Home freenode/#haskell: Logs Calendar

Logs: freenode/#haskell

←Prev  Next→
Page 1 .. 74 75 76 77 78 79 80 81 82 83 84 .. 5022
— 502,152 events total
2020-09-19 19:00:51 <davean> NonGrungyScoping
2020-09-19 19:01:04 <ski> yea. i remember i complained about this, when i first learned of `ScopedTypedVariables'
2020-09-19 19:01:22 <davean> yah, ScopedTypeVariables is exactly the opposite of what makes sense.
2020-09-19 19:01:33 <davean> its one of the reasons it has to be a report thing to clean it up.
2020-09-19 19:01:47 <davean> Things have rotted because of it being an extension.
2020-09-19 19:01:51 <monochrom> BoundTypeVariables
2020-09-19 19:01:52 <ski> also .. i'm missing `PatternSignatures' ..
2020-09-19 19:02:09 <dolio> The only way it makes 'sense' is that it doesn't change the meaning of any Haskell 98 programs.
2020-09-19 19:02:25 <davean> dolio: which is exactly what we have a report to do!
2020-09-19 19:02:44 × gestone quits (~gestone@c-73-97-137-216.hsd1.wa.comcast.net) (Ping timeout: 260 seconds)
2020-09-19 19:02:51 <dmwit> But there are other extensions which change the meaning of Haskell 98 programs.
2020-09-19 19:02:59 × wavemode quits (~wavemode@097-070-075-143.res.spectrum.com) (Ping timeout: 240 seconds)
2020-09-19 19:03:09 <davean> Yah but I think they're all much newer?
2020-09-19 19:03:14 <dolio> Yeah. GHC tries to avoid it a lot, though.
2020-09-19 19:03:14 <dmwit> MonoLocalBinds or whatever that one's called, ExtendedDefaulting, etc.
2020-09-19 19:03:17 <ski> first, sometimes i want to write stuff like `zip (xs :: [a]) (ys :: [b]) :: [(a,b)] = ..a..b..xs..ys..', annotating the return type directly (not after the body)
2020-09-19 19:03:47 <dolio> Almost everything is designed to be a sort of conservative extension.
2020-09-19 19:03:56 <ski> second, i'm not sure this would have to be mixed in with the explicit `forall's, and explicit type applications, extensions
2020-09-19 19:04:33 <monochrom> There was a time Haskell Reports made good-bye-to-the-past changes. (I refuse to say "breaking changes".)
2020-09-19 19:04:34 <ski> (also, i suppose, existential arguments)
2020-09-19 19:04:57 <dolio> If only people had decided change was bad in 1.4. :)
2020-09-19 19:05:17 <monochrom> In particular especially transitioning from the [Response]->[Request] I/O system to today's IO I/O system.
2020-09-19 19:05:19 ski . o O ( `map',`(++)',monad comprehensions,`Eval' )
2020-09-19 19:05:30 <monochrom> Yeah that one.
2020-09-19 19:05:56 <monochrom> Look at how there was not even a bridging library offered.
2020-09-19 19:06:47 <monochrom> The types Response and Request simply vanished in a puff of logic. Old code simply couldn't be re-compiled.
2020-09-19 19:06:50 <wwwww> But still it doesn't explain why my program runs with more threads in linux rather than osx which only spawns single thread, i'm using stack so there should be no compiler or any configuration difference except stack version, merijn
2020-09-19 19:07:00 <monochrom> We need that bravity again today.
2020-09-19 19:07:34 geekosaur asks again: how are you determining how many OS threads are used?
2020-09-19 19:08:02 <ski> dmwit : btw, in case it wasn't clear, i was suggesting both styles of defining (in the example) `zip' above should be possible. both with explicit `forall' (not scoping over defining equations), possibly combined with type application patterns, and implicit `forall', tyvars scoping over definition
2020-09-19 19:08:36 <monochrom> Yeah I only get one thread in Linux, unless I go out of my way to use non-default settings.
2020-09-19 19:08:45 <dolio> I wonder how many people were using [Response] -> [Request]. There was also the continuation-based wrappers, which were basically monadic IO with different types.
