Logs: freenode/#haskell

←Prev Next→

Page 1 .. 879 880 881 882 883 884 885 886 887 888 889 .. 5022

— 502,152 events total

2020-10-26 20:47:17	→	glguy_ joins (x@freenode/staff/haskell.developer.glguy)
2020-10-26 20:47:17	glguy_	is now known as glguy
2020-10-26 20:47:21	→	justHaunted joins (~justache@unaffiliated/justache)
2020-10-26 20:47:40	×	blackdog quits (~blackdog@198.211.112.85) (Ping timeout: 246 seconds)
2020-10-26 20:48:03	<noname234234>	I have imagined product for and operation and sum for `or` operation
2020-10-26 20:48:04	→	justanotheruser joins (~justanoth@unaffiliated/justanotheruser)
2020-10-26 20:48:14	glguy	is now known as ghoulguy
2020-10-26 20:48:16	→	aldessa joins (~Hugh@host-92-21-7-179.as13285.net)
2020-10-26 20:48:26	<noname234234>	I believe my reasoning is correct
2020-10-26 20:48:35	<noname234234>	but I don't really know how would I express this in haskell
2020-10-26 20:49:01	<tomsmeding>	you managed Sum, didn't you? :)
2020-10-26 20:49:02	→	srid joins (sridmatrix@gateway/shell/matrix.org/x-mkesouyylxzbcafd)
2020-10-26 20:49:20	→	MikeD joins (ad5822ca@cpe-173-88-34-202.columbus.res.rr.com)
2020-10-26 20:49:25	→	alvinsj[m] joins (alvinsjmat@gateway/shell/matrix.org/x-vcggqutodlsvbbci)
2020-10-26 20:49:38	<tomsmeding>	though it's more natural to write: newtype Sum = Sum { getSum :: Bool }
2020-10-26 20:49:44	<tomsmeding>	instead of parametrising on a variable `a'
2020-10-26 20:50:06	→	blackdog joins (~blackdog@198.211.112.85)
2020-10-26 20:50:09	<noname234234>	https://paste.tomsmeding.com/w73fxoro
2020-10-26 20:50:24	<dminuoso>	noname234234: Your code is almost spot on, really. If you want to just "get it done", enable the extension by adding `{-# LANGUAGE FlexibleInstances #-}` to the top
2020-10-26 20:50:32	<dminuoso>	Or change the newtype like tomsmeding has suggested
2020-10-26 20:50:41	→	materialfuture[m joins (materialfu@gateway/shell/matrix.org/x-jvhencngpnxshide)
2020-10-26 20:51:02	→	jiribenes1 joins (jbjiribene@gateway/shell/matrix.org/x-fxavqhwgliqrsecu)
2020-10-26 20:51:03	→	tersetears[m] joins (tersetears@gateway/shell/matrix.org/x-njimwnsggeorbzzd)
2020-10-26 20:51:09	→	wi[m] joins (w1gzmatrix@gateway/shell/matrix.org/x-yfarknebhuzfmgwp)
2020-10-26 20:51:16	<tomsmeding>	I'd say, because this Sum data type is intended to be used only with Bool, I think, I would write it as I suggested
2020-10-26 20:51:46	→	mmynsted[m] joins (mmynstedtc@gateway/shell/matrix.org/x-zszjcyokedzkvvvw)
2020-10-26 20:51:58	→	steve[m] joins (stevetrout@gateway/shell/matrix.org/x-fkuthzdpqeujhzmp)
2020-10-26 20:52:03	<dminuoso>	(Note I didnt bother to check whether your boolean algebra is right, but oh well, that's on you :-)
2020-10-26 20:52:04	<noname234234>	so I want to understand the real issue here let me write it as I understood -> currently generic types are not supported in instance definition in haskell? sorry for `generic type` term but I see that as `a` I defined there
2020-10-26 20:52:14	<dminuoso>	noname234234: So
2020-10-26 20:52:21	→	michaelpj joins (michaelpjm@gateway/shell/matrix.org/x-oigadqxzdyumsikb)
2020-10-26 20:52:23	<dminuoso>	If we consider a general intsance declaration as:
2020-10-26 20:52:35	<tomsmeding>	dminuoso: side note: why didn't the haskell spec allow "flexible instances", so to speak? Didn't they want to mess with overlap detection algorithms?
