Logs: liberachat/#haskell
| 2021-07-27 19:45:11 | <Clint> | transliterating them to hanzi? |
| 2021-07-27 19:45:18 | <monochrom> | Yes. |
| 2021-07-27 19:45:24 | Clint | nods. |
| 2021-07-27 19:45:45 | <Cale> | ＦｉｒｓｔＮａｍｅ・ＬａｓｔＮａｍｅ |
| 2021-07-27 19:45:53 | <lechner> | doesn't the last name come first? |
| 2021-07-27 19:46:05 | <monochrom> | No. |
| 2021-07-27 19:46:23 | <Cale> | I guess that's why the dot |
| 2021-07-27 19:46:42 | <lechner> | sorry, i got confused with my chinese name. yours is for transliteration |
| 2021-07-27 19:46:46 | <Cale> | Well, apart from the fact that it would be awkward to figure out where one name ends and the other begins |
| 2021-07-27 19:47:09 | <lechner> | it does not matter |
| 2021-07-27 19:47:24 | <lechner> | for one person |
| 2021-07-27 19:48:44 | × | fossdd quits (~fossdd@sourcehut/user/fossdd) (Ping timeout: 250 seconds) |
| 2021-07-27 19:48:51 | → | eggplantade joins (~Eggplanta@108-201-191-115.lightspeed.sntcca.sbcglobal.net) |
| 2021-07-27 19:48:53 | <lechner> | is using too many $ must be a beginner's mistake |
| 2021-07-27 19:49:14 | → | fossdd joins (~fossdd@sourcehut/user/fossdd) |
| 2021-07-27 19:49:44 | × | Obo quits (~roberto@70.pool90-171-81.dynamic.orange.es) (Ping timeout: 252 seconds) |
| 2021-07-27 19:50:01 | <geekosaur> | it's often recommended not to use $ at all until you get used to haskell |
| 2021-07-27 19:50:42 | <lechner> | too many brackets. i wear glasses |
| 2021-07-27 19:51:55 | <lechner> | but i am not using a single . ! i was so excited about them. also, the pure parts of my programs are tiny. monad in. monad out. |
| 2021-07-27 19:52:19 | <yushyin> | the general opinion here seems to be that $ is overused and that it is better to use parentheses more often |
| 2021-07-27 19:52:52 | <Cale> | I dunno, there are places where you definitely want to use $ -- particularly if the argument is long enough that it's annoying to find the matching closing parens |
| 2021-07-27 19:52:53 | <monochrom> | The correct way to get rid of parenthesis is to defeat the point of Fortran and go back to 3-address instructions: let tmp1 = f x; tmp2 = g tmp1; tmp3 = h tmp2 in k tmp3 |
| 2021-07-27 19:53:24 | <Cale> | But I'll tend not to chain $'s myself, unless I'm trying to blend in with an existing code style |
| 2021-07-27 19:53:40 | <Cale> | I'll always use composition and then a single $ when I use it |
| 2021-07-27 19:53:57 | <monochrom> | and roll back 60 years of progress in languages and compilers. |
| 2021-07-27 19:54:28 | <Cale> | monochrom: Or RPN |
| 2021-07-27 19:54:44 | → | alx741 joins (~alx741@181.196.69.4) |
| 2021-07-27 19:54:53 | <DigitalKiwi> | don't be silly everyone knows there's no $ in haskell! /s |
| 2021-07-27 19:55:07 | <Cale> | I wish I hadn't lost all my silly dc snippets |
| 2021-07-27 19:55:37 | <Cale> | I remember at one point working out how to do maps and folds over the stack in dc :) |
| 2021-07-27 19:55:44 | <yushyin> | pure $ whatever comes next -- is a common pattern for me |
| 2021-07-27 19:56:12 | <Cale> | I also never use pure unless it's important that my definition is Applicative polymorphic |
| 2021-07-27 19:56:29 | <lechner> | i actually understood that |
