Home freenode/#haskell: Logs Calendar

Logs: freenode/#haskell

←Prev  Next→
Page 1 .. 73 74 75 76 77 78 79 80 81 82 83 .. 5022
— 502,152 events total
2020-09-19 18:45:48 <monochrom> To a large extent every GHC extension paper is of much higher quality than Haskell 2010 quality.
2020-09-19 18:46:19 <monochrom> Every such paper painstakingly writes down typing rules. There is none in the Haskell Reports.
2020-09-19 18:46:27 <sm[m]> nice tomsmeding
2020-09-19 18:46:56 <davean> It would be nice if one could clean-room impliment Haskell from the report without other knowlege than how to read it and anything explicitely referenced. We don't have that.
2020-09-19 18:47:05 <tomsmeding> (formatting is debatable)
2020-09-19 18:47:07 <davean> Requiring that is a weird standard to update other parts of the report.
2020-09-19 18:47:21 <monochrom> At this rate, adding most GHC extensions to the next Haskell Report is trivial, just write down the syntax, there is no need to write down the typing rules or the dynamic semantics, thus "no need", har har har.
2020-09-19 18:47:50 → wavemode joins (~wavemode@097-070-075-143.res.spectrum.com)
2020-09-19 18:48:19 <geekosaur> monochrom, the downside is that the typing rules often assume ghc's internal type system, which I'm not sure anyone wants to take as the basis for a Haskell standard
2020-09-19 18:48:52 → jumper149 joins (~jumper149@ip185225.wh.uni-hannover.de)
2020-09-19 18:48:58 <maerwald> tomsmeding: how do you download a paste?
2020-09-19 18:49:07 <geekosaur> and nobody wants to do the work to specify a minimal type system for the Report
2020-09-19 18:49:21 <tomsmeding> maerwald: there's the [raw] link for the first file
2020-09-19 18:49:31 <maerwald> tomsmeding: I mean all files
2020-09-19 18:49:36 <davean> geekosaur: right.
2020-09-19 18:49:41 <tomsmeding> you don't, currently :p
2020-09-19 18:49:51 <tomsmeding> how would you expect that to work, zip file?
2020-09-19 18:49:58 <dmwit> ski: Counterpoint: the obvious alternative is for forall to take the variable out of scope. But foralls are overloaded now to also be part of the TypeApplications API.
2020-09-19 18:50:02 <tomsmeding> also how often do you really need that
2020-09-19 18:50:14 <lechner> Hi, why do I have to use the <- assignment operator on the result from getArgs, but not on the result from readFile, please? I see that the readLine result is not 'tainted' by the IO monad, but I do not otherwise see a distinction between these commands vs expressions.
2020-09-19 18:50:18 <dmwit> ski: So that seems like it would be sort of a bad conflict in what you're able to express.
2020-09-19 18:50:41 <dmwit> lechner: You *do* have to use <- on the result from readFile. Why do you believe you don't?
2020-09-19 18:50:48 <wwwww> Let's say i would like to run something with forkIO, in OSX only single thread spawns but in linux there are many of them, why?
2020-09-19 18:50:56 <Uniaika> < geekosaur> and nobody wants to do the work to specify a minimal type system for the Report // yeah, it takes time, effort and coordination, and I feel like we're too much in maintenance mode for that kind of thing to happen
2020-09-19 18:50:56 <dmwit> lechner: Also, I challenge your claim that readLine is not 'tainted'.
2020-09-19 18:51:29 <lechner> never mind, i thought i passed the result directly to putStr when i fact I placed a =<< in between
2020-09-19 18:51:29 <dolio> wwwww: No threaded runtime maybe?
2020-09-19 18:51:34 <dmwit> wwwww: Perhaps the compiler you're using on Linux has defaulted to the threaded runtime, but the one on OSX hasn't?
