00:01 (join) dsantiago
00:03 (quit) dsantiago: Client Quit
00:20 (join) dsantiago
00:32 (join) yours_truly
00:34 (quit) yours_truly: Read error: Connection reset by peer
00:54 (quit) dsantiago: Quit: Computer has gone to sleep.
00:59 (join) dsantiago
01:04 (quit) dsantiago: Client Quit
01:05 (join) didi
01:15 (join) mizu_no_oto
01:16 (join) tsion
01:16 (quit) tsion: Changing host
01:16 (join) tsion
01:20 (join) RacketCommitBot
01:20 RacketCommitBot: [racket] plt pushed 16 new commits to master: http://git.io/r54i4g
01:20 RacketCommitBot: racket/master e81f9eb Jens Axel Søgaard: Added bindings to number theoretic functions
01:20 RacketCommitBot: racket/master 8b99eb5 Jens Axel Søgaard: Headline in docs
01:20 RacketCommitBot: racket/master f68cc0a Jens Axel Søgaard: More documentation on number theoretic functions
01:20 (part) RacketCommitBot
01:22 (quit) jonrafkind: Ping timeout: 255 seconds
01:50 (quit) mizu_no_oto: Quit: Computer has gone to sleep.
01:53 aezx: why does commit bot leave after each commit? why not just let it idle?
01:53 didi: aezx: It's shy.
01:54 aezx: ha
01:54 didi doesn't know, obviously :^)
02:03 neilv: easier to program it that way (it doesn't have to be careful not to exceed limits on number of joined channels, doesn't have to manage multiple channels at once, etc.). no question of whether it's spying on channel. less resources (it doesn't get sent lots of channel traffic).
02:03 neilv: or it could just be shy, as you say
02:05 didi: Let's stick to "shy". Always good to anthropomorphize stuff.
02:06 neilv: or misanthropic
02:10 (join) nilyaK
02:15 (quit) didi: Quit: ERC Version 5.3 (IRC client for Emacs)
02:19 (quit) ynniv: Quit: ynniv
02:19 (quit) mithos28: Quit: mithos28
02:21 (join) mithos28
02:32 (quit) nilyaK: Read error: Connection reset by peer
02:49 neilv: still some work to do, but this is about the amount of info you'll see from a 7-year-old canon consumer camera. http://paste.lisp.org/display/133800
02:56 (quit) neilv: Quit: Leaving
03:12 (quit) Nisstyre-laptop: Quit: Leaving
03:20 (quit) cdidd: Remote host closed the connection
03:22 (join) cdidd
03:24 (quit) mithos28: Quit: mithos28
03:35 (quit) shriphani_: Quit: shriphani_
03:37 (join) dsantiago
03:38 (join) shriphani_
03:51 (join) lewis1711
04:10 (join) asvil
04:16 jaimef: can racket exec code say "cat *|racket -e "(for (...))"
04:17 jaimef: ahh excellent
04:29 (join) bitonic
04:29 (join) MayDaniel
04:38 (quit) Shviller: Read error: Connection reset by peer
04:43 (join) Shviller
04:49 (quit) tsion: Quit: Leaving
05:28 (join) soegaard
06:07 (join) masm
06:58 (join) Blkt
07:04 (join) dzhus
07:50 (join) nathanpc
07:52 (join) deu5
07:57 (join) greghendershott
08:23 (quit) greghendershott: Quit: Leaving.
08:25 (quit) nathanpc: Read error: Connection reset by peer
08:26 (join) nathanpc
08:36 (join) wlangstroth
08:38 (join) greghendershott
08:44 (part) lewis1711
08:53 (quit) wlangstroth: Quit: wlangstroth
09:10 (join) Fare
09:15 (quit) greghendershott: Quit: Leaving.
09:27 (join) jeapostrophe
09:27 (quit) jeapostrophe: Changing host
09:27 (join) jeapostrophe
09:29 Cryovat: jaimef: You just taught me cat *
09:29 Cryovat: Thanks :D
09:31 (nick) samth_away -> samth
09:31 (join) SummerWine
09:32 SummerWine: hi, guys! can you comment this one: http://s10.postimage.org/e1ylfxm2x/bloatware.png
09:33 Cryovat: SummerWine: I believe I have a solution for you:
09:33 Cryovat: http://www.amazon.com/How-Win-Friends-Influence-People/dp/1439167346/
09:33 SummerWine: please, stick to the theme
09:34 SummerWine: how is it possible that DrRacket takes much more memory than Visual Studio???
09:34 SummerWine: that's the real question here, not getting friends
09:34 SummerWine: take a look once again: http://s10.postimage.org/e1ylfxm2x/bloatware.png
09:34 SummerWine: and say honestly: is drracket peace of crap???