| 2021-07-27 19:56:32 | × | fossdd quits (~fossdd@sourcehut/user/fossdd) (Ping timeout: 250 seconds) |
| 2021-07-27 19:56:35 | <lechner> | vs return |
| 2021-07-27 19:56:39 | <monochrom> | Backus and Knuth worked so hard to find algorithms that can parse (a+b)*c+d so that if you see that formula in math books you can just copy that and it's valid code. Now entitled millenials trash that and want to go back to 3-address instructions. |
| 2021-07-27 19:56:43 | <geekosaur> | showing your age :þ |
| 2021-07-27 19:56:43 | <Cale> | When I see "pure" in code, I take it as a sign that Applicative is being used in some nontrivial way, and if it's not, I'm mildly annoyed :) |
| 2021-07-27 19:57:15 | → | fossdd joins (~fossdd@sourcehut/user/fossdd) |
| 2021-07-27 19:58:05 | <geekosaur> | bet you hate "monad of no return" |
| 2021-07-27 19:58:25 | <Cale> | I'm less concerned about that |
| 2021-07-27 19:58:33 | <lechner> | others have told me prefer "pure" |
| 2021-07-27 19:58:38 | <lechner> | to |
| 2021-07-27 19:59:00 | <Cale> | I *would* be opposed to return :: Applicative f => a -> f a though |
| 2021-07-27 20:00:01 | <Cale> | It just helps make typechecking in your head that much easier to have that little cue |
| 2021-07-27 20:00:16 | <monochrom> | I use pure when the context uses liftA2 or >>= etc so they are on the same page. |
| 2021-07-27 20:00:23 | <Cale> | Yeah, that's fair |
| 2021-07-27 20:00:36 | <Cale> | I might use pure in cases where I'm also using (<*>) immediately adjacent to it |
| 2021-07-27 20:00:36 | → | drd joins (~drd@93-39-151-19.ip76.fastwebnet.it) |
| 2021-07-27 20:00:53 | <Rembane> | I always use pure, I blame Purescript. |
| 2021-07-27 20:01:17 | <Cale> | Like, maybe I know the thing is a monad, but I'm still using Applicative syntax to combine a bunch of stuff, in that case, pure seems okay. |
| 2021-07-27 20:01:18 | <DigitalKiwi> | Cale: lol so that tweet you sent earlier from d6 (who i follow on twitter) finally showed up in my feed...only after someone else had quoted tweeted it and d6 retweeted *that*... |
| 2021-07-27 20:01:25 | <Cale> | hah |
| 2021-07-27 20:01:28 | <geekosaur> | quite a few people say "just use pure" these days |
| 2021-07-27 20:01:45 | <yushyin> | i roll a dice before every use. even is pure, odd is return |
| 2021-07-27 20:01:46 | <Cale> | geekosaur: I'm aware, but I think it does lose something |
| 2021-07-27 20:01:55 | → | LukeHoersten joins (~LukeHoers@user/lukehoersten) |
| 2021-07-27 20:02:02 | → | pschorf joins (~user@208.127.190.222) |
| 2021-07-27 20:02:28 | <Cale> | Also, I'm of the mindset that "return" is also a pretty good name for what it is most of the time |
| 2021-07-27 20:02:30 | <maerwald> | I use `pure` |
| 2021-07-27 20:02:38 | → | Deide joins (~Deide@217.155.19.23) |
| 2021-07-27 20:02:38 | × | Deide quits (~Deide@217.155.19.23) (Changing host) |
| 2021-07-27 20:02:38 | → | Deide joins (~Deide@user/deide) |
| 2021-07-27 20:02:50 | × | mr-red quits (~drd@93-39-151-19.ip76.fastwebnet.it) (Ping timeout: 258 seconds) |
| 2021-07-27 20:02:58 | <Cale> | i.e. it makes an action which does nothing except to return the given value as its result |
| 2021-07-27 20:03:01 | <monochrom> | I prefer "impurify" >:) |
| 2021-07-27 20:03:14 | <Cale> | haha |
