Home freenode/#haskell: Logs Calendar

Logs: freenode/#haskell

←Prev  Next→
Page 1 .. 673 674 675 676 677 678 679 680 681 682 683 .. 5022
— 502,152 events total
2020-10-16 14:12:49 × karanlikmadde quits (~karanlikm@2a01:c23:6037:1800:e990:a27c:f553:f1d1) (Quit: karanlikmadde)
2020-10-16 14:14:33 <hyperisco> > flip map [1,2,3] \x -> x + 1
2020-10-16 14:14:35 <lambdabot> error:
2020-10-16 14:14:35 <lambdabot> Unexpected lambda expression in function application:
2020-10-16 14:14:35 <lambdabot> \ x -> x + 1
2020-10-16 14:14:44 <hyperisco> is that extension in GHC yet and what is it called?
2020-10-16 14:16:10 × bitmapper quits (uid464869@gateway/web/irccloud.com/x-dkvrjjnwvljgfxqg) (Quit: Connection closed for inactivity)
2020-10-16 14:16:26 → kuribas joins (~user@ptr-25vy0i94ezsnlaecbal.18120a2.ip6.access.telenet.be)
2020-10-16 14:16:57 <ski> % (`map` [1,2,3]) \x -> x + 1
2020-10-16 14:16:58 <yahb> ski: ; <interactive>:58:17: error:; Unexpected lambda expression in function application:; \ x -> x + 1; You could write it with parentheses; Or perhaps you meant to enable BlockArguments?
2020-10-16 14:17:09 <kuribas> I wonder what the added value of recursion schemes is. They are fun to learn, but I feel they ultimately don't solve a real problem.
2020-10-16 14:17:15 <ski> % :set -XBlockArguments
2020-10-16 14:17:15 <yahb> ski:
2020-10-16 14:17:17 <ski> % (`map` [1,2,3]) \x -> x + 1
2020-10-16 14:17:18 <yahb> ski: [2,3,4]
2020-10-16 14:17:26 <dolio> kuribas: Yep.
2020-10-16 14:17:27 <kuribas> you get slightly less verbose code...
2020-10-16 14:18:01 <hyperisco> am I looking in the wrong place? not documented https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/glasgow_exts.html
2020-10-16 14:18:01 <dolio> The value is that endless papers can be written about them.
2020-10-16 14:18:28 <ski> <https://downloads.haskell.org/~ghc/latest/docs/html/users_guide/glasgow_exts.html#extension-BlockArguments> ?
2020-10-16 14:18:33 → plast1k_ joins (~plast1k_@196.207.181.246)
2020-10-16 14:18:55 <kuribas> dolio: indeed
2020-10-16 14:18:57 <hyperisco> apparently I kept typing "black" instead of "block"
2020-10-16 14:20:19 <kuribas> maybe they could better prove termination in a total language?
2020-10-16 14:20:38 <dolio> I've never seen anyone use them in a total language.
2020-10-16 14:20:58 <dolio> Also, some of them are not total.
2020-10-16 14:21:33 <kuribas> anamorphism...
2020-10-16 14:21:52 <merijn> I tried learning recursion schemes once
2020-10-16 14:21:52 × unlink2 quits (~unlink2@p200300ebcf17c500f1cd4c5efbdddd1e.dip0.t-ipconnect.de) (Ping timeout: 260 seconds)
2020-10-16 14:21:53 <phadej> anamorphims is corecursion, not recursion
2020-10-16 14:21:57 <merijn> Decided it wasn't worth it
2020-10-16 14:22:06 <hyperisco> we've all used foldr … the problem is that we have not learned enough as a community
2020-10-16 14:22:07 <merijn> So hard to see what's going
2020-10-16 14:22:14 <dolio> hylo is non-total.
2020-10-16 14:22:30 <merijn> "oh, but by not reimplementing the recursion each time you avoid screwing it up"
2020-10-16 14:22:42 <merijn> Which, in the grand scheme of things seems like a non-issue
2020-10-16 14:22:58 <hyperisco> I'd say it is the wrong perspective
2020-10-16 14:23:00 <merijn> I can't recall the last time I messed up implementing some recursive function
2020-10-16 14:23:11 <kuribas> merijn: I find I can still screw up with cata :)
2020-10-16 14:23:19 <hyperisco> the point should be removing the need to think about recursion, that is what it removes
2020-10-16 14:23:23 <merijn> I spend more time trying to figure out recursion schemes then I ever spend debugging recursion
2020-10-16 14:23:42 <merijn> hyperisco: But thinking about recursion is an order of magnitude easier than thinking about recursion schemes
2020-10-16 14:23:51 → GyroW_ joins (~GyroW@ptr-48ujrfd1ztq5fjywfw3.18120a2.ip6.access.telenet.be)
2020-10-16 14:23:51 × GyroW_ quits (~GyroW@ptr-48ujrfd1ztq5fjywfw3.18120a2.ip6.access.telenet.be) (Changing host)
2020-10-16 14:23:51 → GyroW_ joins (~GyroW@unaffiliated/gyrow)
2020-10-16 14:23:56 <hyperisco> maybe, or maybe it is just an education problem
2020-10-16 14:24:12 → ephemera_ joins (~E@122.34.1.187)
2020-10-16 14:24:25 <hyperisco> and probably, before that, a lack of many people having experience
2020-10-16 14:24:32 × GyroW quits (~GyroW@unaffiliated/gyrow) (Ping timeout: 256 seconds)
2020-10-16 14:24:37 <phadej> merijn: learning a bit about recursion schemes might make you think about ordinary recursive functions in more principled way
2020-10-16 14:24:43 <merijn> Seems like a big coincidence that the 10+ blogposts by different authors, 3 papers and the docs of recursion schemes are all inscrutable, even for the ones I know
