Newest at the top
| 2025-02-12 15:04:52 +0100 | tri | (~tri@ool-44c70bcb.dyn.optonline.net) (Ping timeout: 252 seconds) |
| 2025-02-12 15:00:22 +0100 | tri | (~tri@ool-44c70bcb.dyn.optonline.net) |
| 2025-02-12 14:56:48 +0100 | zungi | (~tory@user/andrewchawk) andrewchawk |
| 2025-02-12 14:55:10 +0100 | alp | (~alp@user/alp) alp |
| 2025-02-12 14:55:10 +0100 | alp | (~alp@5.226.4.112) (Changing host) |
| 2025-02-12 14:55:05 +0100 | alp | (~alp@5.226.4.112) |
| 2025-02-12 14:52:26 +0100 | forell | (~forell@user/forell) forell |
| 2025-02-12 14:51:25 +0100 | forell | (~forell@user/forell) (Quit: ZNC - https://znc.in) |
| 2025-02-12 14:51:12 +0100 | zungi | (~tory@user/andrewchawk) (Ping timeout: 264 seconds) |
| 2025-02-12 14:49:17 +0100 | brachyrhynchos | (~brachyrhy@user/brachyrhynchos) (Remote host closed the connection) |
| 2025-02-12 14:43:15 +0100 | brachyrhynchos | (~brachyrhy@user/brachyrhynchos) brachyrhynchos |
| 2025-02-12 14:38:54 +0100 | alfiee | (~alfiee@user/alfiee) (Ping timeout: 260 seconds) |
| 2025-02-12 14:36:32 +0100 | hgolden | (~hgolden@2603:8000:9d00:3ed1:6ff3:8389:b901:6363) hgolden |
| 2025-02-12 14:35:22 +0100 | <haskellbridge> | <Profpatsch> I mean the decoupling of DecodeAction and actual decoding is super interesting |
| 2025-02-12 14:34:10 +0100 | alfiee | (~alfiee@user/alfiee) alfiee |
| 2025-02-12 14:34:10 +0100 | <bgamari> | but cborg shows that this isn't the only path to a decent parser |
| 2025-02-12 14:32:54 +0100 | <bgamari> | in particular, intermediate structures can be deforested readily |
| 2025-02-12 14:32:36 +0100 | <bgamari> | since it optimises specific parsers reasonably well |
| 2025-02-12 14:31:56 +0100 | <bgamari> | when people think of "high performance" parsing in Haskell they often resort of CPS'd style |
| 2025-02-12 14:31:16 +0100 | <bgamari> | indeed cborg is an interesting design (primarily Duncan's) |
| 2025-02-12 14:31:15 +0100 | tromp | (~textual@92-110-219-57.cable.dynamic.v4.ziggo.nl) |
| 2025-02-12 14:30:46 +0100 | <bgamari> | so evaluation order is irrelevant (since the bytestring is immutable) and duplication doesn't change semantics (for the same reason) |
| 2025-02-12 14:30:00 +0100 | <bgamari> | regarding unsafeDupablePerformIO, the argument is that we "own" the bytestring and none of the operations under `withBsPtr` are side-effecting (e.g. writes) |
| 2025-02-12 14:29:08 +0100 | <bgamari> | since you will gain nothing by suspending the work of the parsing (which is typically fairly "light" computationally) |
| 2025-02-12 14:28:44 +0100 | <bgamari> | the short answer is: when you are parsing and can't incrementally consume the parsed result then you likely want to be strict |
| 2025-02-12 14:27:59 +0100 | <bgamari> | fair enough |
| 2025-02-12 14:27:48 +0100 | <haskellbridge> | <Profpatsch> but even just an explanation of what needs to be strict and what doesn’t would be super helpful since it’s such a black art |
| 2025-02-12 14:27:47 +0100 | Digit | (~user@user/digit) Digit |
| 2025-02-12 14:27:43 +0100 | <bgamari> | yes |
| 2025-02-12 14:27:10 +0100 | <haskellbridge> | <Profpatsch> well I’m assuming "return $!" is used everywhere because you want to be strict in the return values in the ST |
| 2025-02-12 14:27:02 +0100 | Digit | (~user@user/digit) (Read error: Connection reset by peer) |
| 2025-02-12 14:26:44 +0100 | <bgamari> | sure, that would be a great source of inspiration |
| 2025-02-12 14:26:37 +0100 | <bgamari> | what concretely is the question about `return $!`? |
| 2025-02-12 14:26:33 +0100 | <haskellbridge> | <Profpatsch> I could write down some questions |
| 2025-02-12 14:26:21 +0100 | <bgamari> | a great suggestion |
| 2025-02-12 14:26:13 +0100 | <haskellbridge> | <Profpatsch> Haskell Unfolder? :P |
| 2025-02-12 14:25:50 +0100 | sawilagar | (~sawilagar@user/sawilagar) sawilagar |
| 2025-02-12 14:25:40 +0100 | <bgamari> | Profpatch: I don't think it was ever written down |
| 2025-02-12 14:25:38 +0100 | <haskellbridge> | <Profpatsch> are these valid if they are allowed to be inlined? I guess go_fast acts as a natural boundary that is never inlined because of its size |
| 2025-02-12 14:24:38 +0100 | <haskellbridge> | <Profpatsch> Up to go_fast that is, which does not have any pragma |
| 2025-02-12 14:24:25 +0100 | <haskellbridge> | <Profpatsch> One more question: Stuff like https://hackage.haskell.org/package/cborg-0.2.10.0/docs/src/Codec.CBOR.Magic.html#withBsPtr uses unsafeDupablePerformIO, but all of these helpers are INLINE |
| 2025-02-12 14:23:09 +0100 | <haskellbridge> | <Profpatsch> bgamari: ^ I guess same question :) |
| 2025-02-12 14:21:41 +0100 | sawilagar_ | (~sawilagar@user/sawilagar) (Quit: Leaving) |
| 2025-02-12 14:17:20 +0100 | forell | (~forell@user/forell) forell |
| 2025-02-12 14:11:43 +0100 | florida | (~florida@2a02:ab88:7200:6a00:762b:62ff:fe83:1a1b) (Quit: Leaving) |
| 2025-02-12 14:09:39 +0100 | ljdarj1 | ljdarj |
| 2025-02-12 14:09:39 +0100 | ljdarj | (~Thunderbi@user/ljdarj) (Ping timeout: 276 seconds) |
| 2025-02-12 14:09:01 +0100 | ljdarj1 | (~Thunderbi@user/ljdarj) ljdarj |
| 2025-02-12 14:05:18 +0100 | <haskellbridge> | <Profpatsch> "return $!" is another one that I have no clue how it makes things in ST work |
| 2025-02-12 14:02:37 +0100 | Guest34 | (~Guest34@96-8-132-1.block0.gvtc.com) (Client Quit) |