Home freenode/#haskell: Logs Calendar

Logs: freenode/#haskell

←Prev  Next→
Page 1 .. 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 .. 5022
— 502,152 events total
2020-11-09 15:31:06 × heatsink quits (~heatsink@107-136-5-69.lightspeed.sntcca.sbcglobal.net) (Ping timeout: 256 seconds)
2020-11-09 15:33:05 <bqv> man lenses make my head hurt
2020-11-09 15:33:41 <bqv> why are people so twisted on monads when lenses are a thing
2020-11-09 15:34:52 <int-e> Because you can use Haskell without touching lenses, but you're kind of forced to use at least one monad on occasion?
2020-11-09 15:35:13 <bqv> yeah
2020-11-09 15:35:21 <int-e> Monads also have a longer history of being scary.
2020-11-09 15:35:22 <dminuoso> bqv: Also, if you find lenses a bit too hard to use, you can also try optics
2020-11-09 15:35:30 <dminuoso> It offers better readable types and much improved diagnostics
2020-11-09 15:35:49 <maerwald> and good documentation
2020-11-09 15:35:54 <bqv> any functional difference?
2020-11-09 15:35:56 <dminuoso> Part of what makes lens so harsh is that it leaks implementation details in types
2020-11-09 15:36:01 <maerwald> bqv: yes, some
2020-11-09 15:36:13 <bqv> hm
2020-11-09 15:36:13 <dminuoso> bqv: It's a bit safer and doesn't allow some things, also it's missing a few parts left and right
2020-11-09 15:36:15 → adm_ joins (~adm@43.229.88.236)
2020-11-09 15:36:18 <dminuoso> Overall it's mostly a drop in replacement
2020-11-09 15:36:25 × toorevitimirp quits (~tooreviti@117.182.180.38) (Remote host closed the connection)
2020-11-09 15:36:27 <dminuoso> (modulo replacing . with %, and unsafe combinators)
2020-11-09 15:36:48 <bqv> fair enough
2020-11-09 15:36:48 <dminuoso> for example, optics doesnt allow you to `view` through a fold
2020-11-09 15:37:06 <int-e> . o O ( please describe lens in a sentence )
2020-11-09 15:37:06 × ystael quits (~ystael@209.6.50.55) (Quit: Lost terminal)
2020-11-09 15:37:56 <bqv> something something coalgebras costate comonad...
2020-11-09 15:38:00 <dminuoso> "First class selectors for dealing with nested data, like css-selector/xpath"
2020-11-09 15:38:56 <gehmehgeh> dminuoso: hmm, so how would implement something as simply as an array 10x10 array? That is 100 elemnts, selectable by coordinates (x,y) such that x and y are in {0,9} ?
2020-11-09 15:39:00 <gehmehgeh> *simple
2020-11-09 15:39:13 × samlamamma quits (~user@h188-122-129-70.cust.a3fiber.se) (Ping timeout: 264 seconds)
2020-11-09 15:39:15 <dminuoso> gehmehgeh: By writing your own index functions
2020-11-09 15:39:17 <gehmehgeh> still Vector of VEctor?
2020-11-09 15:39:20 <dminuoso> Yes.
2020-11-09 15:39:20 <gehmehgeh> hm
2020-11-09 15:39:23 <gehmehgeh> ok
2020-11-09 15:39:25 <gehmehgeh> will do. Thanks
2020-11-09 15:39:27 → ystael joins (~ystael@209.6.50.55)
2020-11-09 15:39:30 <dminuoso> For just a matrix, you could use any of the matrix libraires instead too
2020-11-09 15:39:50 <dminuoso> like hmatrix if you want a BLAS/LAPACK backed matrix
2020-11-09 15:40:04 <gehmehgeh> hmmm
2020-11-09 15:40:19 × pjb quits (~t@2a01cb04063ec50070ecdd787f2214a5.ipv6.abo.wanadoo.fr) (Ping timeout: 272 seconds)
2020-11-09 15:40:25 <dminuoso> For higher dimensional tensors, it depends. Do you need this with dynamic/arbitrary dimensions?
2020-11-09 15:40:28 <dminuoso> Or just fixed 3d?
2020-11-09 15:41:29 <gehmehgeh> At the moment, I'm just experimenting
2020-11-09 15:42:06 <gehmehgeh> I was thinking about a fixed 10x10 matrix for now, though
2020-11-09 15:42:33 <dminuoso> bqv: "coalgebras of the costate comonad" is not just a joke, it's also a sad truth of lens. There's of course folks like edwardk who like being directly exposed to the internals of optics, others like me are not fond of having to decode failed constraints into "what optic did I build, what optic is expected" left and right.
2020-11-09 15:43:27 <bqv> i thought i saw a ghc plugin somewhere that improved the errors a bit
2020-11-09 15:43:42 <bqv> maybe i just saw the optics library you mentioned
2020-11-09 15:43:42 <dminuoso> That's the thing, you can do this in all (ghc-extended) haskell without plugins
2020-11-09 15:43:52 <bqv> hm
2020-11-09 15:43:53 <dminuoso> optics uses just the standard ghc facilities
2020-11-09 15:44:20 <dminuoso> • A_Fold cannot be used as A_Getter
2020-11-09 15:44:52 → da39a3ee5e6b4b0d joins (~textual@cm-171-98-71-231.revip7.asianet.co.th)
2020-11-09 15:45:03 <dminuoso> * No instance for (Monoid Int) arising from a use of `folded'
2020-11-09 15:45:20 × da39a3ee5e6b4b0d quits (~textual@cm-171-98-71-231.revip7.asianet.co.th) (Client Quit)
2020-11-09 15:45:46 <dminuoso> I'll take the former over the latter
