wizard1337 2017-03-07 20:51:09
why does reid barton work on ghc
cocreature 2017-03-07 20:58:04
why shouldn't he?
wizard1337 2017-03-07 20:58:54
seems like a bit of a jump from pure math
cocreature 2017-03-07 20:59:31
some people have multiple interests :)
wizard1337 2017-03-07 21:00:16
is there any relation to his experience at rentech?
wizard1337 2017-03-07 21:00:23
and is he paid to work on ghc?
ezyang 2017-03-07 21:01:33
"because it's fun and intellectually stimulating"
n1 2017-03-07 21:03:49
'pure math' is a matter of perspective!
wizard1337 2017-03-07 21:04:21
what are the major sources of funding for ghc?
Cale 2017-03-07 21:09:30
wizard1337: Well, it's an open source project, but I suppose you could have a look at the places that the developers are working.
wizard1337 2017-03-07 21:09:41
so do people who are actually good with haskell get snapped up
wizard1337 2017-03-07 21:10:05
or is it what i have read, which is that there are tons of people who are great with haskell who are desperate to find a position anywhere
Cale 2017-03-07 21:10:18
It's a bit of both.
wizard1337 2017-03-07 21:10:18
*and are
wizard1337 2017-03-07 21:10:30
those are sorta mutually exclusive
n1 2017-03-07 21:10:37
I believe SPJ works on ghc for microsoft research
wizard1337 2017-03-07 21:10:57
right, so that would be 1 job
Cale 2017-03-07 21:11:06
wizard1337: Well, there are a fair number of us who are employed and spend our day jobs writing Haskell
wizard1337 2017-03-07 21:11:16
i guess it's appropriate to respond with an existence proof?
wizard1337 2017-03-07 21:11:25
;-P
Cale 2017-03-07 21:11:33
I'm one of them :)
wizard1337 2017-03-07 21:12:33
right,... but it's not like there are agencies which will be absorbing pretty much anybody who is tall enough to ride the roller coaster
wizard1337 2017-03-07 21:12:43
or are there...?
Cale 2017-03-07 21:13:23
There's Takt, they've been snapping up lots of people lately :)
Cale 2017-03-07 21:13:41
(They're a startup, but they have quite a lot of funding from what it seems)
wizard1337 2017-03-07 21:14:07
lol ive worked at startups with about 30 times that in funding :)
wizard1337 2017-03-07 21:14:08
30 m is significant but not remarkable
Cale 2017-03-07 21:14:46
Well, it's enough to hire a bunch of people and to do a bunch of business with the other Haskell development companies
Cale 2017-03-07 21:15:01
The entire community is only so large to begin with.
wizard1337 2017-03-07 21:15:13
i see
ezyang 2017-03-07 21:15:14
GHC dev is basically MSR + Well Typed + Army of PhDs
ezyang 2017-03-07 21:15:44
with MSR and other industrial people funding the Well Typed committment
Cale 2017-03-07 21:16:12
A couple people at Obsidian also have been getting into it a bit lately, if a bit in anger :)
wizard1337 2017-03-07 21:16:54
carmack likes haskell
wizard1337 2017-03-07 21:17:15
said that he started a project in it
wizard1337 2017-03-07 21:17:28
but other people working on it stormed his office and said "what the hell is a monad?"
Cale 2017-03-07 21:18:19
Monads are not a big deal. It's just a mathematical name for a pattern that's been showing up in functional programming for much longer than we had a name for it.
wizard1337 2017-03-07 21:18:35
yeah, i'm not saying these people had the right idea
Cale 2017-03-07 21:20:41
The harder part of learning to program in Haskell is usually going to have more to do with the change in mindset required regarding programming with immutable data structures most of the time.
wizard1337 2017-03-07 21:21:12
yaeh
Cale 2017-03-07 21:27:37
wizard1337: Also, when it comes to IO, the fact that IO is a monad, though it seems to be what people often fixate on, is not usually going to be any kind of a problem -- it just means we have a certain API for sticking these things together. The weird part that takes some getting used to is that IO is a type at all, and that we have a type of values which represent effectful programs, and their execution is a separate process from
Cale 2017-03-07 21:27:37
evaluation of expressions.
wizard1337 2017-03-07 21:28:28
i don't fully understand monads but everything i've seen about them so far isn't troublesome
wizard1337 2017-03-07 21:28:45
containing state is a very sensible thing
Cale 2017-03-07 21:28:49
You basically learn a bunch of examples of them
wizard1337 2017-03-07 21:28:51
i did that naturally when i wrote imperative code
wizard1337 2017-03-07 21:29:03
purity, totality, state containment,...
Cale 2017-03-07 21:29:08
and gradually get a good sense that way for what kinds of things to usually expect
maerwald 2017-03-07 21:30:39
what's a nice xml parser in haskell without lens, TH or other complexity I don't care about
wizard1337 2017-03-07 21:30:45
lol
wizard1337 2017-03-07 21:30:58
if you don't like complexity, why are you using xml
maerwald 2017-03-07 21:31:09
because it's the input
Cale 2017-03-07 21:31:21
maerwald: What do you need it for? hexpat-pickle is old, but it might be good.
cocreature 2017-03-07 21:31:25
maerwald: I've used xml-conduit and found it quite nice
maerwald 2017-03-07 21:31:32
Cale: parsing the capec database
cocreature 2017-03-07 21:31:50
you don't need to use conduit for it to be useful
wizard1337 2017-03-07 21:33:20
use kaggle and hire someone to train a neural net to parse xml
rom1504 2017-03-07 21:35:17
great idea
wizard1337 2017-03-07 21:41:05
oh, it's the gghc
wizard1337 2017-03-07 21:41:07
not ghc
