Home freenode/#haskell: Logs Calendar

Logs: freenode/#haskell

←Prev  Next→
Page 1 .. 480 481 482 483 484 485 486 487 488 489 490 .. 5022
— 502,152 events total
2020-10-07 07:11:51 → jinblack- joins (~JinBlack@2001:41d0:8:4b28::1)
2020-10-07 07:12:30 → lyxia_ joins (~lyxia@poisson.chat)
2020-10-07 07:12:50 → jluttine_ joins (~jluttine@87-95-204-180.bb.dnainternet.fi)
2020-10-07 07:13:03 × vicfred quits (~vicfred@unaffiliated/vicfred) (Quit: Leaving)
2020-10-07 07:13:03 × jinblack quits (~JinBlack@new.jinblack.it) (Ping timeout: 272 seconds)
2020-10-07 07:13:03 × iteratee_ quits (~kyle@162.211.154.4) (Read error: Connection reset by peer)
2020-10-07 07:13:03 × lyxia quits (~lyxia@poisson.chat) (Ping timeout: 272 seconds)
2020-10-07 07:13:03 × jluttine quits (~jluttine@87-95-204-180.bb.dnainternet.fi) (Ping timeout: 272 seconds)
2020-10-07 07:13:03 × deu quits (de@uio.re) (Ping timeout: 272 seconds)
2020-10-07 07:13:23 → deu joins (de@uio.re)
2020-10-07 07:16:29 <dminuoso> % setr (each . _Left) 1 [Left 'c', Left 'd', Right "foo"] -- ski
2020-10-07 07:16:30 <yahb> dminuoso: ([1,1],[Left 1,Left 1,Right "foo"])
2020-10-07 07:16:36 <dminuoso> % setr :: Traversal s t a b -> b -> s -> ([b], t); setr l b = runRemember #. l (\_ -> Remember ([b], b))
2020-10-07 07:16:37 <yahb> dminuoso:
2020-10-07 07:16:48 <dminuoso> % newtype Remember b t = Remember ([b], t) deriving (Eq, Ord, Functor)
2020-10-07 07:16:49 <yahb> dminuoso:
2020-10-07 07:16:59 <dminuoso> % instance Applicative (Remember b) where pure a = Remember ([], a); Remember (fs, f) <*> Remember (gs, g) = Remember ((fs <> gs), f g)
2020-10-07 07:16:59 <yahb> dminuoso:
2020-10-07 07:17:51 <dminuoso> Also for you, unclechu, ^-
2020-10-07 07:18:55 × GyroW_ quits (~GyroW@unaffiliated/gyrow) (Quit: Someone ate my pie)
2020-10-07 07:19:06 → carlomagno1 joins (~cararell@inet-hqmc01-o.oracle.com)
2020-10-07 07:19:10 → GyroW joins (~GyroW@ptr-48ujrfd1ztq5fjywfw3.18120a2.ip6.access.telenet.be)
2020-10-07 07:19:10 × GyroW quits (~GyroW@ptr-48ujrfd1ztq5fjywfw3.18120a2.ip6.access.telenet.be) (Changing host)
2020-10-07 07:19:10 → GyroW joins (~GyroW@unaffiliated/gyrow)
2020-10-07 07:19:11 <dminuoso> It also occured to me, that you can just use (a,) instead of Remember.
