<samueldr>
don't believe their "marketing", all their towers do
<gchristensen>
hard-refresh this a few times: http://147.75.96.102/ -- its being served by two servers on the same IP :D
<joepie91>
gchristensen: eh... :P
<joepie91>
gchristensen: how's that set up?
<gchristensen>
BGP :)
<samueldr>
monkeys unplugging and plugging really fast
<joepie91>
I prefer samueldr's explanation
<gchristensen>
as far as I know, maybe that is how BGP works
<joepie91>
then again-
<samueldr>
Beautiful Giant Primates
<joepie91>
damnit
<joepie91>
cut off my joke there :P
disasm has quit [Ping timeout: 240 seconds]
disasm has joined #nixos-chat
lassulus_ has joined #nixos-chat
lassulus has quit [Ping timeout: 264 seconds]
lassulus_ is now known as lassulus
sir_guy_carleton has quit [Quit: WeeChat 2.0]
<infinisil>
Idris just might have been the culprit for my freeze, after having coded in it for a while now it used like 8GB of memory, and usually my system is using about 15GB already
<infinisil>
It's really cool though, what I just coded in it
<infinisil>
Namely, a way to do versioning upgrades between types [a, b, ...], by just specifying the types and the migrations between them, and it will work out what to do when it reads a certain version number
<infinisil>
E.g. let's say you use a type Int in the first version of your program, so you set `stateVersions = [Int]`, then later you want to migrate everything to a String, so you do `stateVersions = [Int, String]` and provide a transition function `transitions = [(\int => show int)]`
<infinisil>
And everything works when the types check!
<infinisil>
Now I'm not 100% done, but the most important function is written
Lisanna has quit [Remote host closed the connection]
Lisanna has joined #nixos-chat
<andi->
might as well be haskell... I will never find the patience to read through the "bloat" of control characters..
obadz has quit [Ping timeout: 260 seconds]
obadz has joined #nixos-chat
obadz has quit [Ping timeout: 268 seconds]
obadz has joined #nixos-chat
__monty__ has joined #nixos-chat
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-chat
Jason_Grossman has joined #nixos-chat
<infinisil>
andi-: you do have a point heh
<gchristensen>
control characters, andi-?
<andi->
gchristensen: well all the operators )(:: -> for me (as a non haskell/idris/.. guy it looks very uncommon :-)
<andi->
I started learning erlang latelty.. so It isn't all that foreign to me
<gchristensen>
ohhh yeah it does look like nothing to me
<gchristensen>
erlang is fun :)
<gchristensen>
no types makes me anxious, though
<andi->
yeah..
<gchristensen>
and its mechanism for trusting other erlang servers is terrifyingly bad
<gchristensen>
haha
<gchristensen>
I do like it though
<andi->
I have to dealw ith a army of python developers that are not used to having to specify types at the office.... uargh
<gchristensen>
oh andi
<gchristensen>
you poor thing
<infinisil>
I need mah types
<infinisil>
It's so reassuring
<andi->
I started adding comments to reviews like "any reasonable language wouldn't accept this"
<andi->
It's reallt frustrating without types :/
<__monty__>
I'm not sure carrying type discipline over to non statically typed languages is a good idea. Sounds like a recipe for frustration and pissing off people not used to types.
<andi->
Yes and no... I sometimes just wish there was a compiler that hits their fingers on some lines of codes...
<andi->
And that thing should not be optional
<gchristensen>
are you using python 3.6? :)
<andi->
yes
<andi->
I am using type annotations ;-)
<gchristensen>
you can start adding type
<gchristensen>
nice
<andi->
and mypy
<andi->
the python4 annoucement read nice as well.. with python4 there will be some nice changes to the type systems.. forgot which but the annotations will get better (somehow)
<infinisil>
I used python a couple times.. i just can't do it, the dynamic nature of it was such a big problem
<gchristensen>
nice
<gchristensen>
tbh I'd rather reach for bash than python mostly .... X)
<andi->
I used it heavily in the past. Still do.. I would recommend it as a great first/2nd language.. Everything beyond that is personal flavor. Since we are in #nixos-chat I guess most people are a bit different from the mainstream already. ;-)
<gchristensen>
:D
<infinisil>
There's so many Haskell people in the nix community I love it
<andi->
tbh I did have some anger against the whole haskell on nixos thing a few times already.. but in general I don't mind it
<infinisil>
andi-: What kinda things?
