Home freenode/#haskell: Logs Calendar

Logs: freenode/#haskell

←Prev  Next→
Page 1 .. 539 540 541 542 543 544 545 546 547 548 549 .. 5022
— 502,152 events total
2020-10-09 18:10:49 <dsal> I tend to reach for megaparsec early, but I also don't parse for performance much.
2020-10-09 18:11:00 <geekosaur> depends. Read itself has a decent amount of overhead
2020-10-09 18:11:21 <tomsmeding> @hoogle ByteString -> (Double, ByteString)
2020-10-09 18:11:22 <lambdabot> Algebra.RealRing splitFraction :: (C a, C b) => a -> (b, a)
2020-10-09 18:11:22 <lambdabot> NumericPrelude splitFraction :: (C a, C b) => a -> (b, a)
2020-10-09 18:11:22 <lambdabot> NumericPrelude.Numeric splitFraction :: (C a, C b) => a -> (b, a)
2020-10-09 18:11:25 × GyroW_ quits (~GyroW@unaffiliated/gyrow) (Ping timeout: 240 seconds)
2020-10-09 18:11:29 <tomsmeding> now if that would've existed :p
2020-10-09 18:11:56 <tomsmeding> @hoogle ByteString -> Maybe (Double, ByteString)
2020-10-09 18:11:57 <lambdabot> Data.ByteString.Lex.Integral readDecimal :: Integral a => ByteString -> Maybe (a, ByteString)
2020-10-09 18:11:57 <lambdabot> Data.ByteString.Lex.Integral readHexadecimal :: Integral a => ByteString -> Maybe (a, ByteString)
2020-10-09 18:11:57 <lambdabot> Data.ByteString.Lex.Integral readOctal :: Integral a => ByteString -> Maybe (a, ByteString)
2020-10-09 18:12:05 <tomsmeding> there we go, let's try that
2020-10-09 18:14:30 × ClaudiusMaximus quits (~claude@unaffiliated/claudiusmaximus) (Quit: ->)
2020-10-09 18:14:58 <ski> @type ((listToMaybe .) . evalStateT . (>>= \ns -> do "" <- StateT lex; return ns) . many . StateT) (reads :: ReadS Double)
2020-10-09 18:15:00 <lambdabot> String -> Maybe [Double]
2020-10-09 18:17:58 × Ariakenom quits (~Ariakenom@193.254.193.1) (Ping timeout: 258 seconds)
2020-10-09 18:19:12 <tomsmeding> ski: that's ~60% slower than my simple read pipeline
2020-10-09 18:19:37 <tomsmeding> and my read pipeline additionally has the advantage of not using the list monad in an obscure manner :p
2020-10-09 18:20:11 <tomsmeding> (on my machine)
2020-10-09 18:21:16 × cosimone quits (~cosimone@93-47-228-249.ip115.fastwebnet.it) (Quit: cosimone)
2020-10-09 18:21:51 → cosimone joins (~cosimone@2001:b07:ae5:db26:fe52:4567:a4e4:fc3a)
2020-10-09 18:21:56 → ButterflyOfFire joins (~Butterfly@185.163.110.116)
2020-10-09 18:22:13 × nyd quits (~lpy@unaffiliated/elysian) (Ping timeout: 260 seconds)
2020-10-09 18:22:37 × GyroW quits (~GyroW@unaffiliated/gyrow) (Quit: Someone ate my pie)
2020-10-09 18:22:47 → GyroW joins (~GyroW@d54c03e98.access.telenet.be)
2020-10-09 18:22:48 × GyroW quits (~GyroW@d54c03e98.access.telenet.be) (Changing host)
2020-10-09 18:22:48 → GyroW joins (~GyroW@unaffiliated/gyrow)
2020-10-09 18:23:53 <tomsmeding> ski: though I'm curious how you came up with that monstrosity :p
2020-10-09 18:24:44 → Guest18 joins (4f7755cb@gateway/web/cgi-irc/kiwiirc.com/ip.79.119.85.203)
2020-10-09 18:24:54 <tomsmeding> ah the bytestring way seems _significantly_ faster
2020-10-09 18:25:09 <tomsmeding> (as expected, in hindsight)
2020-10-09 18:25:23 <Guest18> Hello, I have a question regarding unbalanced binary search trees in haskell
2020-10-09 18:25:31 → ashbreeze joins (~mark@72-161-255-112.dyn.centurytel.net)
2020-10-09 18:25:42 <ski> dwts : you could compare with <https://srfi.schemers.org/srfi-67/srfi-67.html#node_sec_4.5>
2020-10-09 18:25:48 <Athas> Roughly, when you have a monad transformer stack, is it more efficient to have an Error/Except monad innermost, or an error monad transformer outermost?
