<infinisil>
elvishjerricco: Hm, I'm not a fan of libraries based on IO, because it's basically a do-whatever-you-want effect, launching missiles and that
drakonis has quit [Ping timeout: 248 seconds]
<infinisil>
Actually restricted effects would make it much cleaner what functions do
<elvishjerricco>
infinisil: Well I mean if you supply an interpreter that runs in IO, it's the same thing
<infinisil>
And of course, testing an IO-based library can get a bit ugly
<elvishjerricco>
The interpreter can fire missiles just like an IO function can
<infinisil>
Yeah, but you only have to review the interpreter to know what it can do
<elvishjerricco>
That's true
<infinisil>
And this would in a zfs library just be to call the zfs binary
drakonis has joined #nixos-chat
<infinisil>
I'd really like to see effect-based libraries to succeed, because everything else just feels not right
<infinisil>
s/feels not right/doesn't feel right
<infinisil>
elvishjerricco: ALso, I believe it should be possible to provide an IO layer above an effect system
<elvishjerricco>
I'd need to really look into polysemy a lot more and kinda figure out it
<elvishjerricco>
Like I need to look at how it implements catchError (have to step away from my computer and use mobile for now)
<elvishjerricco>
What I'd love to see is a version of polysemy that uses type classes instead of interpreters. It'd still solve the n^2 instances problem, and you wouldn't have to really so heavily on inlining and the optimizer
<__monty__>
Looks like european evening is busiest?
<infinisil>
Peak at 22:00-23:00, low at 6:00-7:00
<infinisil>
Yup
waleee-cl has joined #nixos-chat
__monty__ has quit [Ping timeout: 268 seconds]
__monty__ has joined #nixos-chat
Synthetica has quit [Quit: Connection closed for inactivity]
makefu has quit [Quit: WeeChat 2.6]
makefu has joined #nixos-chat
psyanticy has joined #nixos-chat
* etu
held a 30 minute presentation at work for the entire tech team of the swedish equivalent of Craigslist today
<etu>
A room with 100+ people
<etu>
Don't think there was any recordings :/
endformationage has joined #nixos-chat
<etu>
But from what I've heard I got the most questions about the topic. Because some people seemed interested :)
<etu>
And I think I managed to sell direnv to some people. I heard them discussing nodejs versions per project during the break
<gchristensen>
yay!!
<gchristensen>
that sounds great, etu!
<talyz>
etu: \o/
wildtrees has joined #nixos-chat
<rycee>
etu: Blocket?
<etu>
rycee: yeah
<etu>
rycee: I work in the same building as them. And they had a tech day today where they invited external speakers to talk about interesting subjects. Mostly to give new perspectives on things I think :)
<rycee>
Nice!
<etu>
And of course I touched subjects around things I know they use that I know that Nix can do well ;D
<etu>
But that's not just to try to get them to use Nix, it's also to keep the interest of the audience
drakonis1 has joined #nixos-chat
psyanticy has quit [Quit: Connection closed for inactivity]
<gchristensen>
look in Nix rev a5ece7d016e72a61ca69a401e833314f538518f9
<__monty__>
Well, that's gross.
<gchristensen>
Nix back in the early 2000's was a pretty wild place
<gchristensen>
.fix files and whatnot
<__monty__>
Good thing I discovered it late. Might not have stuck around : )
<gchristensen>
yeah, well, very few knew of Nix at th etime
<gchristensen>
I find it interesting to go back and see the different ways Nix developed and changed to become what it is now
<elvishjerricco>
Heh, my script from yesterday turned out useful afterall. The router on the remote network went berserk and broke the connection. Got that fixed, ran the script, and it started right where it left off. Neat
<__monty__>
Code archeology is definitely interesting.
<__monty__>
elvishjerricco: Wouldn't a receive -s (or what did your script use) call have done that as well?
<elvishjerricco>
receive -s is what I used, yea
<elvishjerricco>
But my script was mostly so I could get my pretty `pv` output :P
<elvishjerricco>
including ETA and stuff
<gchristensen>
nice, elvishjerricco
<gchristensen>
did you use resume tokens?