2020-10-16 14:24:53 <phadej> merijn: not much different than learning a bit of Haskell might make you better Python programmer
2020-10-16 14:24:56 <merijn> phadej: Sure, but so far the ROI seems slim to non-existent
2020-10-16 14:25:09 <phadej> merijn: maybe
2020-10-16 14:25:48 <hyperisco> several arguments can be brought out that are hard to distinguish from, say, an OO programmer bemoaning the needless obscurity of FP
2020-10-16 14:26:35 <typetetris> Someone here using ghc8102 for his production software already?
2020-10-16 14:26:40 <merijn> I'll reconsider/reevaluate my opinion if/when someone shows me a tangible example that's improve
2020-10-16 14:26:47 <merijn> typetetris: Define production
2020-10-16 14:26:58 <hyperisco> I'd say all the same things about lens :P
2020-10-16 14:27:04 <merijn> hyperisco: I don't use lens either
2020-10-16 14:27:08 <typetetris> merijn: run software compiled with ghc8.10.2 to earn money.
2020-10-16 14:27:26 <dolio> Lens is way more useful than fancy recursion schemes.
2020-10-16 14:27:39 <merijn> dolio: Oh, I'm sold on the usefulness of lens
2020-10-16 14:27:48 <merijn> dolio: I just haven't done anything where it was worth bothering
2020-10-16 14:28:21 <merijn> I'm all about Luddite haskell :p
2020-10-16 14:28:33 <merijn> (he says, after enabling TypeFamilies and GADTs everywhere...)
2020-10-16 14:28:50 <kuribas> lens is great, but most stuff in lens is unnecessary IMHO
2020-10-16 14:29:02 → DavidEichmann joins (~david@43.240.198.146.dyn.plus.net)
2020-10-16 14:29:14 <phadej> merijn: UTerm in unification-fd is a very neat example of using recursion schemes ideas
2020-10-16 14:29:18 <hyperisco> I am finding that some things just serve an implementation, and that isn't really where the challenge is
2020-10-16 14:29:27 hackage lucid-cdn 0.2.2.0 - Curated list of CDN imports for lucid. https://hackage.haskell.org/package/lucid-cdn-0.2.2.0 (locallycompact)
2020-10-16 14:29:37 <hyperisco> I feel like lens is that sort of thing… but I am pretty stupid about lens.
2020-10-16 14:29:37 <phadej> UTerm though being the same as Free
2020-10-16 14:29:56 <merijn> I'm still undecided about free too :p
2020-10-16 14:30:12 <phadej> it's used very differently though
2020-10-16 14:30:15 <phadej> UTerm is very data-like
2020-10-16 14:30:17 <merijn> I was trying to do a fancy AST with Free, ended with tons of boilerplate
2020-10-16 14:30:22 <phadej> Free is hardly ever presented as it
2020-10-16 14:30:46 <merijn> Talked to Ed about it and he told me he was just using a regular AST type instead of free for that reason
2020-10-16 14:30:48 <phadej> boilerplate is worth if you can have something "for free" (like unification)
2020-10-16 14:30:55 × djellemah quits (~djellemah@2601:5c2:100:96c:e008:b638:39fe:6a54) (Ping timeout: 240 seconds)
2020-10-16 14:31:02 <phadej> for abitrary `f`
2020-10-16 14:31:19 <phadej> but if you don't need (or cannot have) that kind of separation
2020-10-16 14:31:23 <merijn> phadej: Sure, but I feel some of these solutions get over-hyped and then drag beginners into abstraction tarpits :)
2020-10-16 14:31:45 <phadej> merijn: yes. I do think that all extensible-effects things are overhyped
2020-10-16 14:31:59 <hyperisco> I think incorporating a generalisation and only using it in special cases is often increasing burden without benefit
2020-10-16 14:32:01 <phadej> most people don't have the problems they solve
2020-10-16 14:32:23 × oxide quits (~lambda@unaffiliated/mclaren) (Ping timeout: 260 seconds)
2020-10-16 14:33:07 <lortabac> what problems do they solve?
2020-10-16 14:33:12 × ph88 quits (~ph88@2a02:8109:9e40:2704:21a2:9414:d2e6:266d) (Quit: Leaving)
2020-10-16 14:33:27 <merijn> lortabac: Staying employed as PL researcher ;)
2020-10-16 14:33:30 merijn ducks
2020-10-16 14:33:31 <phadej> "separation of concerns"
2020-10-16 14:33:41 <hyperisco> getting Monad machinery from just a Functor is helpful when your total effort needs not be much more than obtaining Monad machinery
2020-10-16 14:34:17 <phadej> which in that case is decoupling syntax from semantics
2020-10-16 14:34:19 <hyperisco> if you then have to go on defining several more things, the benefit of Free becomes comparatively small and the burden multiplies over your remaining effort
2020-10-16 14:34:39 <phadej> (specifically having different semantics for the same syntax)
2020-10-16 14:34:49 <phadej> if these words don't mean anything to you, you don't have that problem ;)
2020-10-16 14:35:04 <lortabac> I think I don't have that problem :)
2020-10-16 14:35:06 <hyperisco> in other words, Free life seems to start with your Functor, then you start adding on more definitions, then you realise what Free is giving you isn't worth it
←Prev  Next→
Page 1 .. 673 674 675 676 677 678 679 680 681 682 683 .. 5022

All times are in UTC.