2020-10-07 07:19:22 × fredcy quits (~fredcy@mail.yankowski.com) (Quit: ZNC - http://znc.in)
2020-10-07 07:19:24 × Graypup_ quits (Graypup@lfcode.ca) (Quit: ZNC 1.6.1 - http://znc.in)
2020-10-07 07:19:38 → fredcy joins (~fredcy@mail.yankowski.com)
2020-10-07 07:19:38 → Graypup__ joins (Graypup@lfcode.ca)
2020-10-07 07:20:00 × polux2001 quits (~polux@51.15.169.172) (Quit: Ping timeout (120 seconds))
2020-10-07 07:20:01 × drupol quits (sid117588@gateway/web/irccloud.com/x-rynwclhcvdkvjpek) (Read error: Connection reset by peer)
2020-10-07 07:20:01 × alexknvl quits (sid259568@gateway/web/irccloud.com/x-shhushfytlhxjajz) (Read error: Connection reset by peer)
2020-10-07 07:20:05 × carlomagno quits (~cararell@inet-hqmc01-o.oracle.com) (Ping timeout: 240 seconds)
2020-10-07 07:20:12 → alexknvl joins (sid259568@gateway/web/irccloud.com/x-esxzjxwhwfjmhmwu)
2020-10-07 07:20:17 → polux2001 joins (~polux@51.15.169.172)
2020-10-07 07:20:18 → drupol joins (sid117588@gateway/web/irccloud.com/x-botdgnvcemakmhms)
2020-10-07 07:20:25 × abhixec quits (~abhixec@c-67-169-141-95.hsd1.ca.comcast.net) (Ping timeout: 240 seconds)
2020-10-07 07:20:44 × vonfry quits (~user@li1870-189.members.linode.com) (Remote host closed the connection)
2020-10-07 07:20:49 <dminuoso> % setr :: Traversal s t a b -> b -> s -> ([b], t); setr l b = l (\_ -> ([b], b))
2020-10-07 07:20:49 <yahb> dminuoso:
2020-10-07 07:20:53 <dminuoso> Oh yeah, that's much much simpler. :)
2020-10-07 07:21:52 → snakemas1 joins (~snakemast@213.100.206.23)
2020-10-07 07:22:22 <Squarism> dminuoso, was that for me?
2020-10-07 07:23:23 <dminuoso> No
2020-10-07 07:27:08 → mmohammadi98126 joins (~mmohammad@5.238.172.236)
2020-10-07 07:30:20 → raichoo joins (~raichoo@213.240.178.58)
2020-10-07 07:33:33 → plutoniix joins (~q@175.176.222.7)
2020-10-07 07:34:16 → justsomeguy joins (~justsomeg@unaffiliated/--/x-3805311)
2020-10-07 07:35:06 × GyroW quits (~GyroW@unaffiliated/gyrow) (Quit: Someone ate my pie)
2020-10-07 07:35:11 → livvy joins (~livvy@gateway/tor-sasl/livvy)
2020-10-07 07:35:24 → GyroW joins (~GyroW@ptr-48ujrfd1ztq5fjywfw3.18120a2.ip6.access.telenet.be)
2020-10-07 07:35:25 × GyroW quits (~GyroW@ptr-48ujrfd1ztq5fjywfw3.18120a2.ip6.access.telenet.be) (Changing host)
2020-10-07 07:35:25 → GyroW joins (~GyroW@unaffiliated/gyrow)
2020-10-07 07:41:47 × nineonine quits (~nineonine@50.216.62.2) (Remote host closed the connection)
2020-10-07 07:41:59 → frot-lab joins (3b47f017@gateway/web/cgi-irc/kiwiirc.com/ip.59.71.240.23)
2020-10-07 07:42:27 × aarvar quits (~foewfoiew@50.35.43.33) (Ping timeout: 240 seconds)
2020-10-07 07:44:00 × heatsink quits (~heatsink@107-136-5-69.lightspeed.sntcca.sbcglobal.net) (Remote host closed the connection)
2020-10-07 07:44:39 <siraben> If Foo is a newtype constructor, is map Foo a no-op?
2020-10-07 07:46:10 <dminuoso> % newtype Foo a = Foo a
2020-10-07 07:46:11 <yahb> dminuoso:
2020-10-07 07:46:40 <dminuoso> siraben: I dont think so.
2020-10-07 07:46:50 <dminuoso> But you can use coerce if you want it to be free.
2020-10-07 07:46:50 <siraben> dminuoso: I would need to use coerce right?
