Logs: freenode/#haskell

←Prev Next→

Page 1 .. 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 .. 5022

— 502,152 events total

2020-11-26 14:13:00	hackage	lsp 1.0.0.1 - Haskell library for the Microsoft Language Server Protocol https://hackage.haskell.org/package/lsp-1.0.0.1 (luke_)
2020-11-26 14:13:53	→	acarrico joins (~acarrico@dhcp-68-142-39-249.greenmountainaccess.net)
2020-11-26 14:14:40	→	sondr3 joins (~sondr3@cm-84.211.56.132.getinternet.no)
2020-11-26 14:19:02	×	hekkaidekapus_ quits (~tchouri@gateway/tor-sasl/hekkaidekapus) (Quit: hekkaidekapus_)
2020-11-26 14:21:06	→	hekkaidekapus joins (~tchouri@gateway/tor-sasl/hekkaidekapus)
2020-11-26 14:21:44	<dminuoso>	maerwald: Anyway, interesting. If I dont case-of, I usually have some `note` function instead.
2020-11-26 14:22:25	<dminuoso>	Or I move it into pattern matching on separate functions, where I might have `fooEither (Left ...) = ...; fooEither (Right ...) = ...'` and use that as a continuation
2020-11-26 14:22:38	<dminuoso>	But either/maybe seems just strange to me for some reason I cant explain
2020-11-26 14:22:58	<boxscape>	do you use foldr?
2020-11-26 14:23:57	<dminuoso>	sure
2020-11-26 14:24:02	<boxscape>	hm
2020-11-26 14:24:44	<boxscape>	So an aversion to Church encodings of non-recursive types but not to those of recursive types, to overly extrapolate
2020-11-26 14:24:56	<dminuoso>	Yes, very much.
2020-11-26 14:26:51	×	jonatanb_ quits (~jonatanb@83.24.220.252.ipv4.supernova.orange.pl) (Remote host closed the connection)
2020-11-26 14:27:14	<dminuoso>	The non-recursive types encode a choice, list does not
2020-11-26 14:27:34	<dminuoso>	Or I guess that depends on the perspective
2020-11-26 14:28:10	<dminuoso>	since a list could be `n + (n^2) + (n^3) + (n^4) + ....`
2020-11-26 14:28:36	<dminuoso>	Maybe it's just that to me it's not clear which side of `maybe` does what
2020-11-26 14:28:51	<dminuoso>	Perhaps I also just lack the intuition because I dont use it
2020-11-26 14:28:58	<c_wraith>	doesn't the type tell you?
2020-11-26 14:29:08	<maerwald>	dminuoso: if I convert a maybe to some other type (e.g. Either), I find maybe to be quite idiomatic... but I can't stand fromMaybe
2020-11-26 14:29:11	<c_wraith>	:t maybe
2020-11-26 14:29:12	<lambdabot>	b -> (a -> b) -> Maybe a -> b
2020-11-26 14:29:31	<c_wraith>	there's only one case either of those arguments can apply in.
2020-11-26 14:29:49	<dminuoso>	c_wraith: Sure, but when you look at `maybe foo bar h`, then it's just not obvious to me
2020-11-26 14:29:57	<c_wraith>	Now, I understand that complaint about bool....
2020-11-26 14:30:05	<dminuoso>	Mmm
2020-11-26 14:30:25	<maerwald>	I always mix up second and third parameter of foldr/foldl :p
2020-11-26 14:30:30	<hpc>	dminuoso: there's a way to remember it as a specific instance of a more general thing
2020-11-26 14:30:38	<hpc>	data Maybe a = Nothing \| Just a
2020-11-26 14:30:46	<hpc>	the parameters for maybe go in the same order as the data definition
2020-11-26 14:30:53	<dminuoso>	hpc: Sure, the first constructor comes first. But that requires memorizing the data definition
2020-11-26 14:30:54	<hpc>	same for foldr, they go in the same order
2020-11-26 14:31:11	<hpc>	dminuoso: the smallest constructor is first
2020-11-26 14:31:30	<hpc>	(for deriving Ord and such)
2020-11-26 14:31:32	<dminuoso>	hpc: so what about These?
2020-11-26 14:31:39	→	fendor joins (~fendor@77.119.130.234.wireless.dyn.drei.com)
2020-11-26 14:31:40	<dminuoso>	What type signature would `these` have?
2020-11-26 14:32:10	×	darjeeling_ quits (~darjeelin@122.245.217.23) (Ping timeout: 256 seconds)
2020-11-26 14:32:17	<dminuoso>	these :: (a -> d) -> (b -> d) -> (a -> b -> d) -> These a b -> d
2020-11-26 14:32:18	<c_wraith>	intuition says: (a -> c) -> (b -> c) -> (a -> b -> c) -> These a b -> c
2020-11-26 14:32:21	<dminuoso>	these :: (b -> d) -> (a -> d) -> (a -> b -> d) -> These a b -> d
2020-11-26 14:32:41	<dminuoso>	According to your definition, Id have to look up the definition of These, and I couldnt infer it from the size
2020-11-26 14:32:42	<maerwald>	easy: this, that, these
2020-11-26 14:32:44	<maerwald>	xD
2020-11-26 14:33:24	<hpc>	it derives Ord so you can check in ghci
2020-11-26 14:33:44	<dminuoso>	Well if I have to look it up, I might as well just check hoogle
2020-11-26 14:33:46	<boxscape>	% :i These
2020-11-26 14:33:47	<yahb>	boxscape: ; <interactive>:1:1: error: Not in scope: `These'
2020-11-26 14:33:52	<boxscape>	:/
2020-11-26 14:34:00	<dminuoso>	% import Data.Strict.These
2020-11-26 14:34:00	<yahb>	dminuoso: ; <no location info>: error:; Could not find module `Data.Strict.These'; It is not a module in the current program, or in any known package.
