Logs: freenode/#haskell
| 2020-11-11 16:57:56 | <boxscape> | would you say there's a relationship between purely functional programming and not using null references? |
| 2020-11-11 16:58:10 | <hc> | I think so |
| 2020-11-11 16:58:30 | <boxscape> | I guess trying to do something with a null reference that you can't do usually results in side effects? |
| 2020-11-11 16:58:48 | → | texasmynsted joins (~texasmyns@64.44.55.44) |
| 2020-11-11 16:58:55 | × | adm_ quits (~adm@43.229.89.44) (Ping timeout: 258 seconds) |
| 2020-11-11 16:59:04 | <hc> | you don't have/use regularily references in purely functional programming languages ;) |
| 2020-11-11 16:59:15 | × | timCF quits (~i.tkachuk@m91-129-105-245.cust.tele2.ee) (Client Quit) |
| 2020-11-11 16:59:17 | <hc> | at least not directly; it's all hidden away by smart compilers |
| 2020-11-11 16:59:19 | <maerwald> | hc: why not? |
| 2020-11-11 16:59:26 | <hc> | maerwald: less need |
| 2020-11-11 16:59:35 | → | timCF joins (~i.tkachuk@m91-129-105-245.cust.tele2.ee) |
| 2020-11-11 16:59:54 | <hc> | maerwald: due to referential transparency mostly, i think |
| 2020-11-11 17:00:01 | <maerwald> | boxscape: only practically. I don't think it matters theoretically? |
| 2020-11-11 17:00:03 | × | texasmynsted quits (~texasmyns@64.44.55.44) (Client Quit) |
| 2020-11-11 17:00:37 | <boxscape> | Just thinking about whether I can find a good segway to talk about avoiding null pointers in a talk about purely functional programming :) |
| 2020-11-11 17:00:49 | <boxscape> | er |
| 2020-11-11 17:00:51 | <boxscape> | segue |
| 2020-11-11 17:01:15 | <maerwald> | that is, it isn't obvious to me why you can't have a purely functional programming language with null pointers |
| 2020-11-11 17:01:16 | × | raehik quits (~raehik@cpc95906-rdng25-2-0-cust156.15-3.cable.virginm.net) (Read error: Connection reset by peer) |
| 2020-11-11 17:01:59 | <p0a> | scheme right? |
| 2020-11-11 17:02:08 | <hc> | maerwald: because what's the use of a null pointer when you don't have pointers/references in the first place? |
| 2020-11-11 17:02:09 | <boxscape> | I guess having "null pointers" doesn't necessarily mean regular variables/constant can have a value of null, which is how I was thinking about it |
| 2020-11-11 17:02:26 | <maerwald> | hc: I'm not talking about the use, but the definition |
| 2020-11-11 17:02:31 | <maerwald> | I don't think there's a hard correlation |
| 2020-11-11 17:02:40 | <hc> | agree, it's a soft correlation ;) |
| 2020-11-11 17:02:44 | → | raehik joins (~raehik@cpc95906-rdng25-2-0-cust156.15-3.cable.virginm.net) |
| 2020-11-11 17:03:39 | <hc> | maerwald: do pointers/references make sense in a purely functional language? |
| 2020-11-11 17:03:42 | <Chousuke> | null pointers aren't really the problem, it's when any pointer type can have null values that you run into trouble |
| 2020-11-11 17:03:54 | × | jorjoro quits (4f70b8ab@79.112.184.171) (Remote host closed the connection) |
| 2020-11-11 17:03:56 | <timCF> | Hello guys! I know it sounds stupid, but I'll ask anyway. For whatever reason my linker is extremely slow, when stack do linking phase - I literally can have time for a cup of tea or two. Maybe I'm missing something. How to make it fast? |
| 2020-11-11 17:04:09 | → | Lycurgus joins (~niemand@cpe-45-46-134-163.buffalo.res.rr.com) |