2020-11-09 15:46:26 <dminuoso> Also, with optics you can just say `f % g :: _` and let GHC infer the type for you. And it wont give you some constraint algebra you have to do in your head, but straight up tell you
2020-11-09 15:46:26 × Tario quits (~Tario@198.252.153.28) (Read error: Connection reset by peer)
2020-11-09 15:46:31 <dminuoso> "This is an AffineTraversal"
2020-11-09 15:46:46 <dminuoso> (Okay bad example, since lens cant represent affine traversals)
2020-11-09 15:47:01 <bqv> to be honest, i am rather interested in the implementation details anyway, so i'm not against the arcane type signatures
2020-11-09 15:47:15 <bqv> but it does obviously make debugging basically black magic until i'm a master of profunctors
2020-11-09 15:48:06 <dminuoso> btw, lens already does a mixed van laarhoven/profunctor implementation
2020-11-09 15:48:25 <dminuoso> the implementation of optics is not more complicated than that of lens, really
2020-11-09 15:48:48 <dminuoso> (I guess much of the weirdnesses of lens are about generalization and performance)
2020-11-09 15:48:55 → ddellacosta joins (dd@gateway/vpn/mullvad/ddellacosta)
2020-11-09 15:49:02 <dminuoso> and very opaque naming
2020-11-09 15:49:48 <dminuoso> Dont get me wrong, I dont have any fundamental problems with lens, but I like not having to decipher type errors. :)
2020-11-09 15:49:53 <dminuoso> And I like type inference being on my side
2020-11-09 15:49:57 × bliminse quits (~bliminse@host109-156-197-211.range109-156.btcentralplus.com) (Ping timeout: 256 seconds)
2020-11-09 15:50:41 × cosimone quits (~cosimone@2001:b07:ae5:db26:d849:743b:370b:b3cd) (Remote host closed the connection)
2020-11-09 15:50:43 <bqv> isn't there a lot of stuff that's targeted at compatibility with lens specifically - would those translate?
2020-11-09 15:50:46 → bliminse joins (~bliminse@host109-156-197-211.range109-156.btcentralplus.com)
2020-11-09 15:51:05 → cosimone joins (~cosimone@2001:b07:ae5:db26:d849:743b:370b:b3cd)
2020-11-09 15:51:25 <dminuoso> Well you can translate lenses from VL to optics forth and back
2020-11-09 15:51:45 <dminuoso> at least some
2020-11-09 15:51:59 bqv . o O ( lens isos? )
2020-11-09 15:52:11 <dminuoso> Or.. well one direction anyway
2020-11-09 15:52:13 <dminuoso> foldVL :: (forall f. Applicative f => (a -> f u) -> s -> f v) -> Fold s a
2020-11-09 15:52:15 <dminuoso> e.g.
2020-11-09 15:52:17 ← iqubic` parts (~user@2601:602:9500:4870:f97f:76da:55:8502) ("ERC (IRC client for Emacs 28.0.50)")
2020-11-09 15:52:30 <dminuoso> traversalVL :: TraversalVL s t a b -> Traversal s t a b
2020-11-09 15:53:07 → alp joins (~alp@88.126.45.36)
2020-11-09 15:53:18 <dminuoso> It would be kind of nice if such packages were written with backpack to have switchable optics implementation
2020-11-09 15:53:38 <bqv> yeah..
2020-11-09 15:53:49 <gtk> what is the file extension for Happy grammer file?
2020-11-09 15:54:49 <dminuoso> bqv: But honestly, if you write a library, I'd be careful not to expose lenses as part of the main package
2020-11-09 15:55:09 <dminuoso> If you wanted to do this at all, try and make a `lib-lenses` package instead, that way someone else can make `lib-optics` too if they desired
2020-11-09 15:55:24 <dminuoso> And users who dont want to drag in the transitive dependencies can use it without lens/optics
2020-11-09 15:55:31 <bqv> oh, huh. ok
2020-11-09 15:56:40 → petersen joins (~petersen@redhat/juhp)
2020-11-09 15:57:19 <idnar> dminuoso: one nice thing about `lens` is that you don't need to depend on it to export lenses
2020-11-09 15:57:56 <dminuoso> idnar: otoh optics-core is very light on dependencies
2020-11-09 15:57:57 × mputz quits (~Thunderbi@dslb-084-058-211-084.084.058.pools.vodafone-ip.de) (Ping timeout: 260 seconds)
2020-11-09 15:58:16 <dminuoso> (microlens otoh can be too small)
2020-11-09 16:01:00 → Tario joins (~Tario@198.252.153.28)
2020-11-09 16:02:44 → pjb joins (~t@2a01cb04063ec50054866239aff0100c.ipv6.abo.wanadoo.fr)
2020-11-09 16:02:48 × Tario quits (~Tario@198.252.153.28) (Read error: Connection reset by peer)
2020-11-09 16:04:13 → tsaka__ joins (~torstein@ppp-2-84-29-23.home.otenet.gr)
2020-11-09 16:04:15 × cosimone quits (~cosimone@2001:b07:ae5:db26:d849:743b:370b:b3cd) (Quit: cosimone)
2020-11-09 16:04:44 × acidjnk_new2 quits (~acidjnk@p200300d0c718f646f97e112dfc4bc61a.dip0.t-ipconnect.de) (Ping timeout: 240 seconds)
2020-11-09 16:05:24 × thir quits (~thir@p200300f27f0b7e00f4e9381c2bf90854.dip0.t-ipconnect.de) (Ping timeout: 240 seconds)
2020-11-09 16:06:51 × adm_ quits (~adm@43.229.88.236) (Remote host closed the connection)
←Prev  Next→
Page 1 .. 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 .. 5022

All times are in UTC.