2020-10-26 20:52:36	→	lnxw37d4 joins (lnxw37d4ma@gateway/shell/matrix.org/x-amgtuukfolghuwuq)
2020-10-26 20:52:38	<dminuoso>	instance ClassName (TypeConstructor t1 t2 t3 ...)
2020-10-26 20:52:38	→	maralorn joins (maralornma@gateway/shell/matrix.org/x-pkelfgrohvpzfolu)
2020-10-26 20:52:49	<dminuoso>	then t1, t2, t3, etc must only be type variables
2020-10-26 20:52:59	<dminuoso>	they may not be type constructors like Bool
2020-10-26 20:53:18	<noname234234>	type variable is typeclasses?
2020-10-26 20:53:18	<dminuoso>	tomsmeding: I dont know for sure, but I think there's two parts to it
2020-10-26 20:53:40	<dminuoso>	tomsmeding: For one, FlexibleInstances essentially incurs FlexibleContexts, and you can write incoherent code with that alone (see the GHC manual)
2020-10-26 20:53:44	tomsmeding	would say "type names", not "type constructors"; I've seen someone use "type constructor" where they meant "data constructor" in this channel
2020-10-26 20:54:00	<dminuoso>	Another one I've heard mentioned is that their implementation is non-trivial
2020-10-26 20:54:13	<dminuoso>	At least if you didn't know how to do it beforehand.
2020-10-26 20:55:21	<monsterchrom>	I avoid "type constructor" because people then go anal about whether Int counts, and independently whether (Either Int) counts.
2020-10-26 20:55:21	<dminuoso>	(It takes some real effort to write incoherent code with it, which is why I called it fairly benigh above)
2020-10-26 20:55:46	<dminuoso>	monsterchrom: a type constructor is something you can pattern match on in tyfams/tyinsts!
2020-10-26 20:55:53	<dminuoso>	just like data constructors can be pattern matched on in case-of
2020-10-26 20:55:55	<dminuoso>	:p
2020-10-26 20:56:02	<noname234234>	so type variable is for example `Num ` class?
2020-10-26 20:56:29	<tomsmeding>	noname234234: no, that's a class name ;)
2020-10-26 20:56:36	<tomsmeding>	it's something like `a`
2020-10-26 20:56:43	<noname234234>	ah oaky
2020-10-26 20:56:53	<tomsmeding>	like the `a` in your `newtype Sum a = Sum { getSum :: a }`
2020-10-26 20:57:31	<noname234234>	ah okay its like generic type
2020-10-26 20:57:34	<noname234234>	inother languages
2020-10-26 20:57:39	<dminuoso>	noname234234: no
2020-10-26 20:57:54	<dminuoso>	a type variable is the `a` itself, not what you can do with it
2020-10-26 20:57:57	<tomsmeding>	it's like the type parameter to a generic type/function in other languages, I guess
2020-10-26 20:58:01	<dminuoso>	% :t id
2020-10-26 20:58:01	<yahb>	dminuoso: a -> a
2020-10-26 20:58:12	<dminuoso>	noname234234: in that type signature above, the `a` is sort of like a placeholder
2020-10-26 20:58:20	<tomsmeding>	in java or c++, the T in the <>'s
2020-10-26 20:58:40	<dminuoso>	noname234234: so a type variable means "you can pick whatever you want there"
2020-10-26 20:58:41	<noname234234>	okay thank you
2020-10-26 20:58:47	<dminuoso>	Bool
2020-10-26 20:58:48	×	justHaunted quits (~justache@unaffiliated/justache) (Read error: Connection reset by peer)
2020-10-26 20:58:49	<dminuoso>	is not a type variable
2020-10-26 20:58:58	<dminuoso>	It's a concrete/particular type
2020-10-26 20:59:37	<noname234234>	aha get it
2020-10-26 20:59:54	<noname234234>	got it*, so I will now work on the example,
