Home freenode/#haskell: Logs Calendar

Logs: freenode/#haskell

←Prev  Next→
Page 1 .. 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 .. 5022
— 502,152 events total
2020-11-19 11:39:19 <dminuoso> It's how the function we replace the cons with associates
2020-11-19 11:39:39 <boxscape> dminuoso I know, but for that you need to have some mental image of what left and right means in a list, right?
2020-11-19 11:39:58 <boxscape> which isn't fundamentally part of the type
2020-11-19 11:40:16 <kuribas> (a <> (b <> (c <> d))) should be the same as (((a <> b) <> c) <> d)
2020-11-19 11:40:25 <kuribas> but one may be more efficient to calculate
2020-11-19 11:40:34 <boxscape> so if you swap directions of that mental image, you rename foldr to foldl and foldl to foldr, and perhaps write the constructor as [a] : a, and I imagine that's what a snoc list would be
2020-11-19 11:40:38 <boxscape> functionally the same as a list
2020-11-19 11:40:55 <dminuoso> Ill await nshepperd's response. :)
2020-11-19 11:40:59 <dminuoso> To clarify
2020-11-19 11:40:59 <boxscape> that's fair
2020-11-19 11:41:00 <kuribas> > foldMap' (<>) [a, b, c, d]
2020-11-19 11:41:02 <lambdabot> <Expr -> Expr>
2020-11-19 11:41:24 <kuribas> > foldMap' Sum [a, b, c, d]
2020-11-19 11:41:26 <lambdabot> Sum {getSum = 0 + a + b + c + d}
2020-11-19 11:41:30 <kuribas> > foldMap Sum [a, b, c, d]
2020-11-19 11:41:32 <lambdabot> Sum {getSum = a + (b + (c + (d + 0)))}
2020-11-19 11:41:51 <dminuoso> kuribas: The difference matters in two respects: Is the input data structure infinite, and do we need to be lazy on the output (do we build a chain of data constructors)
2020-11-19 11:42:18 <kuribas> those aren't the same value
2020-11-19 11:42:24 <kuribas> are they?
2020-11-19 11:42:24 × chaosmasttter quits (~chaosmast@p200300c4a70b2a01441f1455f36b3658.dip0.t-ipconnect.de) (Ping timeout: 240 seconds)
2020-11-19 11:42:59 <kuribas> ah they are since 0 is the empty element
2020-11-19 11:43:16 × invaser quits (~Thunderbi@31.148.23.125) (Ping timeout: 240 seconds)
2020-11-19 11:43:20 <dminuoso> kuribas: by monoid laws they are the same
2020-11-19 11:43:41 <dminuoso> because precisely, a monoid is associative (so it doesnt matter how they associate) and mempty is the identity element (it doesnt matter where you slap it onto)
2020-11-19 11:44:15 → ft joins (~ft@shell.chaostreff-dortmund.de)
2020-11-19 11:44:47 <dminuoso> Guess that's the sort of things that makes foldMap so cool. You can just replace it with foldMap' without worrying whether it changes its meaning
2020-11-19 11:44:55 <dminuoso> with foldr/foldl
2020-11-19 11:45:07 <dminuoso> you have to manually check for associativity and identity
2020-11-19 11:46:55 × zyklotomic quits (~ethan@unaffiliated/chocopuff) (Quit: WeeChat 2.9)
2020-11-19 11:46:58 → Alleria_ joins (~AllahuAkb@2604:2000:1484:26:ef:21ba:eb1b:b066)
2020-11-19 11:47:05 <boxscape> hmm I wonder if a monoid with separate left and right identity elements could be useful
2020-11-19 11:47:31 <tomjaguarpaw> If you had left and right identities el and er what would el <> er be?
2020-11-19 11:48:26 × Stanley00 quits (~stanley00@unaffiliated/stanley00) ()
2020-11-19 11:48:53 <boxscape> that's a very good question
2020-11-19 11:49:17 <boxscape> I guess it wouldn't be useful then
2020-11-19 11:49:36 × cfricke quits (~cfricke@unaffiliated/cfricke) (Quit: WeeChat 2.9)
2020-11-19 11:49:53 × Amras quits (~Amras@unaffiliated/amras0000) (Ping timeout: 272 seconds)
2020-11-19 11:50:32 <kuribas> dminuoso: it does change meaning wrt bottom
2020-11-19 11:50:35 → jeet1 joins (~jeet@183.83.41.219)
2020-11-19 11:50:48 <boxscape> Although, I guess you could have a (not-quite-)monoid that *only* has left identity (or only right-identity)
2020-11-19 11:51:22 <kuribas> > foldMap First [1, 2, undefined]
2020-11-19 11:51:24 <lambdabot> error:
2020-11-19 11:51:24 <lambdabot> • No instance for (Num (Maybe ())) arising from a use of ‘e_112’
2020-11-19 11:51:24 <lambdabot> • In the expression: e_112
2020-11-19 11:51:50 <nshepperd> yeah a snoc list is just a cons list that has had left and right renamed
2020-11-19 11:52:41 × cosimone quits (~cosimone@2001:b07:ae5:db26:d849:743b:370b:b3cd) (Remote host closed the connection)
2020-11-19 11:53:02 → cosimone joins (~cosimone@2001:b07:ae5:db26:d849:743b:370b:b3cd)
2020-11-19 11:53:04 <kuribas> > foldMap First [Just 1, Just 2, undefined]
2020-11-19 11:53:06 <lambdabot> First {getFirst = Just 1}