2020-09-19 18:51:46 <maerwald> tomsmeding: tar.gz
2020-09-19 18:51:49 <davean> wwwww: forkIO is not forkOS
2020-09-19 18:52:06 <wwwww> I know they are different but i get same result
2020-09-19 18:52:09 <maerwald> zip is trash
2020-09-19 18:52:15 × irc_user quits (uid423822@gateway/web/irccloud.com/x-jxviwzlqotrdnqhx) (Quit: Connection closed for inactivity)
2020-09-19 18:52:17 <lechner> does <- force an evaluation, in some sense ?
2020-09-19 18:52:43 <tomsmeding> maerwald: fair :p
2020-09-19 18:52:48 <dmwit> lechner: No, it forces an execution.
2020-09-19 18:52:55 <tomsmeding> how many times have you wanted such functionality already
2020-09-19 18:53:02 <maerwald> all the time man
2020-09-19 18:53:10 <ski> dmwit : hm, point. however, if we get `foo @a ... = ..a..', then that could be combined with `foo :: forall a. ..a..'
2020-09-19 18:53:11 <davean> wwwww: forkIO is a greenthread. Why would it produce a new OS thread generally? Thats a RTS decision.
2020-09-19 18:53:19 <lechner> dmwit: in the world of haskell, that's not the same thing?
2020-09-19 18:53:22 tabemann_ is now known as tabemann
2020-09-19 18:53:29 <dmwit> lechner: "evaluation" is the process of reducing expressions to other expressions. "execution" is the process of submitting terms to the runtime system to request that effects be performed.
2020-09-19 18:53:35 <davean> lechner: not at all the same.
2020-09-19 18:53:49 → MaoZeDong_ joins (~yuri@2a00:1370:8135:91d4:956f:2ce4:9ce3:de8c)
2020-09-19 18:53:58 <dmwit> ski: Neat idea!
2020-09-19 18:54:05 × knupfer quits (~Thunderbi@mue-88-130-61-068.dsl.tropolys.de) (Ping timeout: 260 seconds)
2020-09-19 18:54:10 <tomsmeding> maerwald: personally I'd barely ever use it, only on the wildest pastes with >3 files, but I guess other people work differently :p
2020-09-19 18:54:20 <monochrom> Well, beginners from imperative backgrounds do not already know that, or why, we need two separate "eval" and "exec" notions.
2020-09-19 18:54:24 <dolio> lechner: It's kind of jargon invented here.
2020-09-19 18:54:37 <ski> dmwit : i dunno whether it will happen, but i can still complain about it ;)
2020-09-19 18:54:40 <dolio> People tend to use it before explaining it for some reason. :)
2020-09-19 18:54:43 <dmwit> ski: Hm, I'm not sure I understand that, though. How is the order of variables determined from `foo @a ... = ...`?
2020-09-19 18:54:49 <merijn> davean: eh, note that forkOS doesn't spawn an OS thread either
2020-09-19 18:54:50 <geekosaur> I've used the gist.github.com equivalent a few times. Not often, but I also haven't encountered that many multi-file pastes
2020-09-19 18:54:57 <lechner> dmwit: doesn't an execution necessitate an evaluation?
2020-09-19 18:54:59 <monochrom> So you need to be gentle in the sense that you can't just paint it as "everyone else already knows" truth. Every else doesn't already know.
2020-09-19 18:54:59 <dmwit> ski: Are you suggesting that the `a` in `@a` must have the same name as in the type signature?
2020-09-19 18:55:05 <ski> dmwit : `foo :: forall a. forall b. ..a..b..; foo @a @b ... = ...'
2020-09-19 18:55:06 <dmwit> lechner: Very often, yes. But not always.
2020-09-19 18:55:09 <ski> dmwit : no
2020-09-19 18:55:14 <davean> merijn: yes, I was getting there.
2020-09-19 18:55:34 × supercoven quits (~Supercove@dsl-hkibng32-54fb54-166.dhcp.inet.fi) (Read error: Connection reset by peer)
2020-09-19 18:55:38 <dmwit> ski: Then I don't understand how it helps clear up the conflict I proposed.