2020-09-19 19:09:48 <ski> not sure, but i'd suspect more were using the CPS, since it was easier to get compositional
2020-09-19 19:10:02 <geekosaur> +RTS -N, which I think requires building with --rts-options
2020-09-19 19:10:13 <dolio> Yeah, it seems like the CPS stuff would be obviously superior.
2020-09-19 19:10:14 <ski> (and also to not get out-of-synch, or premature-forcing, bugs)
2020-09-19 19:10:24 <geekosaur> (but I'd expect more than one thread anyway due to the I/O manager using several internally)
2020-09-19 19:10:51 × wagle quits (~wagle@quassel.wagle.io) (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.)
2020-09-19 19:11:04 <monochrom> But multiple OS threads requires firstly someone explicitly saying "-threaded" at link time.
2020-09-19 19:11:27 → wagle joins (~wagle@quassel.wagle.io)
2020-09-19 19:11:30 <dmwit> wwwww: Didn't my very first response from ages ago propose an explanation for that? I feel as though dolio and I have been ignored this whole time.
2020-09-19 19:11:36 <ski> hm, i thought i saw a dialogue-based I/O implemented on top of `IO' (using lazy I/O), some time ?
2020-09-19 19:11:54 <merijn> You can just set the number of threads from the code, though
2020-09-19 19:12:02 <dmwit> Only if you have the right runtime.
2020-09-19 19:12:06 <monochrom> OK I'm out of that threading question because geekosaur's question nails it.
2020-09-19 19:12:07 <merijn> true
2020-09-19 19:12:21 <lechner> If anyone is using Debian: In which -dev package can i find one of the apparently many SHA-512 cryptographic primitives available in Haskell, please?
2020-09-19 19:12:22 <ski> (but then there's the extensibility problem over the raw response-to-request mapping, especially wrt FFI)
2020-09-19 19:14:16 <monochrom> ski: Do you have "interact" in mind? Or is it one of those free monads?
2020-09-19 19:14:20 <sm[m]> lechner: why do you want a debian -dev package, did the haskell package give a build error ? If so you could paste that
2020-09-19 19:14:43 → irc_user joins (uid423822@gateway/web/irccloud.com/x-yamugsoddzfdaipx)
2020-09-19 19:14:55 <ski> monochrom : something along the lines of it, but more involved, to deal with the richer `Request' and `Response' types
2020-09-19 19:15:01 → Thra11 joins (~Thra11@5.1.169.217.in-addr.arpa)
2020-09-19 19:15:19 <geekosaur> we tend to prefer getting packages from hackage or stackage over using OS packages, because the latter are often old
2020-09-19 19:15:23 → sfvm joins (~sfvm@37.228.215.148)
2020-09-19 19:15:37 <geekosaur> and stack will usually ignore them anyway
2020-09-19 19:16:31 <sm[m]> oh, I wasn't sure if they are looking for a debian c package or a debian haskell package
2020-09-19 19:16:32 <wwwww> dmwit: Forgive me, you gave me good point for further research about this kind of thing, but one of the reason i use stack instead of just ghc is avoid this kind of situations, i provide my compiler version in one of those .yml files and program should behave same in every platform, at least thats what i expect
2020-09-19 19:16:54 <lechner> sm[m]: i'll use stackage. i am just trying to figure out how
2020-09-19 19:17:13 <dmwit> Well, that just seems naive.
2020-09-19 19:17:39 <dmwit> I'm not even sure what "behave same" even means, TBH.
2020-09-19 19:17:44 <sm[m]> wwwww: did you answer the question about how you're counting threads ?
2020-09-19 19:18:37 <dmwit> You need some theory of observations that accounts for the differences between OS's, and then some relation describing "same" observations on the two. Seems complicated to set that all up carefully.
2020-09-19 19:18:54 × justsomeguy quits (~justsomeg@unaffiliated/--/x-3805311) (Ping timeout: 256 seconds)
2020-09-19 19:19:07 <wwwww> sm[m]: i'm watching the htop and i hope it is the right way to count them
2020-09-19 19:19:14 <dmwit> e.g. look at Wine, an attempt to formalize the connection between Windows observations and Linux observations. It is many hundreds of megabytes of description to get that right.
