nitrix 2017-01-27 08:45:20
Liskni_si: Which is completely crazy. What `kawaii` depends on shouldn't be none of the executable's business.
nitrix 2017-01-27 08:45:40
Liskni_si: In fact, what happens as soon as the library depends on a different version? Whoops, builds breaking.
Liskni_si 2017-01-27 08:46:00
nitrix: maybe you want to use cabal-install instead of stack, then
nitrix 2017-01-27 08:46:28
Liskni_si: It should be that the executable depends on `kawaii` and that's it. And kawaii should have the dependencies commented out and that would make a lot more sense.
Liskni_si 2017-01-27 08:46:29
nitrix: stack is about reproducible builds, and it won't go all around the internet looking for random packages you might need, it just won't, wasn't designed to
nitrix 2017-01-27 08:46:51
Liskni_si: It isn't going around the internet lookinf for random packages. These are local dependencies.
nitrix 2017-01-27 08:47:19
Liskni_si: `kawaii` is a local dependency of that executable.
Liskni_si 2017-01-27 08:48:07
nitrix: stack doesn't care where those dependencies are. If it doesn't know about the package, it won't see it. And it only knows about what's in the resolver (lts-7.9 in your case) and extra-deps.
nitrix 2017-01-27 08:48:41
Liskni_si: But it ISN'T, an extra dep.
Liskni_si 2017-01-27 08:48:49
nitrix: Again, stack.yaml aren't meant to nest. It isn't looking at stack.yaml in kawaii! It's looking at kawaii.cabal, maybe, but not kawaii/stack.yaml.
nitrix 2017-01-27 08:48:51
It's a package listed as a local dependency.
nitrix 2017-01-27 08:48:58
packages:
nitrix 2017-01-27 08:49:00
- '.'
nitrix 2017-01-27 08:49:02
- kawaii
nitrix 2017-01-27 08:49:04
How do I have to work this.
nitrix 2017-01-27 08:49:06
*word.
nitrix 2017-01-27 08:49:18
It'll, literally, look up that folder.
Liskni_si 2017-01-27 08:49:31
It doesn't matter what kind of dependency it is, local or remote or nonexistent.
nitrix 2017-01-27 08:49:48
Well, that is bogus.
nitrix 2017-01-27 08:49:55
Dependencies should be self-contained.
Liskni_si 2017-01-27 08:50:20
If you come from rebar3, you might well expect them to be, yeah. :-)
Liskni_si 2017-01-27 08:50:53
If you come from anything else, say, cpan, maven, whatever, not so much.
nitrix 2017-01-27 08:51:35
I can understand for packages that are external (on hackage) or internal (on stackage), but here we clearly state that kawaii can checkout this repository and build the library using these files.
nitrix 2017-01-27 08:51:59
Then the executable should just be able to use the library that was just built as, that library, is a local dependency.
nitrix 2017-01-27 08:52:16
It can just use it. It doesn't make sense for the executable to re-list all the stuff that that library needs.
Liskni_si 2017-01-27 08:53:08
I'm not sure if you want an explanation of why things are the way they are or a way to solve this without sacrificing your beliefs. :-/
Liskni_si 2017-01-27 08:53:22
Either way, I'm hungry, so I can't give you either. :-(
nitrix 2017-01-27 08:53:49
Alright. I'll see with the FPComplete dev team.
ph88 2017-01-27 09:11:01
monochrom, i've checked the docs it says i don't need an explicit forall in ghc and i already have ScopedTypeVariables extension enabled
monochrom 2017-01-27 09:11:28
http://www.vex.net/~trebla/humour/tautologies.html #4
ph88 2017-01-27 09:13:16
ya that's right, i don't know the cause -___-
geekosaur 2017-01-27 09:14:11
something like this was discussed in #ghc recently, iirc currently there is no way to synchronize those type variables
ph88 2017-01-27 09:16:17
is there a workaround ?
geekosaur 2017-01-27 09:19:28
http://ircbrowse.net/browse/ghc?id=44779×tamp=1484535315#t1484535315 ff. was discussion I was remembering
ph88 2017-01-27 09:20:36
thanks, ill take a look
geekosaur 2017-01-27 09:20:37
ph88, what ghc version?
ph88 2017-01-27 09:20:42
8.0.1
geekosaur 2017-01-27 09:20:51
http://ircbrowse.net/browse/ghc?id=44866×tamp=1484536793#t1484536793
geekosaur 2017-01-27 09:21:10
is the relevant part, I *think* this is something that changed in 8.0.2
geekosaur 2017-01-27 09:21:43
but not directly related to your issue, just doing something you might benefit from as a side effect
byorgey 2017-01-27 09:24:14
dgpratt: um. There are probably, like, thousands.
ph88 2017-01-27 09:24:21
i keep being surprised that there is still heavy development activity in ghc ^^
byorgey 2017-01-27 09:24:30
oops, I was scrolled up =P
ph88 2017-01-27 09:24:50
by the way, is there also a forum for questions ?
geekosaur 2017-01-27 09:26:46
glasgow-haskell-users mailing list, possibly
AWizzArd 2017-01-27 09:29:37
Control.Monad.forM vs mapM – what's the difference?
monochrom 2017-01-27 09:29:52
parameter order
monochrom 2017-01-27 09:29:59
is that right/
monochrom 2017-01-27 09:30:03
@type forM
lambdabot 2017-01-27 09:30:05
(Traversable t, Monad m) => t a -> (a -> m b) -> m (t b)
monochrom 2017-01-27 09:30:16
yeah, parameter order
AWizzArd 2017-01-27 09:30:28
So it's just interesting for style? If I have a big body I use forM, otherwise if it's just a function name mapM?
monochrom 2017-01-27 09:30:41
yes
zennist 2017-01-27 09:30:48
For composition and product between functors / applicatives - what are the conventional constructors to use?
AWizzArd 2017-01-27 09:31:01
monochrom: oki thx
zennist 2017-01-27 09:31:45
I found: for composition: Data.Functor.Compose or :.:; for Product, only found :*: ?
ph88 2017-01-27 09:31:47
geekosaur, what i see there in the chat and also what i read in the documentation are examples whereby the type defined in the class is matched with the type of the function in the instance. But in my case i just want to specify the type of a function that resides in another instance (of the same class nevertheless). That's why i can't really relate those things back to my situation
zennist 2017-01-27 09:32:35
I suppose (:.:) and (:*:) are more consistent?
geekosaur 2017-01-27 09:32:42
the most relevant part was discussing just scoping every type variable automatically over the whole thing, instead of the hack added to 8.0.2 that scoped certain type variables
monochrom 2017-01-27 09:32:47
You found Data.Functor.Compose but not Data.Functor.Product?
geekosaur 2017-01-27 09:33:25
(that is not something that is currently implemented)
tibbe 2017-01-27 09:34:15
How do I get the Emacs haskell-mode to use stack instead of cabal?
zennist 2017-01-27 09:37:02
actually - I found Data.Functor.Product; maybe this is better
monochrom 2017-01-27 09:39:30
The X product :>*<: and its duel, the TIE product :<*>:
clmg 2017-01-27 09:42:14
How do cursors/outputs work in vty? When I setCursorPos via `outputIfFace vty` I don't see a cursor.
Tuplanolla 2017-01-27 09:42:59
I was about the ask the same thing today, clmg.
clmg 2017-01-27 09:44:40
Vty is great so far. I just wish vty-examples was more complete.
Tuplanolla 2017-01-27 09:44:59
I'm using it with Brick here. It's quasipleasant.
