Logs: freenode/#haskell

←Prev Next→

Page 1 .. 29 30 31 32 33 34 35 36 37 38 39 .. 5022

— 502,152 events total

2020-09-17 18:04:57	<dumptruckman>	s/define/
2020-09-17 18:04:59	<ski>	yes, that's how instances work
2020-09-17 18:05:05	<dumptruckman>	well yeah..
2020-09-17 18:05:07	<tomsmeding>	well, like, there's generics
2020-09-17 18:05:31	<dumptruckman>	but like, i don't have to implement anything to use Show for most stuff
2020-09-17 18:05:42	<tomsmeding>	because Show is hard-wired :)
2020-09-17 18:05:43	<dumptruckman>	so it has some kind of default implementation that "just works"?
2020-09-17 18:05:43	<ski>	because `Show' can be derived
2020-09-17 18:05:55	<tomsmeding>	you can use generics to do what you want, I think
2020-09-17 18:06:02	<ski>	(and, usually, should be derived)
2020-09-17 18:06:03	<tomsmeding>	but it's not very pretty
2020-09-17 18:06:04	<geekosaur>	but you must derive it for it to be there
2020-09-17 18:06:14	<dumptruckman>	right
2020-09-17 18:06:28	<dumptruckman>	so i guess it's not an instance of the typeclass if i simply derive that typeclass
2020-09-17 18:06:55	<dumptruckman>	so i suppose i'm wondering if there's a way to do this without making my data types an instance of it
2020-09-17 18:07:05	<monochrom>	Rather, "deriving" is not for arbitrary classes of your own design.
2020-09-17 18:07:25	<tomsmeding>	except for obscure ghc extensions mumble mumble
2020-09-17 18:07:40	→	eric joins (~eric@2804:431:c7d4:b75:69f1:6cfd:8841:8fad)
2020-09-17 18:07:54	<tomsmeding>	but, yes, in vanilla haskell, "deriving" is only for built-in things
2020-09-17 18:08:42	<tomsmeding>	dumptruckman: alternative track: what if you define a type 'data CachedShow a = CachedShow String a'
2020-09-17 18:08:44	<dumptruckman>	ahh
2020-09-17 18:08:52	→	nyd joins (~nyd@unaffiliated/elysian)
2020-09-17 18:09:17	<tomsmeding>	and then define 'data Something a = Something (CachedShow a)'
2020-09-17 18:09:20	<dumptruckman>	ahhh
2020-09-17 18:09:23	<dumptruckman>	that's a good idea
2020-09-17 18:09:25	<tomsmeding>	where that 'a' is whatever you want of course
2020-09-17 18:09:30	<sm[m]>	I got an SDL app working in GHCI by using -fno-ghci-sandbox per https://stackoverflow.com/questions/61842338/creating-an-sdl2-window-inside-ghci . But I'm having trouble getting rid of the window when the app exits. destroyWindow having no effect. Has anyone got this working ?
2020-09-17 18:09:49	<tomsmeding>	then if you say 'instance Show (CachedShow a) where show (CachedShow s _) = s'
2020-09-17 18:09:57	<tomsmeding>	and just use 'deriving (Show)' on Something
2020-09-17 18:10:14	<tomsmeding>	then you get 'Something [whatever that string has]', which is not exactly what you asked, but it's close
2020-09-17 18:10:31	<tomsmeding>	downside is that now there's a CachedShow constructor in between the Something constructor and the contained data
2020-09-17 18:10:41	<dumptruckman>	or i could simply do: `data Name = Name String` `name :: Name -> String` ` name (Name s) = s` `data Something = Something name a b c`
2020-09-17 18:10:43	<dumptruckman>	i think
2020-09-17 18:10:43	<tomsmeding>	but if you're okay with that, this is a very vanilla way of sort-of getting that :)
2020-09-17 18:11:09	<tomsmeding>	well 'Something Name a b c', not 'Something name a b c', but yes
2020-09-17 18:11:17	<dumptruckman>	yes
2020-09-17 18:11:20	<dumptruckman>	that's what i meant
2020-09-17 18:11:22	<tomsmeding>	but then you still need to implement Show for Something yourself
2020-09-17 18:11:22	<dumptruckman>	oh but wait
2020-09-17 18:11:29	<tomsmeding>	so you haven't won anything
2020-09-17 18:11:32	<dumptruckman>	and i'd still have to implement name
2020-09-17 18:11:40	<dumptruckman>	name :: Show -> String
2020-09-17 18:11:48	<dumptruckman>	er, name :: Something -> String
2020-09-17 18:11:49	<ski>	tomsmeding : that would be bad, imho :)
2020-09-17 18:11:55	→	fendor joins (~fendor@77.119.129.27.wireless.dyn.drei.com)
2020-09-17 18:11:59	→	Gigabitten joins (~Somn@ip-99-203-18-60.pools.cgn.spcsdns.net)
2020-09-17 18:12:01	<tomsmeding>	the point of my solution, with wrapping the rest of the Something contents _inside_ that Name, you can just 'deriving (Show)' and it works
2020-09-17 18:12:18	<tomsmeding>	ski: yes you haven't won anything and you've made it more complicated :p
2020-09-17 18:12:25	<tomsmeding>	or are you referring to my CachedShow thing
2020-09-17 18:12:26	<ski>	(imho, that `Show' instance is incorrect)
2020-09-17 18:12:27	<ski>	yes
2020-09-17 18:12:29	×	eric quits (~eric@2804:431:c7d4:b75:69f1:6cfd:8841:8fad) (Ping timeout: 272 seconds)
2020-09-17 18:12:32	<tomsmeding>	sure