| 2021-07-27 20:04:08 | <monochrom> | class Monad m => Animate m where anthropomorphize :: a -> m a |
| 2021-07-27 20:04:21 | <Cale> | Of course, people who just came from imperative programming sometimes find "return" a bit jarring, because it doesn't have the control effects that they're used to. |
| 2021-07-27 20:05:07 | × | juhp quits (~juhp@128.106.188.220) (Ping timeout: 245 seconds) |
| 2021-07-27 20:05:26 | → | Obo joins (~roberto@70.pool90-171-81.dynamic.orange.es) |
| 2021-07-27 20:05:28 | <Cale> | Imagine being used to having all your stuff implicitly wrapped in callCC |
| 2021-07-27 20:06:49 | → | juhp joins (~juhp@128.106.188.220) |
| 2021-07-27 20:06:53 | → | chris_ joins (~chris@81.96.113.213) |
| 2021-07-27 20:08:32 | <lechner> | return actually makes monads less scary. would it be fair to somehow compare monads to imperative function prototypes? |
| 2021-07-27 20:09:33 | <jiribenes> | it makes monads _look_ less scary, but its semantics confuse the heck out of imperative programmers |
| 2021-07-27 20:10:18 | <monochrom> | The difference between "I understand, meaning I feel so good" and "I understand, meaning I write correct code". |
| 2021-07-27 20:10:37 | <lechner> | does it have semantics |
| 2021-07-27 20:10:52 | <lechner> | it does nothing |
| 2021-07-27 20:10:57 | <monochrom> | A little bit of semantics from the monad laws. |
| 2021-07-27 20:11:06 | <geekosaur> | but tey expect semantics |
| 2021-07-27 20:11:13 | <monochrom> | Beyond that, it's up to actual instances. |
| 2021-07-27 20:11:20 | → | gehmehgeh joins (~user@user/gehmehgeh) |
| 2021-07-27 20:11:37 | → | P1RATEZ joins (piratez@user/p1ratez) |
| 2021-07-27 20:12:00 | <monochrom> | But do look at "return x >>= k = k x". This tells you a lot. For example, this tells you "not an exit". |
| 2021-07-27 20:12:34 | <jiribenes> | ^ this |
| 2021-07-27 20:13:04 | <monochrom> | I know of a monad tutorial (that even makes it to Haskell Weekly News) that gets this wrong. |
| 2021-07-27 20:13:18 | <monochrom> | Subsequently I shamed them in an exam question. |
| 2021-07-27 20:13:53 | <Cale> | haha, nice |
| 2021-07-27 20:15:07 | <Cale> | You also have (x >>= \v -> return v) = x, which is another way to say that return isn't doing a whole lot |
| 2021-07-27 20:15:38 | <Cale> | But yeah, the actual meaning of return (and (>>=)) varies based on which monad we're talking about |
| 2021-07-27 20:15:55 | → | lavaman joins (~lavaman@98.38.249.169) |
| 2021-07-27 20:16:27 | <Cale> | So while there's a sort of "imperative-ish" feel to things, sometimes it'll subvert your expectations a bit. Monads can have backtracking, state can flow in the opposite direction from the rest of the computation, etc. |
| 2021-07-27 20:17:03 | <maerwald> | is that a good thing? |
| 2021-07-27 20:17:18 | <Cale> | Can be. |
| 2021-07-27 20:17:20 | <maerwald> | I'd argue it basically makes reasoning harder |
| 2021-07-27 20:17:38 | <maerwald> | about monadic combinators at least |
| 2021-07-27 20:17:51 | <Cale> | It's good because we get to define a bunch of functions generically that'll work in any monad, and then use them in every monad we define. |
| 2021-07-27 20:18:00 | <maerwald> | "work" how? |
| 2021-07-27 20:18:11 | <maerwald> | rather "somehow" |
All times are in UTC.