2020-11-26 14:34:08	<dminuoso>	% import Data.Functor.These
2020-11-26 14:34:08	<yahb>	dminuoso: ; <no location info>: error:; Could not find module `Data.Functor.These'; Perhaps you meant; Data.Functor.Base (from recursion-schemes-5.1.3); Data.Functor.Const (from base-4.14.0.0); Data.Functor.Plus (from semigroupoids-5.3.4)
2020-11-26 14:34:42	<dminuoso>	hpc: I get your point, Im not saying its not discoverable, Im just suggesting that from just staring at its usage, it might not be obvious
2020-11-26 14:34:48	<hpc>	sometimes
2020-11-26 14:34:54	<dminuoso>	But arguably the same could be said about `foldr`
2020-11-26 14:35:01	<c_wraith>	and every function
2020-11-26 14:35:01	<hpc>	but then for example, how would you write a fold over a tree?
2020-11-26 14:35:08	<dminuoso>	hpc: easy! foldMap!
2020-11-26 14:35:12	<dminuoso>	;)
2020-11-26 14:35:15	<hpc>	heh
2020-11-26 14:36:39	<hpc>	but yeah, if you know the structure it's pretty easy, foldTree leaf branch Leaf = leaf; foldTree leaf branch (Branch value left right) = branch value (foldTree left) (foldTree right)
2020-11-26 14:36:57	<hpc>	or something similar depending on what type of tree you're dealing with
2020-11-26 14:38:50	→	jonatanb joins (~jonatanb@83.24.220.252.ipv4.supernova.orange.pl)
2020-11-26 14:38:51	<hpc>	i wonder if there's anything that writes these functions automatically
2020-11-26 14:38:54	→	bitmapper joins (uid464869@gateway/web/irccloud.com/x-ntgjmwmudxzzxssg)
2020-11-26 14:40:11	<c_wraith>	recursion-schemes?
2020-11-26 14:40:26	<merijn>	c_wraith: That's not automatic
2020-11-26 14:40:32	<c_wraith>	it has TH
2020-11-26 14:40:34	<merijn>	That's just "reusing handwritten ones"
2020-11-26 14:41:06	<c_wraith>	every library is reusing something someone wrote by hand...
2020-11-26 14:42:19	<merijn>	c_wraith: Right, but I meant as opposed to DeriveFunctor. I guess TH counts, but I didn't know recursion-schemes had that, because my brain doesn't like recursion schemes :p
2020-11-26 14:42:43	×	jonatanb quits (~jonatanb@83.24.220.252.ipv4.supernova.orange.pl) (Remote host closed the connection)
2020-11-26 14:42:55	<c_wraith>	And I don't like using TH. :P
2020-11-26 14:43:01	dminuoso	likes TH
2020-11-26 14:43:16	<c_wraith>	I like writing it. But I don't like writing code that uses it.
2020-11-26 14:43:24	<merijn>	I like TH, but they fucked up the specification
2020-11-26 14:43:31	<dminuoso>	There is a specification?
2020-11-26 14:43:34	<dminuoso>	:>
2020-11-26 14:43:40	<merijn>	dminuoso: My point
2020-11-26 14:44:03	<dminuoso>	My main annoyance with TH is just one: staging restriction
2020-11-26 14:44:06	<merijn>	TH should've had a clear "target" and "host" distinction for everything
2020-11-26 14:44:09	<hpc>	hmm, that technically writes it but by way of free shennanigans
2020-11-26 14:44:21	<dminuoso>	Having to fiddle with extra modules just to satisfy the staging restriction is sometimes frustrating
2020-11-26 14:44:36	<dminuoso>	Especially because we cant have circular module references
2020-11-26 14:44:38	<merijn>	dminuoso: Staging is annoying, but you can workaround it
2020-11-26 14:44:57	<hpc>	like if i was using [] and had to pick between foldr and whatever recursion-schemes gives me, recursion-schemes would be harder to use
2020-11-26 14:45:07	<merijn>	dminuoso: In contrast to the lack of distinction between target/host which makes cross-compiled TH fundamentally impossible
2020-11-26 14:45:18	<dminuoso>	merijn: cross compiling is not something I care about
2020-11-26 14:45:27	<merijn>	(well, the Asterius guys seem to be attempting to fix/work around it via heroic amounts of work)
2020-11-26 14:45:29	<hpc>	i just want to write data Foo ..., makeFold 'Foo, and get foo :: whatever -> Foo ... -> ...
2020-11-26 14:45:34	<merijn>	dminuoso: Not yet
2020-11-26 14:45:44	<dminuoso>	Perhaps
2020-11-26 14:45:53	<merijn>	dminuoso: Compiling windows executables from linux/etc. would've been nice :p
2020-11-26 14:45:58	<dminuoso>	Is that some reference to the dominion of ARM processors?
2020-11-26 14:46:00	<dminuoso>	Oh
2020-11-26 14:46:05	<merijn>	dminuoso: ARM too
2020-11-26 14:46:32	<merijn>	dminuoso: Everything is just slightly more inconvenient without cross-compilation
2020-11-26 14:47:04	<merijn>	And proper cross-compilation support isn't even that hard. It's just a herculean task to retrofit cross-compilation support if you didn't engineer it in from the beginning

←Prev Next→

Page 1 .. 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 .. 5022

All times are in UTC.