2020-09-17 18:12:41	×	z0 quits (~z0@bl15-162-186.dsl.telepac.pt) (Quit: leaving)
2020-09-17 18:12:43	<tomsmeding>	was just trying to satisfy dumptruckman's wishes :p
2020-09-17 18:12:53	<tomsmeding>	but I guess yes, perhaps this shouldn't be Show, dumptruckman
2020-09-17 18:12:56	<ski>	well .. they didn't seem to ask for `name' to be related to `show'
2020-09-17 18:13:01	×	Gigabitten quits (~Somn@ip-99-203-18-60.pools.cgn.spcsdns.net) (Client Quit)
2020-09-17 18:13:02	<tomsmeding>	OW
2020-09-17 18:13:10	<tomsmeding>	my bad
2020-09-17 18:13:27	<tomsmeding>	okay fair point gimme 1min
2020-09-17 18:14:47	<dumptruckman>	yeah, i just have a bunch of different data types that all have a name
2020-09-17 18:14:57	<dumptruckman>	and instead of having a separate name function for each i'd like to just use name
2020-09-17 18:15:02	×	gestone quits (~gestone@c-73-97-137-216.hsd1.wa.comcast.net) (Ping timeout: 260 seconds)
2020-09-17 18:15:37	<tomsmeding>	okay so am I allowed to cheat and use GeneralizedNewtypeDeriving
2020-09-17 18:15:45	<monochrom>	Do you really have an algorithm that cannot be written if every type has a separate name function?
2020-09-17 18:15:46	→	Kaiepi joins (~Kaiepi@nwcsnbsc03w-47-55-157-9.dhcp-dynamic.fibreop.nb.bellaliant.net)
2020-09-17 18:15:54	<dumptruckman>	no
2020-09-17 18:16:00	<dumptruckman>	it's more of a clarity thing i guess
2020-09-17 18:16:00	<monochrom>	Then YAGNI
2020-09-17 18:16:16	→	livvy joins (~livvy@gateway/tor-sasl/livvy)
2020-09-17 18:16:28	<monochrom>	"clarity by overloading names" is in perpetual controversy.
2020-09-17 18:16:33	<dumptruckman>	so i don't have to do `worldName world` `playerName player` `sharableName sharable` etc
2020-09-17 18:16:54	<dumptruckman>	it maybe be a case of YAGNI but at the same time, i'd love to learn how to do this
2020-09-17 18:17:08	<tomsmeding>	dumptruckman: implementation of my wacked idea: https://paste.tomsmeding.com/bSSsxieh
2020-09-17 18:17:20	<tomsmeding>	1. the newtype is crucial, 2. you need that HasName in between
2020-09-17 18:17:36	<tomsmeding>	if you don't want the HasName, you need to use generics
2020-09-17 18:17:43	<tomsmeding>	or not do this at all :)
2020-09-17 18:17:46	<geekosaur>	this sounds like the kind of record field problem lens packages are designed for
2020-09-17 18:17:53	→	frdg joins (~user@pool-71-184-143-249.bstnma.fios.verizon.net)
2020-09-17 18:17:58	<ski>	dumptruckman : "i just have a bunch of different data types that all have a name","and instead of having a separate name function for each i'd like to just use name" -- that's what i figured possibly was the case :)
2020-09-17 18:18:24	<dumptruckman>	sorry, i should've just said that but it wasn't immediately obvious to myself
2020-09-17 18:18:24	→	ahmr88 joins (~ahmr88@cpc85006-haye22-2-0-cust131.17-4.cable.virginm.net)
2020-09-17 18:18:37	<dumptruckman>	tomsmeding: only thing is the tuple feels weird
2020-09-17 18:18:40	<monochrom>	For learning purpose, just handwrite every instance. You said it, "learning".
2020-09-17 18:18:48	<tomsmeding>	yeah the tuple is "whatever data you want there"
2020-09-17 18:18:58	<tomsmeding>	had to put something :)
2020-09-17 18:18:59	<ski>	dumptruckman : so .. since it seems you don't want to write overloaded polymorphic operations wrt this type class, i'd suggest that it's probably better to not make such a type class
2020-09-17 18:19:06	<tomsmeding>	RT monochrom
2020-09-17 18:19:13	<monochrom>	For thinking purpose, http://www.cs.utoronto.ca/~trebla/CSCC24-2020-Summer/05-haskell-types-2.html#why
2020-09-17 18:19:19	×	pingiun quits (~pingiun@ip-213-124-184-182.ip.prioritytelecom.net) (Quit: Textual IRC Client: www.textualapp.com)
2020-09-17 18:19:28	<dumptruckman>	i'd still like to be able to use the format like `Player { playerName :: String }`
2020-09-17 18:19:49	<dumptruckman>	i tried to do it where each type was like `Player { name :: String }` but it yelled at me
2020-09-17 18:20:24	<dumptruckman>	monochrom: haha nice
2020-09-17 18:21:32	<ski>	monochrom : the law thing is a good point to bring up. perhaps you'd want to add the "have overloaded polymorphic operations defined in terms of methods" point to that, as well ?
2020-09-17 18:21:53	<monochrom>	That's "can be used to build useful general algorithms"
2020-09-17 18:21:58	<ski>	tomsmeding : "RT" ?
2020-09-17 18:22:01	<tomsmeding>	retweet
2020-09-17 18:22:08	<ski>	mhm
2020-09-17 18:22:12	<dolio>	I don't see what's wrong with using type classes to overload something like this.

←Prev Next→

Page 1 .. 29 30 31 32 33 34 35 36 37 38 39 .. 5022

All times are in UTC.