2020-11-19 11:53:12 <kuribas> > foldMap' First [Just 1, Just 2, undefined]
2020-11-19 11:53:14 <lambdabot> First {getFirst = Just 1}
2020-11-19 11:53:47 × noctux quits (~noctux@unaffiliated/noctux) (Read error: Connection reset by peer)
2020-11-19 11:53:48 <nshepperd> you won't often find them in the wild
2020-11-19 11:54:25 → noctux joins (~noctux@unaffiliated/noctux)
2020-11-19 11:54:36 <kuribas> > foldMap All [True, False, undefined]
2020-11-19 11:54:37 <lambdabot> All {getAll = False}
2020-11-19 11:54:41 <kuribas> > foldMap' All [True, False, undefined]
2020-11-19 11:54:42 <lambdabot> All {getAll = False}
2020-11-19 11:54:49 <kuribas> hmmm
2020-11-19 11:55:10 → m0rphism joins (~m0rphism@HSI-KBW-095-208-098-207.hsi5.kabel-badenwuerttemberg.de)
2020-11-19 11:55:36 <dminuoso> nshepperd: That sounds.. like a list.
2020-11-19 11:56:04 <nshepperd> well yeah
2020-11-19 11:56:38 <nshepperd> you can just do things with lists, and remember to reverse things in your head when reading code
2020-11-19 11:57:27 × agrajag` quits (~agrajag`@84.39.117.57) (Remote host closed the connection)
2020-11-19 11:57:31 <nshepperd> but sometimes it's easier to have a different type so that the syntax corresponds to the intended ordering
2020-11-19 11:57:56 <dminuoso> So roughly a `data Tsil a = Snoc a (Tsil a) | Lin` differs from [] by conceptually reversing before folding/traversing?
2020-11-19 11:58:15 <dminuoso> (or in case of traverse, revere, traverse, and then reverse again)
2020-11-19 11:58:15 <kuribas> they are the same even with bottom then?
2020-11-19 11:59:14 <boxscape> the list matches the intended left right order better with Snoc (Tsil a) a: Lin `SNoc` 3 `Snoc` 2 `Snoc` 1
2020-11-19 11:59:17 <nshepperd> generally you write SnocList a = Nil | Snoc (SnocList a) a
2020-11-19 11:59:29 → chaosmasttter joins (~chaosmast@p200300c4a70b2a01441f1455f36b3658.dip0.t-ipconnect.de)
2020-11-19 11:59:48 <dminuoso> Why not just use Dual?
2020-11-19 12:00:02 × jeet1 quits (~jeet@183.83.41.219) (Quit: WeeChat 2.9)
2020-11-19 12:00:11 <nshepperd> performance
2020-11-19 12:00:23 <dminuoso> Mmm, fair
2020-11-19 12:00:34 <nshepperd> also, pattern matching
2020-11-19 12:00:44 × Lycurgus quits (~niemand@cpe-45-46-142-188.buffalo.res.rr.com) (Quit: Exeunt)
2020-11-19 12:00:45 → cfricke joins (~cfricke@unaffiliated/cfricke)
2020-11-19 12:00:46 × pfurla quits (~pfurla@190.15.195.173.client.static.strong-in52.as13926.net) (Ping timeout: 272 seconds)
2020-11-19 12:00:56 <dminuoso> boxscape: Ill take a look at your gist later tonight, think I need to sit down and employ the good ol' fashioned Feynman algorithm
2020-11-19 12:00:57 → __monty__ joins (~toonn@unaffiliated/toonn)
2020-11-19 12:01:18 <dminuoso> nshepperd: well, the pattern matching is a mood argument, since it's no different from (:)
2020-11-19 12:01:37 → pfurla joins (~pfurla@ool-182ed2e2.dyn.optonline.net)
2020-11-19 12:01:44 <nshepperd> nooo
2020-11-19 12:01:58 <dminuoso> Whether you write `1:2:3:[]` or `[]:%3:%2:%1` seems no different at all
2020-11-19 12:02:07 <nshepperd> i mean, you can't use Dual when you're doing type level lists
2020-11-19 12:02:07 × AWizzArd quits (~code@unaffiliated/awizzard) (Read error: Connection reset by peer)
2020-11-19 12:02:10 <dminuoso> ah
2020-11-19 12:02:41 → AWizzArd joins (~code@gehrels.uberspace.de)
2020-11-19 12:03:42 <boxscape> dminuoso In a way I think these proofs are even harder to understand in Haskell than in some other languages, because the way constraint resolution makes it so it's not obvious how things are actually plugged together (though at least these are small proofs). But replacing the rhss with typed holes should help.
2020-11-19 12:03:44 × xff0x quits (~fox@2001:1a81:534b:a000:20f1:4040:a0e2:f16) (Ping timeout: 240 seconds)
2020-11-19 12:03:58 <boxscape> s/the way//
2020-11-19 12:04:32 <kuribas> I find myself going back from typed holes to tuples
2020-11-19 12:04:45 → xff0x joins (~fox@2001:1a81:534b:a000:65df:cebc:ef3d:70f7)
2020-11-19 12:04:48 <kuribas> the error message is much better
2020-11-19 12:05:06 <__monty__> Do you mean unit?
2020-11-19 12:05:13 <kuribas> __monty__: yes
2020-11-19 12:05:36 <kuribas> my-fun = () $ partial-implementation
2020-11-19 12:06:11 <__monty__> That does change how associativity is parsed, no?
2020-11-19 12:06:42 <boxscape> oh huh
←Prev  Next→
Page 1 .. 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 .. 5022

All times are in UTC.