2020-10-07 07:46:52 <dminuoso> Right
2020-10-07 07:47:23 <tomsmeding> which begs the question: why _isn't_ it a no-op?
2020-10-07 07:47:23 <siraben> Ok
2020-10-07 07:47:46 <siraben> tomsmeding: looks like it's in the paper https://www.microsoft.com/en-us/research/uploads/prod/2018/05/coercible-JFP.pdf
2020-10-07 07:47:59 <siraben> The optimiser in the Glasgow Haskell Compiler (GHC) is powerless to fix the problem because it
2020-10-07 07:47:59 <siraben> works over a typed intermediate language; the MkHTML constructor changes the type of its operand, and hence cannot be optimised away.
2020-10-07 07:48:34 <tomsmeding> I see
2020-10-07 07:48:45 <tomsmeding> that's unfortunate
2020-10-07 07:51:19 <tomsmeding> still, that claim is untrue
2020-10-07 07:51:27 <tomsmeding> the map could be replaced with a coerce!
2020-10-07 07:51:31 <tomsmeding> and the result would still typecheck
2020-10-07 07:52:06 <tomsmeding> that's quite a specific optimisation, but "cannot be optimised away" is too strong I think
2020-10-07 07:52:19 <siraben> can it be optimized away without coerce
2020-10-07 07:53:02 <dminuoso> tomsmeding: Given the 4 authors of the paper, Im quite convinced they do know what they're talking about. :p
2020-10-07 07:53:10 <tomsmeding> dminuoso: fair point :p
2020-10-07 07:53:29 <tomsmeding> perhaps they meant that it could not be optimised without coerce, as siraben said
2020-10-07 07:53:34 <tomsmeding> but then now we have coerce...? :p
2020-10-07 07:53:40 <tomsmeding> anyhow
2020-10-07 07:53:58 <dminuoso> Well, coerce is the *result* of that paper.
2020-10-07 07:53:59 <siraben> tomsmeding: the paper is about implementing coerce...
2020-10-07 07:54:04 <tomsmeding> I know!
2020-10-07 07:54:19 → borne joins (~fritjof@2001:638:708:30da:5a9f:5f48:16a:a1d1)
2020-10-07 07:54:21 <tomsmeding> I'm asking about now, 4 years after that paper :p
2020-10-07 07:54:32 <siraben> I think claim is still true then? Seems like you would have to insert coerce manually
2020-10-07 07:54:33 <tomsmeding> why map Foo is apparently not a no-op
2020-10-07 07:54:35 <siraben> Would have to check GHC Core to be sure.
2020-10-07 07:54:53 <tomsmeding> so my core (heh) question is, why doesn't ghc insert a coerce as an optimisation
2020-10-07 07:55:48 <siraben> fmap Foo is a no-op because of the functor law, right?
2020-10-07 07:55:55 <siraben> fmap id = idF
2020-10-07 07:56:11 <dminuoso> 09:55:48 siraben | fmap Foo is a no-op because of the functor law, right?
2020-10-07 07:56:18 <dminuoso> GHC does not use these laws.
2020-10-07 07:56:26 <dminuoso> It cant because it doesn't know whether they are even satisfied.
2020-10-07 07:57:05 <dminuoso> (This is why things like Co/Yoneda are so useful, because they help GHC fuse without knowing about functor laws)
2020-10-07 07:57:29 × cole-h quits (~cole-h@c-73-48-197-220.hsd1.ca.comcast.net) (Ping timeout: 265 seconds)
2020-10-07 07:57:50 <siraben> Ah I have not used the yoneda library
2020-10-07 07:58:06 <dminuoso> So it'd have to reconstruct that knowledge from the definition of map (if its inlined at all!)
2020-10-07 07:58:21 <dminuoso> And I dont know whether GHC can do that at all.
←Prev  Next→
Page 1 .. 480 481 482 483 484 485 486 487 488 489 490 .. 5022

All times are in UTC.