09:47 asvil: SummerWine: what is about memory limit for repl?
10:09 (join) mizu_no_oto
10:27 (quit) asvil: Ping timeout: 268 seconds
10:30 (join) rmathews
10:36 (quit) bjz: Quit: Leaving...
10:49 (join) mithos28
10:58 (quit) mizu_no_oto: Quit: Computer has gone to sleep.
11:08 (quit) cdidd: Remote host closed the connection
11:10 (join) nilyaK
11:13 (join) anRch
11:13 (quit) deu5: Remote host closed the connection
11:14 (quit) cipher: Ping timeout: 252 seconds
11:18 (join) karswell
11:26 SummerWine: http://redd.it/13egnb
11:34 (quit) acarrico: Ping timeout: 260 seconds
11:34 (quit) mithos28: Quit: mithos28
11:34 samth: SummerWine: seriously?
11:34 (quit) SummerWine: Ping timeout: 245 seconds
11:38 (join) mithos28
11:41 (quit) jrslepak: Quit: What happened to Systems A through E?
11:52 (join) acarrico
11:53 (quit) anRch: Quit: anRch
11:54 (quit) karswell: 
11:57 (quit) nilyaK: Read error: Connection reset by peer
12:01 (join) mye
12:05 (join) hash_table
12:10 (join) masm1
12:11 (quit) masm: Ping timeout: 276 seconds
12:12 (quit) rmathews: Ping timeout: 260 seconds
12:13 (join) rmathews
12:38 (join) RacketCommitBot
12:38 RacketCommitBot: [racket] plt pushed 10 new commits to master: http://git.io/WR5nAw
12:38 RacketCommitBot: racket/master 4041e65 Matthew Flatt: JIT simplification...
12:38 RacketCommitBot: racket/master be4ce3e Matthew Flatt: skip some unneeded fixnum tests in JIT-generated code
12:38 RacketCommitBot: racket/master 42f74b5 Matthew Flatt: JIT tweak: stack-clearing operations do not disturb registers
12:38 (part) RacketCommitBot
12:41 noam: SummerWine: I've been asking that myself for a while now too... but let me try and ask here a bit more constructive question, Is there some tool that lets you see WHAT the memory of a racket program is used for? (Like X Mb were allocated in this part of the program and Y MB in that part etc.)
12:46 aezx: maybe some downvotes will solve the problem...
12:48 (join) jrslepak
12:55 (join) greghendershott
12:56 (quit) rmathews: Quit: Bye..
13:07 (quit) mye: Ping timeout: 246 seconds
13:11 (quit) acarrico: Ping timeout: 252 seconds
13:11 (join) mye
13:12 (join) mye_
13:15 (quit) mye: Read error: Operation timed out
13:15 (nick) mye_ -> mye
13:15 samth: noam: yes, you can use custodians for that
13:16 samth: noam: see the docs here: http://docs.racket-lang.org/reference/eval-model.html#%28part._custodian-model%29
13:16 rudybot: http://tinyurl.com/ahonprf
13:17 samth: noam: and the paper here: http://www.cs.utah.edu/plt/publications/ismm04-wf.pdf
13:18 mithos28: samth: is there a reason that all of the for/...: code is just trying to annotate existing for/... stuff instead of rewriting it
13:18 samth: mithos28: to avoid the complication of re-implementing it, and probably getting it wrong
13:18 mithos28: I just wrote a simple for/hash: and for*/hash: that desugar to for/fold: and for*/fold:
13:19 (quit) walter: Remote host closed the connection
13:19 mithos28: the difference is mine actually typecheck, while the standard ones do not
13:19 samth: mithos28: if you're using for/fold, then that's reasonable
13:20 mithos28: https://gist.github.com/4106637
13:20 samth: i meant not reimplementing the whole for machinery
13:20 mithos28: yeah, that would be unreasonable
13:21 samth: mithos28: those look like good replacements for the current implementations of `for/hash:`
13:21 mithos28: ok, cool.
13:21 mithos28: They have the issue that they need the return-annotation
13:22 mithos28: even where the context provides enough information
13:22 samth: mithos28: probably the fix for that is a generalization of the 'called-in-tail-position syntax property
13:22 samth: something like 'used-in-tail-position
13:23 mithos28: ok, will that let the (make-immutable-hash null) be checked in the same context as the full expression
13:24 mithos28: and that is also not enough because for/fold: requires the annotations on the loop variables
13:25 mithos28: So I think I also need to figure out why that restriction exists when it should be inferable from the context
13:27 samth: yeah, that's the idea
13:28 (join) acarrico
13:28 noam: samth: Thanks for the pointer, but how would one use custodians to analyze the memory usage of drracket?