| 2020-11-11 17:04:25 | <boxscape> | Chousuke so you're saying the type of the pointer should indicate whether it can point to null? |
| 2020-11-11 17:04:55 | <bqv> | p0a: i think you can do it by modifying the c-indent family of variables |
| 2020-11-11 17:05:09 | <bqv> | but i just avoid asking emacs to indent for me |
| 2020-11-11 17:05:17 | <bqv> | because it always does it 'wrong' |
| 2020-11-11 17:05:19 | <Chousuke> | boxscape: essentially, yes. Or in other words, just support Optional/Maybe |
| 2020-11-11 17:05:21 | × | christo quits (~chris@81.96.113.213) (Remote host closed the connection) |
| 2020-11-11 17:05:26 | <boxscape> | yeah, that makes sense |
| 2020-11-11 17:05:29 | <hc> | maerwald: i'm not saying by any means i'm right about this... i merely want to understand it :) |
| 2020-11-11 17:05:36 | <p0a> | bqv: what do you do instead? |
| 2020-11-11 17:05:43 | <bqv> | use my spacebar? :p |
| 2020-11-11 17:05:53 | <p0a> | hc: yes they do |
| 2020-11-11 17:05:59 | <p0a> | hc: You can have tape and ... pointers on it |
| 2020-11-11 17:06:05 | <hc> | timCF: i'm regularily waiting for 5 minutes for the rust linker to finish in release/optimizing mode |
| 2020-11-11 17:06:19 | <maerwald> | hc: purely function is defined as: call-by-name, call-by-need and call-by-value are weakly equivalent in that language (modulo bottom) |
| 2020-11-11 17:06:32 | <maerwald> | I don't see why you can't meet that criteria with null pointers |
| 2020-11-11 17:06:52 | <hc> | maerwald: what use would a null pointer be if you can never assign another value to the variable that holds it? |
| 2020-11-11 17:06:59 | <maerwald> | because it's bottom if you deref it |
| 2020-11-11 17:07:09 | <hc> | hmm, so an error case? |
| 2020-11-11 17:07:43 | <bqv> | hc: i mean i'm currently using a pointerlike thing using the Store comonad, and as it happens my keys are 'Ptr a' so i'm abusing nullptr to be a zero value rather than Maybe-ing it |
| 2020-11-11 17:07:45 | <p0a> | hc: what use are pointers if you can't manipulate the pointees? |
| 2020-11-11 17:07:54 | <maerwald> | what matters is that the expressions are equivalent under different eval strategies if there is NO bottom |
| 2020-11-11 17:08:05 | → | __monty__ joins (~toonn@unaffiliated/toonn) |
| 2020-11-11 17:08:06 | <int-e> | p0a: they allow sharing |
| 2020-11-11 17:08:24 | <hc> | p0a: precisely. they aren't of any use as far as i can see it |
| 2020-11-11 17:08:28 | <p0a> | int-e: Not sure what you mean :) but I was trying to say that hc's point is not a good point |
| 2020-11-11 17:08:51 | <int-e> | pah, context |
| 2020-11-11 17:09:00 | × | lawr3nce quits (~lawr3nce@gateway/tor-sasl/lawr3nce) (Quit: Leaving) |
| 2020-11-11 17:09:04 | <bqv> | i prefer lualatex |
| 2020-11-11 17:09:07 | <hc> | okay, different question: *how* do i make a pointer of something in haskell? |
| 2020-11-11 17:09:10 | <p0a> | hc: Haskell is not "without side effects" |
| 2020-11-11 17:09:20 | <timCF> | hc, do you think it's normal for dev environment, without O2 flag, which is supposed to be fast I guess.. |
| 2020-11-11 17:09:22 | <hc> | p0a: i know about iorefs |
| 2020-11-11 17:09:24 | → | ransom joins (~c4264035@undergraduate-jvossen-9690.mines.edu) |