2020-09-19 19:20:11 <monochrom> I wouldn't say it's hundreds of megabytes. I would follow Shannon's advice and gzip it. Then that's a better measure. >:)
2020-09-19 19:20:32 <davean> wwwww: except also the platform isn't the same and you're asking a platform thing.
2020-09-19 19:20:38 → iqubic joins (~user@2601:602:9500:4870:958d:950f:29cc:a6c4)
2020-09-19 19:20:45 <dmwit> It's certainly more than what you would comfortably transmit into an IRC channel full of people...
2020-09-19 19:21:32 <davean> monochrom: several thens of megabytes then :-p
2020-09-19 19:21:37 <sm[m]> wwwww: cool. You know about htop's setting for showing just processes or all threads, I guess
2020-09-19 19:21:41 <monochrom> Yeah!
2020-09-19 19:21:44 <iqubic> What packages do people recommend for writing Haskell code in Emacs on NixOS?
2020-09-19 19:21:49 × Thra11 quits (~Thra11@5.1.169.217.in-addr.arpa) (Ping timeout: 264 seconds)
2020-09-19 19:22:03 <lechner> on the subject of stackage, is the advice given from here usually sounds? https://www.fpcomplete.com
2020-09-19 19:22:08 <maerwald> iqubic: packages?
2020-09-19 19:22:15 <maerwald> emacs packages?
2020-09-19 19:22:31 <iqubic> yeah. I want to know which Emacs Packages people recommend.
2020-09-19 19:22:35 × josh quits (~josh@c-67-164-104-206.hsd1.ca.comcast.net) (Remote host closed the connection)
2020-09-19 19:22:50 <sm[m]> lechner: certainly, that's the home of stack
2020-09-19 19:22:52 <davean> monochrom: I'm sorry, its split between a bunch of pieces but its at least 50MiB compressed. Can we round?
2020-09-19 19:23:10 <davean> monochrom: I'm tired of addig up package sizes at 50MiB :-p
2020-09-19 19:23:11 <lechner> iqubic: https://wiki.haskell.org/Emacs ?
2020-09-19 19:23:14 <monochrom> Yes, no need to be very precise.
2020-09-19 19:23:25 <dolio> davean: The slackware download is like 67MB.
2020-09-19 19:23:38 <davean> dolio: Ok, maybe I almost got to the end.
2020-09-19 19:23:59 <iqubic> Is lsp-haskell still under development?
2020-09-19 19:24:26 <dolio> Thank goodness slackware still apparently exists.
2020-09-19 19:24:27 × MaoZeDong_ quits (~yuri@2a00:1370:8135:91d4:956f:2ce4:9ce3:de8c) (Ping timeout: 272 seconds)
2020-09-19 19:24:35 <sm[m]> iqubic: I think it's called haskell-language-server (or hls) now, unless that's something else. And it's under very active development in #haskell-ide-engine
2020-09-19 19:24:48 <iqubic> Is it ready for personal use?
2020-09-19 19:24:49 <monochrom> Well if COBOL still lives, why not slackware too. :)
2020-09-19 19:25:04 <merijn> iqubic: Define ready :p
2020-09-19 19:25:06 × mirrorbird quits (~psutcliff@2a00:801:44a:a00b:20c3:c64:eb15:73a2) (Remote host closed the connection)
2020-09-19 19:25:06 <davean> monochrom: I don't know people being paid $500/hour to use slackware.
2020-09-19 19:25:16 <sm[m]> iqubic: actually yes, I would say it is, using VS Code and hls 0.4 release
2020-09-19 19:25:27 <iqubic> Is it better than https://github.com/jyp/dante?
2020-09-19 19:25:29 <monochrom> And WinFax for that matter.
2020-09-19 19:25:36 <sm[m]> yes, it's better than everything so far
←Prev  Next→
Page 1 .. 74 75 76 77 78 79 80 81 82 83 84 .. 5022

All times are in UTC.