<elvishjerricco>
gchristensen: Yep
<gchristensen>
nice
<elvishjerricco>
Yea that's a great feature to have
<elvishjerricco>
I am intentionally throttling this to take a week long to keep network saturation down, so I think -s was completely necessary
<gchristensen>
nice
<gchristensen>
very nice
qyliss has quit [Quit: bye]
qyliss has joined #nixos-chat
qyliss has quit [Client Quit]
qyliss has joined #nixos-chat
<ashkitten>
is it a packaging issue that the info page for ftp is the same as the manpage?
<ashkitten>
(side note it annoys me to no end that gnu util manpages are incomplete and just tell you to refer to the info page instead)
__monty__ has quit [Quit: leaving]
<kraem>
what is the best way to indicate to a reviewer of a PR that there's been changes? is he/she notified when there's been a new commit pushed or when a conversation has been a
<kraem>
marked resolved*
<worldofpeace>
kraem: I think it used to or should send a notification on force pushes. But idk, github is funny. The best thing to do is to comment and @ the reviewer, or request review via github ui
<kraem>
even if the person already is a participant and asked for changes?
<kraem>
had to force push anyway since i reintroduced an old typo i had previously removed with a commit :P
<worldofpeace>
Well the true flow on github should be like: Open PR -> Author Requests reviewers -> Reviewer Requests changes -> Author does a push event with changes -> Author Requests review
<worldofpeace>
just normal back and forth like a conversation I guess
drakonis1 is now known as drakonis
<adisbladis>
worldofpeace: I gotta say I really like you <3 We need more of your goofyness in nix :)
<infinisil>
Explanation of why that person switched away from NixOS ^
<gchristensen>
ouch. I'm at least partially responsible for that
<gchristensen>
I Think
<gchristensen>
"And for minor changes, I have an open PR from October 2018 that allows you to set the executable bit on $HOME/.background-image for things like nitrogen --restore. Bikeshedding led to me abandoning it." this one, I get
<worldofpeace>
"The incessant bikeshedding." has happened for me
<worldofpeace>
oh I remember that PR
<worldofpeace>
or wait, maybe it predates me 🤣
<gchristensen>
(I get: as in, I understand the reason we were't really thrilled)
<worldofpeace>
I get this person's experience, have been actively trying ways to remedy it.
<gchristensen>
(I guess we're all responsible, and it'd be good to think about how and why, and how to get out)
<infinisil>
But tbh, that's just some discussion, I'm not sure why that's bikeshedding now
<gchristensen>
just sitting around without a decision sucks
<infinisil>
Though I guess it is a pretty minor issue
<gchristensen>
I think a better thing could have been saying look, we're just not going to support that, this is a way you could get what you want
<pie_[bnc]>
also might be in scope for home-manager
<infinisil>
I think it might be better to just say we do support it for the sake of showing that we do care, it's not a big deal to have this functionality
<infinisil>
Though the check should be a bit more exhaustive than just -x file
<pie_[bnc]>
but that means you gotta pull in home manager or whatever
<pie_[bnc]>
my "bikeshed" moment is when I dont get things into lib :P but I guess I can kind of understand that
<pie_[bnc]>
it still bothers me tho :p
<gchristensen>
merging it just to show we care is not the right kind of care to show
<gchristensen>
there are good reasons to not merge this PR, and they're right -- handling all the edge cases is quite complicated -- plausibly a good reason to not include it, and replying with why is showing the same kind of care merging it would
<pie_[bnc]>
"I would much rather remove the entire section and let people set their own colours and wallpapers than try to work around every possible edge case here." doesnt sound like a bad next step, though I didnt read the whole thread
<infinisil>
Hm
<pie_[bnc]>
somwhat related, i think nixpkgs needs a lot more emphasis on overridability
<infinisil>
Yeah maybe you're right gchristensen
<pie_[bnc]>
but i can derail with that again later
<pie_[bnc]>
would be nice to get some design principles together *big handwave*
<gchristensen>
same kind of care merging it would *if it were a patch which was really great*
<gchristensen>
pie_[bnc]: that is part of my first quarter objectives! literally part of my job :D
<pie_[bnc]>
heh
<gchristensen>
(to propose that is)
<pie_[bnc]>
gchristensen: well its not like ill be able to collect my thoughts but i think deduplication of work for the Software Repositort Of Mankind should be a major goal
<pie_[bnc]>
making overrides easy so that you dont need to copy paste code from the immutable repo seems kindof orthogonal to that but yeah