13:28 (quit) Fare: Ping timeout: 240 seconds
13:28 mithos28: on a slight side note, is there a way to introduce bindings in a for loop that aren't part of a sequence
13:29 samth: mithos28: see `in-val`
13:29 mithos28: (for ((av a-list) (match-define (a name fields) av) (f fields)) ...)
13:29 samth: it's kind of a trick, but that's what it's for
13:29 samth: maybe in-value?
13:30 samth: noam: drracket doesn't expose that information to users at the moment
13:30 mithos28: that is actually a for*
13:30 samth: yes
13:30 samth: mithos28: for/fold should allow un-annotated variables
13:30 samth: for example, this works: (ann (for/fold ([x 0]) ([i (in-range 10)]) (+ i x)) Integer)
13:30 samth: so we don't always need them
13:31 mithos28: ok, I'll try using my code for a bit, and if there are no issues I'll send a pull request
13:31 samth: mithos28: aweseom
13:32 mithos28: I just got sick of having to crazy gymnastics to do a for*/hash
13:32 mithos28: It involved (apply hash-union (make-immutable-hash null) (for/list ((..)) (for/hash ((..)) …))))
13:32 mithos28: + annotations
13:33 noam: well there should be a way to measure memory usage, otherwise it would never get optimized...
13:34 samth: noam: there are certainly ways, but there isn't a way that's exposed to users right now
13:34 noam: what do you mean?
13:35 samth: noam: I mean that you can modify the source of drracket to provide more information about memory use, for example
13:36 noam: Ok I see
13:37 noam: I hope somone more familiar with racket then I am would go after finding what makes drracket use so much memory though...
13:38 samth: noam: i don't think drracket uses "so much memory"
13:38 noam: almost 500MB is quite much
13:38 bremner: resident?
13:39 mithos28: is it a bug if core racket macros just give a bad syntax error instead of a descriptive message?
13:40 mithos28: rudybot: init racket
13:40 rudybot: mithos28: your sandbox is ready
13:41 samth: mithos28: yes
13:41 mithos28: ok, will file away
13:43 (quit) sethalves: Read error: Connection reset by peer
13:43 (join) sethalve_
13:44 (quit) jrslepak: Quit: What happened to Systems A through E?
13:47 (quit) sethalve_: Ping timeout: 248 seconds
13:48 (join) sethalve_
13:54 (quit) greghendershott: Quit: Leaving.
14:04 aezx: i think racket is a modern language for modern machines.
14:04 aezx: 500mb isnt a lot of memory these days
14:04 offby1: mithos28: don't keel heem!!
14:08 (quit) mithos28: Quit: mithos28
14:10 (join) untrusted
14:17 (quit) bitonic: Ping timeout: 268 seconds
14:27 (join) mithos28
14:28 mithos28: is anyone else having issues seeing today's irc logs
14:28 mithos28: it is trying to download them for me instead of view them in my browser
14:29 samth: mithos28: works fine for me in FF
14:29 samth: mithos28: chrome wants to download them though
14:29 mithos28: what about yesterday's
14:29 samth: mithos28: i blame you personally for all of your company's software
14:30 samth: yesterday works right in both
14:30 mithos28: I'll forward it to people who actually work on chrome
14:30 samth: :)
14:30 mithos28: I'm guessing something is up with the server then
14:30 (join) dca
14:31 samth: maybe, i'm hard-pressed to guess what, though
14:31 mithos28: different disposition
14:32 mithos28: http://www.ietf.org/rfc/rfc1806.txt
14:32 mithos28: content-disposition
14:32 mithos28: or its sending it it as a different mimetype
14:33 noam: aezx: I can't even begin to describe how wrong is that thinking. I don't mean that every program should be a super optimized C program, but there is a limit where inefficient becomes bloated. And imho 500mb is too much for what drracket does. A computer is not for running just one or two programs simultaneously, what if every small-ish program you run needed 500mb of memory?
14:33 (join) jrslepak
14:40 offby1: noam: it's a question of economics, really.  Is drracket _worth_ 500Mb?  To you, clearly not; to other people, presumably, yes.
14:43 bremner: especially for people who have to bear the cost of improving the memory footprint ;)
14:44 mithos28: noam: The issue is ram is cheap, and optimizing programs is expensive
14:48 bremner: hmm. Emacs and Xulrunner are bothing sitting at about 200M resident for me.
14:50 bremner: DrRacket is at about 400M resident.