<andi->
It is not really haskell specific but the whole mirroring an entire other package manager thing.. And the fallout that comes with it.. It just feels wrong (on a stable branch)
<infinisil>
Heh, well I love it, never need to package anything
<andi->
sometimes I like that we are incorporating such other repositories and sometimes I feell they'd be better off in their own overlay and can hack around there as they wish (and it's out of my sight ^^)
<gchristensen>
someday :)
<gchristensen>
man, figuring out how to use BGP in Packet is making so much stuff possible
<andi->
how are they doing it? APIPA/LL neighbours?
<__monty__>
andi-: I actually like it this way (granted I'm biased because I use haskell). Only reason every languages comes up with their own packaging system is no package manager was good enough to serve their needs but nix *is*. Also, how is it *in* your sight in the first place? Not like haskell packages turn up in a regular query.
<simpson>
__monty__, andi-: The Monte language more or less uses Nix as the package manager; I've been working towards shipping a Monte-upon-nixpkgs which integrates directly with the rest of nixpkgs.
<andi->
__monty__: reading commits, reading #nixos-* following issues... :-)
<__monty__>
I wouldn't mind overlays as an organizational management tool though. Have the cross-section of nix & haskell people manage the haskellPackages part, as long as it comes included with nixpkgs and doesn't take an extra step to configure.
<andi->
__monty__: no hard feelings.. It is just that sometimes I think one way or another is better... It flips. Both have pros & cons. I was aware when I wrote that, that I will not make many friends with that statement ;-)
<__monty__>
andi-: I actually agree that nixpkgs is too big to maintain. People smarter than me should come up with a nice compositional architecture to solve that problem.
<infinisil>
__monty__: "Compositional architecture"? I have no idea what that would mean
<andi->
why should one language be included in nixpkgs vs another? Or would you still ship all per default?
<__monty__>
simpson: I've mused about that as well, it's a huge part of technical and infrastructure issues that just dissappear.
<simpson>
infinisil: The idea is presumably that many small pieces would be composed together at runtime to provide the entire tree, and then presumably each piece could be independently maintained.
<__monty__>
andi-: All the things. Nix should rule the world ; )
<simpson>
__monty__: Kind of? You need to design your language for it. For example, Haskell's module system is not really great for portable modules.
<__monty__>
infinisil: I'm thinking something like having haskellPackages in an overlay instead, and ditto for other packaging subuniverses, like gnome for example. I'm just not familiar enough with nix/nixpkgs to see whether overlays are the right unit of composition.
<infinisil>
Well to be exact, all of nixpkgs is already just a bunch of overlays
<__monty__>
manveru: Just that it sounds like AUR. But the AUR is such a second class citizen on arch I haven't looked into it.
<simpson>
__monty__: It's a problem because the only way to deal with Haskell modules is Haskell packages, which have terrible composition.
<manveru>
i wouldn't call it second class... it just standardizes third party nixpkgs-alike repos
<__monty__>
infinisil: Then maybe moving some overlays to seperate repositories manage by subcommunities can lighten the maintenance load?
Jason_Grossman has quit [Remote host closed the connection]
Jason_Grossman has joined #nixos-chat
Jason_Gr` has joined #nixos-chat
Jason_Gr` has quit [Remote host closed the connection]
Jason_Grossman has quit [Ping timeout: 276 seconds]
<andi->
I don't think the maintenance load will change/reduce with splitting stuff... It will probabl if at all increase since you now have to syncrhonize changes between repos
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-chat
<__monty__>
Moving the discussion that relates to a specific subcommunity to a seperate issue/PR tracker would help though, surely?
<gchristensen>
help what?
<__monty__>
Lighten the load on the nixpkgs maintainers.
<__monty__>
Like andi- said, if you're a nixpkgs maintainer who has nothing to do with haskellPackages, all discussion/issues/PRs about that are noise making maintenance harder/more of a chore or at least a tad more annoying.
<__monty__>
(I'm paraphrasing, that's not what andi- said exactly.)
<andi->
I don't think it is annoying everyone.. I just sometimes get upset about stuff where I have not knowledge of.. (can be ignored ;-)
Lisanna has quit [Ping timeout: 264 seconds]
ma27 has quit [Ping timeout: 264 seconds]
ma27 has joined #nixos-chat
NinjaTrappeur has quit [Ping timeout: 245 seconds]
pie__ has quit [Remote host closed the connection]
pie__ has joined #nixos-chat
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-chat
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-chat
NinjaTrappeur has joined #nixos-chat
__monty__ has quit [Quit: leaving]
NinjaTrappeur has quit [Ping timeout: 245 seconds]