2020-10-09 18:25:54 <Athas> Or does it depend on the specifics?
2020-10-09 18:26:08 <ski> tomsmeding : hm, i thought it was fairly straightforward
2020-10-09 18:27:00 → bifunc2 joins (bifunc2@gateway/vpn/protonvpn/bifunc2)
2020-10-09 18:27:01 <ski> "my read pipeline additionally has the advantage of not using the list monad in an obscure manner" -- is that an advantage ? ;p
2020-10-09 18:27:20 <tomsmeding> questionably :
2020-10-09 18:27:20 <ski> (but i don't really think it's that obscure ..)
2020-10-09 18:27:21 <tomsmeding> * :p
2020-10-09 18:27:31 <tomsmeding> but apart from how obscure your pipeline is or not
2020-10-09 18:27:37 × _ashbreeze_ quits (~mark@184-157-33-36.dyn.centurytel.net) (Ping timeout: 246 seconds)
2020-10-09 18:27:39 <tomsmeding> why did you think it would be faster than my simple read pipeline?
2020-10-09 18:28:09 <bifunc2> If unsigned integers are serialized in big endian, order is preserved. (smaller number gets a smaller bytestring)
2020-10-09 18:28:09 <bifunc2> Is there any such way to serialize signed integers, Doubles, and Floats?
2020-10-09 18:28:30 → raym joins (~ray@115.187.50.175)
2020-10-09 18:29:05 tomsmeding has no idea regarding Athas' question, pass to another willing answerer
2020-10-09 18:29:40 <Guest18> I made a small type that is a binary search tree and some functions for it, like insert, delete, etc. All is well so far, except for the function that returns the node depth for a given value. for example, if a node has a value of '3' and is 2 levels down from the root, it should return 'Just 2'. If the value searched for isn't in the tree, it
2020-10-09 18:29:40 <Guest18> returns nothing. I have a working example, but it isn't that good. Can anybody provide me with a better implementation or at least some advice?
2020-10-09 18:29:47 <dolio> Innermost vs. outermost might not have the same semantics.
2020-10-09 18:30:25 → mirrorbird joins (~psutcliff@m83-185-82-223.cust.tele2.se)
2020-10-09 18:30:54 × geekosaur quits (82659a05@host154-005.vpn.uakron.edu) (Ping timeout: 245 seconds)
2020-10-09 18:31:16 <tomsmeding> bifunc2: in the standard IEEE floating point format, positive doubles and floats already sort correctly
2020-10-09 18:31:27 <tomsmeding> negative ones sort above positive ones
2020-10-09 18:31:30 <ski> @quote Dr._Seuss
2020-10-09 18:31:30 <lambdabot> Dr._Seuss says: `type Parser a = String -> [(a,String)]' -- "A Parser for Things / is a function from Strings / to Lists of Pairs / of Things and Strings!" -- <https://willamette.edu/~fruehr/haskell/
2020-10-09 18:31:30 <lambdabot> seuss.html>
2020-10-09 18:31:43 <tomsmeding> the same holds for signed integers in 2's complement, when you write them in big endian order
2020-10-09 18:31:54 <bifunc2> tomsmeding, wow, that's mind-blowing, thanks!
2020-10-09 18:32:01 <Guest18> The code here: https://pastebin.pl/view/9ea923e2
2020-10-09 18:32:04 <Cale> Guest18: Seems like something that perhaps the Monad instance for Maybe could help to make clean
2020-10-09 18:32:05 × Jesin quits (~Jesin@pool-72-66-101-18.washdc.fios.verizon.net) (Ping timeout: 246 seconds)
2020-10-09 18:32:07 <tomsmeding> bifunc2: about the signed integers, or about the floats?