| 2020-11-11 17:09:42 | <maerwald> | p0a: IO has no side effects =) |
| 2020-11-11 17:10:03 | <maerwald> | the only side effect in haskell is unsafePerformIO (yes, that breaks the language) |
| 2020-11-11 17:10:04 | <p0a> | maerwald: yeah it's fair to say that. C has no side effects either |
| 2020-11-11 17:10:16 | <maerwald> | p0a: it does |
| 2020-11-11 17:10:20 | <p0a> | Ah I see, alright. I misunderstood you |
| 2020-11-11 17:10:20 | → | geekosaur joins (82659a09@host154-009.vpn.uakron.edu) |
| 2020-11-11 17:10:34 | <int-e> | Oh, I think I see... Well, most haskell implementations pass values by reference which are pointers. There's hardly a point in exposing this implementation detail to the user. |
| 2020-11-11 17:10:39 | × | asheshambasta quits (~user@ptr-e1lysawl9rr13i61o92.18120a2.ip6.access.telenet.be) (Ping timeout: 272 seconds) |
| 2020-11-11 17:10:52 | × | Lycurgus quits (~niemand@cpe-45-46-134-163.buffalo.res.rr.com) (Quit: Exeunt) |
| 2020-11-11 17:10:59 | <maerwald> | but they could |
| 2020-11-11 17:11:07 | <int-e> | Unless you bring actual mutation to the table, for which you'll need IO or ST. |
| 2020-11-11 17:11:47 | <boxscape> | maerwald non-terminating programs don't count as side-effecting either? |
| 2020-11-11 17:12:13 | <maerwald> | boxscape: dunno, that term is a bit more fuzzy |
| 2020-11-11 17:12:31 | <int-e> | boxscape: the usual formal semantics will treat that case as bottom. |
| 2020-11-11 17:12:39 | <bqv> | IO at least has the side effect of p***ing me off a lot |
| 2020-11-11 17:12:45 | <int-e> | the structure of bottoms... is rarely captured formally. |
| 2020-11-11 17:12:45 | <boxscape> | oh, right, I guess that makes sense |
| 2020-11-11 17:12:57 | × | v_m_v_ quits (~vm_v@2a02:aa12:3200:6480:fc4f:fb56:796a:9a4a) (Remote host closed the connection) |
| 2020-11-11 17:13:16 | → | v_m_v_ joins (~vm_v@2a02:aa12:3200:6480:fc4f:fb56:796a:9a4a) |
| 2020-11-11 17:13:40 | → | jamm_ joins (~jamm@unaffiliated/jamm) |
| 2020-11-11 17:13:44 | → | Sanchayan joins (~Sanchayan@122.181.216.76) |
| 2020-11-11 17:13:51 | → | invaser joins (~Thunderbi@128-124-107-154.mobile.vf-ua.net) |
| 2020-11-11 17:13:57 | <hekkaidekapus> | boxscape: If your talk is not geared to any particular language, you can mention ATS about sound pointer manipulation. |
| 2020-11-11 17:13:58 | <hc> | bqv: oh i thought that was io's main purpose? ;p |
| 2020-11-11 17:14:12 | <hc> | timCF: nah, that should be much faster |
| 2020-11-11 17:14:33 | <bqv> | hdh |
| 2020-11-11 17:14:35 | <bqv> | heh* |
| 2020-11-11 17:14:59 | <int-e> | hope dat helps? |
| 2020-11-11 17:15:00 | <int-e> | :) |
| 2020-11-11 17:15:16 | <boxscape> | hekkaidekapus it is geared towards, of all things, Java :) |
| 2020-11-11 17:15:22 | <hekkaidekapus> | For NULL, you handle it as a Maybe like Chousuke said, specifically using a dataview . |
| 2020-11-11 17:15:25 | <maerwald> | boxscape: https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.106.364&rep=rep1&type=pdf |
| 2020-11-11 17:15:29 | <p0a> | If I do import Data.Text (Text) followed by import qualified Data.Text as T, does it make sense? |
| 2020-11-11 17:15:40 | <maerwald> | it doesn't properly define side effect either |
| 2020-11-11 17:15:45 | <bqv> | p0a: yes |
| 2020-11-11 17:15:50 | <p0a> | thank you bq |
All times are in UTC.