it sensed anything sway segfaulted
<andi->
so if you suspended the machine and accidentially touched the screen everything "gone"
<jD91mZM2>
wow
<andi->
that scared me off a bit :)
<jD91mZM2>
I can understand that. Is that necessarily wayland's fault? I mean it's stable enough for gnome to use by default, so...?
<andi->
Not sure whos fault that is... It just means someone has a bad practice on how to handle unknown inputs? Is that the best choice for some scenario?
sir_guy_carleton has joined #nixos-chat
<jD91mZM2>
sphalerite: UGH, another issue: The window doesn't get events from the lockscreen (I guess that's good to prevent keyloggers!), which means this whole thing won't work
simpson has quit [Ping timeout: 252 seconds]
<sphalerite>
jD91mZM2: huh, I'd think that if the keyboard is already grabbed the locker would fail to grab it…
<jD91mZM2>
sphalerite: Maybe that's what happened but the opposite order, I tested this using `sleep 1; cargo run` and then locking while it slept
<sphalerite>
ah right
<jD91mZM2>
Still a problem. Fighting X11 is not fun :P
<sphalerite>
yeah…
<jD91mZM2>
Also, yet another issue: The argument parser won't let me know what arguments belong to what timer if I pass multiple like --time 123 --timer "echo hi" --time 234 --timer "echo hello" --some-arg
<jD91mZM2>
Maybe I can just combine it all as --timer 1m "echo one minute passed"
simpson has joined #nixos-chat
<jD91mZM2>
And then using `s` instead of `m` would use the second resolution, even though that's slower
<jD91mZM2>
s/slower/less efficient
<joepie91>
jD91mZM2: important thing to note where wayland is concerned, is that it's a protocol, not a display server like X11
<joepie91>
so wayland isn't exactly 'stable' or 'unstable', that's a property of the specific wayland compositor implementation you're running
<joepie91>
which is usually a part of whatever provides your desktop environment / window manager
<jD91mZM2>
joepie91: Oh right. But there's a library for it that everyone uses... right?
<joepie91>
jD91mZM2: hm, don't know about that. I think sway is its own implementation?
<joepie91>
but perhaps not
<jD91mZM2>
sphalerite: What do you think of that idea? Leaving the idle detection like it is, but supporting multiple timers using the above syntax?
<jD91mZM2>
Maybe also specifying "primary" somewhere to make it be the thing that happens when you manually trigger the timer using the socket
<zimbatm>
I will try this out this week if I get the time
Synthetica has joined #nixos-chat
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.1)]
<disasm>
me too! Last I checked though, sway didn't work with screen sharing which was a deal breaker for me.
emily has joined #nixos-chat
<jD91mZM2>
zimbatm: Nice, thanks! Any projects using it you wanna share? :)
<gchristensen>
ls
<gchristensen>
d'oh
<disasm>
at least it wasn't a password :)
<gchristensen>
...right....not a password.......
<jD91mZM2>
Why'd you type "not a ********"?
<jD91mZM2>
All I see is stars
<joepie91>
lol
<jD91mZM2>
If anyone of today's very very lucky 10000 are here, that's a reference to http://bash.org/?244321
<joepie91>
that's two references in one line :P
__Sander__ has quit [Quit: Konversation terminated!]
<jD91mZM2>
joepie91: Help me I can't stop
<joepie91>
is that like, pop culture poisoning?
<jD91mZM2>
joepie91: I believe it's an unintended side effect to too much internet exposure. The human body is only physically capable of so much.
drakonis_ has quit [Ping timeout: 244 seconds]
drakonis_ has joined #nixos-chat
drakonis has joined #nixos-chat
drakonis has quit [Quit: WeeChat 2.2]
drakonis has joined #nixos-chat
drakonis_ has quit [Ping timeout: 240 seconds]
drakonis has quit [Ping timeout: 272 seconds]
<Synthetica>
... I just found out the hard way that when you `git checkout <file>` the change doesn't get saved in `git reflog`
<jD91mZM2>
sphalerite: I'm an idiot. It took me until now to realize I won't need a resolution like minute/second either way, since a canceller should only use second precision and every single other timer can use the sleep(delay-idle) trick
<jD91mZM2>
Thanks for the suggestions, and sorry for being slow. Currently implementing it, and it's gonna be great :)
<sphalerite>
jD91mZM2: nothing to apologise for! It's great work you're doing and I appreciate it a great deal
drakonis has joined #nixos-chat
<jD91mZM2>
sphalerite++ for the smart sleep(time-idle) trick, it seems to be working very very well
<sphalerite>
\o/
<jD91mZM2>
Umm {^_^}?
<jD91mZM2>
sphalerite++
<{^_^}>
sphalerite's karma got increased to 35
<jD91mZM2>
there
<jD91mZM2>
Also, when there's no timer left to be invoked I can just sleep for the first timer's duration :D
<sphalerite>
why? Don't you need to catch wakeups?
<jD91mZM2>
sphalerite: True, any timer with a canceller bypasses this and instead polls every second :|
<jD91mZM2>
But apart from that, nope :D
<sphalerite>
I'm not sure I get why it needs to poll
<sphalerite>
oh right you're not doing the event capture thing?
<jD91mZM2>
Yeah no that didn't seem to work :(
<jD91mZM2>
With the lockscreen, I mean
<sphalerite>
aaah right
<jD91mZM2>
Seems to work quite well so far :) I'll push it to a new branch if you want to try it!