2020-10-09 18:32:36 <Guest18> Cale: I don't know how to use monads yet
2020-10-09 18:32:38 <Cale> Guest18: one sec... that's just taking a moment to load here
2020-10-09 18:32:54 <bifunc2> tomsmeding, I'm more interested in the floats. For positive Doubles and Floats, will the "store" package preserve the order on a little endian machine?
2020-10-09 18:32:57 <tomsmeding> for the floats, you can see that by noting that if 0 < A < B, then either A has a smaller exponent than B, or it has the same exponent and a smaller mantissa
2020-10-09 18:33:11 <tomsmeding> ignoring the sign bit of course
2020-10-09 18:33:13 <Cale> Guest18: does it load for you?
2020-10-09 18:33:24 <Cale> Not sure if I'm just having an issue locally...
2020-10-09 18:33:26 <Guest18> Cale: Slowly, but yeah
2020-10-09 18:33:31 <Guest18> It took a while
2020-10-09 18:33:53 <tomsmeding> bifunc2: I have no idea what the 'store' package does, but that claim about the IEEE format only holds for big-endian order
2020-10-09 18:34:02 <Cale> Maybe try dpaste.com?
2020-10-09 18:34:31 <tomsmeding> ski: it's unfortunate that that quote splits over lines halfway through a URL
2020-10-09 18:34:46 <Guest18> Cale: https://dpaste.com/GX8QSKDLD
2020-10-09 18:34:52 <Cale> thanks
2020-10-09 18:35:56 → chaosmasttter joins (~chaosmast@p200300c4a70b4001c004325a2a510724.dip0.t-ipconnect.de)
2020-10-09 18:36:00 <bifunc2> tomsmeding okay. https://hackage.haskell.org/package/store always uses machine endianness. so i'm out of luck there with the order on most machines nowadays.
2020-10-09 18:36:01 <bifunc2> However I can use cereal https://hackage.haskell.org/package/cereal-0.5.8.1/docs/Data-Serialize-IEEE754.html which can encode floats in big-endian
2020-10-09 18:36:03 <Cale> That code looks fairly reasonable to me actually
2020-10-09 18:36:11 <bifunc2> are there any alternatives to cereal that are faster nowadays?
2020-10-09 18:36:12 <Cale> You might want to use compare x root
2020-10-09 18:36:13 <Guest18> Oh, yeah, I should mention that i used to do a lookup first, to ensure that if the element isn't there, i return nothign
2020-10-09 18:36:46 <Cale> Oh, right, depthGo isn't returning Nothing, somehow
2020-10-09 18:36:57 <Guest18> but it seemed to me that it would be too slow and maybe there is a better way, but i couldn't find it
2020-10-09 18:37:02 <Guest18> that's why i came here
2020-10-09 18:37:13 <tomsmeding> bifunc2: not well-versed enough in the library ecosystem to answer that question, unfortunately
2020-10-09 18:37:39 <Guest18> for small trees, it's fine, but doing a lookup on big trees before searching for the depth once again, does add unnecesarry computation
2020-10-09 18:37:39 <Cale> Guest18: I would guess that you want depthGo _ EmptyTree acc = Nothing
2020-10-09 18:38:15 <monochrom> and the x==root case, = Just acc
2020-10-09 18:38:19 <Cale> yeah
2020-10-09 18:38:20 <bifunc2> tomsmeding, np, thanks for the other info!
2020-10-09 18:38:32 <monochrom> So generally the whole depthGo does Maybe, too.
2020-10-09 18:38:37 <Cale> and then take the Maybe stuff out of depth
2020-10-09 18:38:39 <Guest18> so modify the depthGo to return Maybe b as well
2020-10-09 18:38:49 <Cale> (it'll produce Maybe only because depthGo does)
←Prev  Next→
Page 1 .. 539 540 541 542 543 544 545 546 547 548 549 .. 5022

All times are in UTC.