2020-09-19 18:55:39 <davean> merijn: threading is hard
2020-09-19 18:55:40 <monochrom> The next surprising divergence will be how, and why, parsing is different from eval order.
2020-09-19 18:55:41 <wwwww> I tried forkOS/forkIO and as result still i have single thread in OSX davean
2020-09-19 18:55:46 → son0p joins (~son0p@186.159.4.142)
2020-09-19 18:55:49 tomsmeding notes that the people coming here complaining that the channel is quiet should probably return during US waking times
2020-09-19 18:55:55 <dmwit> ski: If forall is used both for variable hiding and variable order, how to I express the order of unhidden variables?
2020-09-19 18:56:03 <davean> wwwww: sure, but my point is theres no reason to tihnk either of them would create a thread in particular.
2020-09-19 18:56:08 <lechner> i guess commands can perhaps expect an expression, too
2020-09-19 18:56:10 <davean> wwwww: How many threads is an RTS decision
2020-09-19 18:56:11 <merijn> wwwww: What are you trying to do?
2020-09-19 18:56:11 <geekosaur> what are you using to determine threads? they look different in OS X vs. Linux
2020-09-19 18:56:17 <davean> which is what the RTS options are about.
2020-09-19 18:56:23 <monochrom> Europe waking time is also pretty active here.
2020-09-19 18:56:25 <ski> dmwit : `zip :: forall a. forall b. [a] -> [b] -> [(a,b)]; zip @u @v (xs :: [u]) (ys :: [v]) = ..u..v..xs..ys.' should be fine
2020-09-19 18:56:28 <davean> geekosaur: yes, process vs. not
2020-09-19 18:56:33 <geekosaur> in particular, on linux they show up in ps becuase of how linux implements threads
2020-09-19 18:56:53 <ski> dmwit : each defining equation could name the type parameters differently, if it wanted to
2020-09-19 18:56:53 <wwwww> merijn: i'm building server for my game
2020-09-19 18:56:58 <dmwit> ski: Ahhh, so the foralls give the order, and the @ patterns give what's hidden?
2020-09-19 18:57:03 <dmwit> (well, what's not hidden)
2020-09-19 18:57:04 <ski> yes
2020-09-19 18:57:12 <dmwit> Got it! Very sensible.
2020-09-19 18:57:13 <davean> ski: I like that.
2020-09-19 18:57:18 <davean> ski: Please give!
2020-09-19 18:57:23 <tomsmeding> monochrom: apparently the Europeans are active in the evening then :)
2020-09-19 18:57:23 <ski> give ?
2020-09-19 18:57:28 <merijn> wwwww: "I need to concurrently handle many connections" (that's what forkIO is for) and "I need multiple OS threads" are orthogonal issues
2020-09-19 18:57:41 <davean> ski: please give me that as how I get to do it :)
2020-09-19 18:57:43 <dmwit> ski: I think he's asking you to implement it and send him a patch. ^_^
2020-09-19 18:57:56 <merijn> wwwww: You can spawn many threads handling independent conncections using a single OS thread, using multiple OS threads is optional
2020-09-19 18:58:30 → exarkun1 joins (~exarkun@193.56.252.210)
2020-09-19 18:58:36 <ski> dmwit : alternatively, you should be able to say `zip :: [a] -> [b] -> [(a,b)]; zip (xs :: [a]) (ys :: [b]) = ..a..b..xs..ys..', of course. but then the order of the type parameters isn't clear (as is already the case with implicit `forall')
2020-09-19 19:00:17 <ski> hm .. i suppose, if this would have any chance of being accepted, it would probably be a differently named language extension than `ScopedTypedVariables'
2020-09-19 19:00:38 <dolio> ski: Yeah, that's the problem.
2020-09-19 19:00:45 <maerwald> south-east asia time is pretty silent
2020-09-19 19:00:45 <dolio> Too late to fix the design.
←Prev  Next→
Page 1 .. 73 74 75 76 77 78 79 80 81 82 83 .. 5022

All times are in UTC.