2020-10-26 20:59:56	<tomsmeding>	dminuoso: I don't see a mention of being able to write incoherent code with FlexibleContexts in the ghc user manual; am I looking in the wrong place? https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/glasgow_exts.html#extension-FlexibleContexts
2020-10-26 20:59:57	×	conal quits (~conal@2600:380:444d:d3f4:d82:fca6:5a7c:8afb) (Quit: Computer has gone to sleep.)
2020-10-26 21:00:02	×	DrGuschtel quits (~DrGuschte@84.39.116.180) ()
2020-10-26 21:00:11	<dminuoso>	tomsmeding: See rules on overlap
2020-10-26 21:00:34	<tomsmeding>	including {-# OVERLAPPABLE #-} pragmas and such? I'd not say that counts
2020-10-26 21:00:43	<dminuoso>	https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/glasgow_exts.html#extension-FlexibleInstances
2020-10-26 21:00:48	<dminuoso>	https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/glasgow_exts.html#instance-overlap
2020-10-26 21:01:01	<dminuoso>	no, without overlap even
2020-10-26 21:01:24	×	mud quits (~mud@unaffiliated/kadoban) (Remote host closed the connection)
2020-10-26 21:01:25	<dminuoso>	(I think)
2020-10-26 21:01:38	×	GyroW quits (~GyroW@unaffiliated/gyrow) (Quit: Someone ate my pie)
2020-10-26 21:02:06	<dminuoso>	And, with the bug report I opened recently, there's a lot of awkwardness with typeclass instance resolution, where it's not even clear what the semantics are
2020-10-26 21:02:15	→	GyroW joins (~GyroW@ptr-48ujrfd1ztq5fjywfw3.18120a2.ip6.access.telenet.be)
2020-10-26 21:02:16	×	GyroW quits (~GyroW@ptr-48ujrfd1ztq5fjywfw3.18120a2.ip6.access.telenet.be) (Changing host)
2020-10-26 21:02:16	→	GyroW joins (~GyroW@unaffiliated/gyrow)
2020-10-26 21:02:29	<dminuoso>	if you look at the haskell report, it doesnt say anything about what context of typeclass instances mean/do
2020-10-26 21:02:40	<tomsmeding>	interesting
2020-10-26 21:02:59	<dminuoso>	tomsmeding: https://gitlab.haskell.org/ghc/ghc/-/issues/18836
2020-10-26 21:03:15	<noname234234>	now I get this error -> `• No instance for (Semigroup Product) arising from the superclasses of an instance declaration• In the instance declaration for ‘Monoid Product’typecheck`
2020-10-26 21:03:27	<noname234234>	for this -> `-- \| Monoid under multiplication.newtype Product = Product { getProduct :: Bool }instance Monoid Product where mempty = Product True Product x `mappend` Product y = Product (x && y)`
2020-10-26 21:03:27	<dminuoso>	A quick look at that note alone makes you wonder how subtle typeclass instance implementations really are. :-)
2020-10-26 21:03:38	<dminuoso>	noname234234: Ah, that might be friction due to changes in GHC Haskell
2020-10-26 21:03:43	×	britva quits (~britva@31-10-157-156.cgn.dynamic.upc.ch) (Quit: This computer has gone to sleep)
2020-10-26 21:03:55	<dminuoso>	noname234234: We split Monoid into Semigroup and Monoid, where mappend was moved into Semigroup as <>
2020-10-26 21:04:13	→	conal joins (~conal@2600:380:444d:d3f4:d82:fca6:5a7c:8afb)
2020-10-26 21:04:13	×	conal quits (~conal@2600:380:444d:d3f4:d82:fca6:5a7c:8afb) (Client Quit)
2020-10-26 21:04:19	→	invaser joins (~Thunderbi@31.148.23.125)
2020-10-26 21:04:28	<dminuoso>	So when you wrote `instance Monoid Foo where mappend = ...; mempty = ...;` you'd now write `instance Semigroup Foo where l <> r = ...; instance Monoid Foo where mempty = ...`
2020-10-26 21:05:10	<noname234234>	I have to separate these two?

←Prev Next→

Page 1 .. 879 880 881 882 883 884 885 886 887 888 889 .. 5022

All times are in UTC.