geekosaur 2017-02-10 16:45:39
the lambda is being run in an IO context, so it can't get at the WaiSession
geekosaur 2017-02-10 16:45:54
I suspect you cannot arrange for the WaiSession to be available at that point
boccato 2017-02-10 16:46:31
Isn't the delete creating the WaiSession?
geekosaur 2017-02-10 16:53:00
no, it wants to run inside one. (getLine likewise does not "create an IO")
notfed 2017-02-10 16:53:53
oh this is where it's at, DANG there's a lotta people in here
notfed 2017-02-10 16:55:13
I'm a newb to IRC, can someone explain how there can be 100+ people in a chat room and absolutely no one talking? I must be doing it wrong.
Xe 2017-02-10 16:55:51
100 clients in a room doesn't mean 100 people looking at this channel actively
jle` 2017-02-10 16:56:14
notfed: most of the people just idle/leave their client logged on while they do their normal stuff
notfed 2017-02-10 16:56:22
True enough I suppose.
boccato 2017-02-10 16:57:43
This one works: https://gist.github.com/boccato/f6de52c0ae18cd31e1972f029e5064da
nitrix 2017-02-10 17:00:29
notfed: Another plausible explanation more specific to #haskell is that we attempt to not step on each others while we talk and also only talk when questions are asked. We have #haskell-offtopic for the noise.
boccato 2017-02-10 17:01:16
I don't understand the concept of running inside.
dmwit 2017-02-10 17:01:54
notfed: The top 10 most active people in here account for about 1/3 of all the content.
boccato 2017-02-10 17:01:59
As I understand, getLine returns (evaluates to) a value of the type IO String.
dmwit 2017-02-10 17:02:15
notfed: So it's one of those long tail sorts of things.
nitrix 2017-02-10 17:02:52
dmwit: 80:20 https://en.wikipedia.org/wiki/Pareto_principle ?
dmwit 2017-02-10 17:03:29
oh no
dmwit 2017-02-10 17:03:44
Way more than 80% of the activity comes from the top 20%.
dmwit 2017-02-10 17:04:04
5408 nicks have been in here in the last month.
nitrix 2017-02-10 17:04:07
20% of 1442 gives 288 people.
Welkin 2017-02-10 17:04:25
I'd probably say it's less than that
Welkin 2017-02-10 17:04:34
maybe 100
dmwit 2017-02-10 17:04:40
http://ircbrowse.net/nicks/haskell
notfed 2017-02-10 17:04:54
Nitrix, I was thinking perhaps no one wanted to say anything lest they change the state of the chat.
dmwit 2017-02-10 17:05:34
100% of the content in the last month came from the top 20% of nicks.
nitrix 2017-02-10 17:05:44
How is ertes speaking more than the bot! He must not be human.
jle` 2017-02-10 17:05:50
notfed: i see what you did there
dmwit 2017-02-10 17:05:50
5408 nicks visited, 1102 nicks spoke
Welkin 2017-02-10 17:06:18
ertes has more than lambdabot lol
Lokathor 2017-02-10 17:07:39
https://github.com/Lokathor/scratch/blob/master/RustForHaskellUsers/Errors.md not everyone knows rust, but if I made any obvious Haskell blunders that'd be nice to know
Welkin 2017-02-10 17:08:57
Lokathor: looks interesting, thanks
Welkin 2017-02-10 17:09:03
I just started learning rust this week
timbod7 2017-02-10 17:13:49
Is it reasonable that lts-7.13 uses ghc-8.0.1, and yet specifies Cabal-1.24.2.0, whilst the Cabal built in to ghc-8.0.1 is Cabal-1.24.0.0?
geekosaur 2017-02-10 17:15:15
Cabal can in general be updated safely (few libraries packaged with the compiler can claim this) so yes, it's reasonable
timbod7 2017-02-10 17:15:21
When I try to build the cairo package, I get errors that indicated inconsistent Cabal versions.
geekosaur 2017-02-10 17:16:47
that tends to be because the Cabal linked into its Setup.hs doesn't match the one cabal/stack was built with, which only becomes an issue with Setup.hs that do complex things (cairo's does). it's a known shortcoming that both cabal-install and stack fight with; Setup.hs is a fragile place
timbod7 2017-02-10 17:17:55
geekosaur So that means that I can only build packages like cairo with stackage snapshots that don't override the ghc supplied cabal?
timbod7 2017-02-10 17:18:12
error here, by the way: http://lpaste.net/352359
geekosaur 2017-02-10 17:18:14
I think it has more to do with what stack *itself* is using
geekosaur 2017-02-10 17:18:28
but I don't understand the full interaction.
nitrix 2017-02-10 17:18:30
You could try: stack setup --upgrade-cabal
nitrix 2017-02-10 17:18:33
At your own risks.
geekosaur 2017-02-10 17:20:14
because stack has to communicate with Setup.hs and this communication can break down if the Cabal stack is built with and the Cabal Setup.hs gets built with don't match. stack tries to ensure Setup.hs is built with *its* Cabal instead of the one in the resolver, but this can fail if the Setup.hs also links against other packages (most of the gtk-related packages do this)
geekosaur 2017-02-10 17:20:18
as I understand it
timbod7 2017-02-10 17:20:30
nitrix looking for a manual entry... what is `stack setup --upgrade-cabal` going to do?
Lokathor 2017-02-10 17:20:55
down with gtk! we need to make our own pure-haskell GUI toolkit! rah rah rah!
geekosaur 2017-02-10 17:21:21
I think both the cabal-install and stack devs would like to see gtk disappear :)
geekosaur 2017-02-10 17:21:53
or at least, gtk that relies on Setup.hs that isn't a default
nitrix 2017-02-10 17:21:56
I indeed learned that command while playing with gtk.
geekosaur 2017-02-10 17:22:22
(gtk packages write out gtk bindings on the fly from Setup.hs because there are so many different gtk versions in the wild)
timbod7 2017-02-10 17:23:36
geekosaur I've experienced perhaps related problems, where I wanted to generate code in my adl compiler, but couldn't reliably depend on another package to provide the code generation tool.
timbod7 2017-02-10 17:23:54
It ended up easier to just commit the generated code to the repository.
Lokathor 2017-02-10 17:23:55
i tried that fltk library, it seemed okay, even on windows
nitrix 2017-02-10 17:24:07
timbod7: It'll upgrade the version of Cabal used by stack; possibly fixing the conflicts it has with the Setup.hs file. You can still delete the sandbox if you have no success with it.
Lokathor 2017-02-10 17:24:14
but the compile times are insane. The docs say that they're "large" but that's a gross understatement
geekosaur 2017-02-10 17:24:33
yes, but every linux distro/release has a different gtk release and every gtk release has a slightly different api
timbod7 2017-02-10 17:24:40
Lokathor: I'm kind of committed to gtk for now. At least the cairo library. The haskell Chart library depends on it.
Lokathor 2017-02-10 17:24:59
ah, well, if you're stuck with it
geekosaur 2017-02-10 17:25:00
gtk cares not about useless things like compatibility between releases
timbod7 2017-02-10 17:25:35
nitrix: thanks. Just trying that now.
timbod7 2017-02-10 17:27:13
sigh: need a faster computer. or a faster ghc.
timbod7 2017-02-10 17:40:07
nitrix That appears to have solved the problem. Thank you again!
nitrix 2017-02-10 17:40:17
:]
aarvar 2017-02-10 17:40:20
does this exist somewhere? data Foo z a = End z | Cons a (Foo z a)