14:50 (join) jonrafkind
14:52 (join) tsion
14:52 (quit) tsion: Changing host
14:52 (join) tsion
14:52 samth: with no tools loaded, drracket is about 310 M
14:52 samth: for me
14:52 samth: presumably less on a 32 bit machine
14:53 (join) SummerWine
14:53 SummerWine: mithos28: the point is: you can't show Racket superiority if your "flagship" product takes more memory than gigantic Microsoft Visual Studio!
14:53 SummerWine: let's just agree in that case: microsoft is better than you, guys :)
14:54 SummerWine: Microsoft is capable of producing low memory programming tool
14:54 SummerWine: and you just cannot
14:54 bremner: lala one troll from ignore list...
14:55 SummerWine: Microsoft, ressurect Microsoft Bob
14:55 SummerWine: :)
14:56 SummerWine: trol or not, it's not importatnt. What's important is ENORMOUS memory consumption of "superior language" DrRacket
14:56 noam: (I don't try to troll, just genuinely concerned about memory usage)
14:57 noam: (and loves lightweight software)
14:57 SummerWine: i really don't understand how this memory issue does'n concern you, academic PLT guys!??
14:57 bremner: yeah, well, if I wasn't ignoring SummerWine I'd point out that the racket interpreter uses a small fraction of the memory of DrRacket
14:58 mithos28: SummerWine: DrRacket uses similar memory to my irc client
14:58 SummerWine: bremner: so what? Microsodt command line c compiler also uses small memory footprint
14:58 SummerWine: so what?
14:58 offby1: bremner: good idea
14:58 SummerWine: but we are talking here about IDE stuff
14:59 mithos28: I don't use MS software so cannot compare to it, but I do know that DrRacket uses much less memory than Eclipse
14:59 SummerWine: a way too much RAM
14:59 (join) nilyaK
14:59 SummerWine: but it's much simpler than eclipse with gazzilion plug-insQ!
14:59 offby1: that's like saying "my cat weighs less than that elephant
14:59 offby1: "
14:59 samth: SummerWine: by default, drracket loads about a dozen plugins
15:00 SummerWine: Visual studio does it too
15:00 noam: mithos28: huh? what irc client do you use?
15:00 SummerWine: but it's many many times smaller
15:00 mithos28: noam: Colloquy
15:00 untrusted: visual studio is just crap
15:01 samth: SummerWine: we, of course, do care about the memory consumption of racket and drracket
15:01 noam: i use xchat and it uses 41mb
15:01 SummerWine: every new release of DrRacket takes more ande more memory
15:01 samth: however, we also care about lots of other things
15:01 samth: such as language expressiveness, type systems, contracts, documentation, extensibility, usability, portability, ...
15:01 mithos28: noam: About 3/4 of my ram is currently unused
15:02 samth: so we spend our time working on all of these issues
15:02 nilyaK uses Pidgin
15:02 mithos28: and that is how it normally stands, so if it ever becomes an issue I might look at a smaller one
15:02 noam: mithos28: welp good for you :P
15:02 SummerWine: the verbs people usualy link to DrRacket (from user groupS) are: sluggish, slow, unresponsive
15:02 SummerWine: but things staxy the same
15:02 mithos28: SummerWine: You do know it is an opensource project right?
15:02 SummerWine: stay
15:03 nilyaK: mythos28 meant: if you would like to submit a fix, you may.
15:03 SummerWine: if it is opensource, it doesn't have to be crapware, right?
15:03 mithos28: SummerWine: It works fine for most of its current users
15:04 SummerWine: hahaha, good joke
15:04 bremner: nilyaK: that's a little cliché ;)
15:04 nilyaK: no, this is a little cliché
15:04 (nick) nilyaK -> SummerWhine
15:05 Kaylin is not having a good morning. -.-
15:05 mithos28: Any one have a rough ballpark of how many racket devs there are, and what percentage of their time they devout to racket work?
15:06 SummerWine: i don't know how many they are, but they are doing the poor job. DrRacket gui is really bad
15:06 samth: mithos28: it's hard to make that count precise
15:06 samth: and it varies pretty widely among people
15:06 SummerWine: i remember once someone criticised gui on your group
15:06 SummerWine: and you were totally unwilling to lisen what that gay said
15:07 mithos28: SummerWine: link?
15:07 mithos28: SummerWine: Also most guis are not cross platform
15:07 SummerWine: something obout "Check syntax" button and how that button change gui appearance in some unlogicall way
15:08 samth: mithos28: there are maybe 15-20 people w/ commit access
15:08 SummerWine: i cant find the link right now
15:08 SummerWine: but i remembered
15:08 samth: plus a few more people who write non-trivial amount of code but aren't committers (like you)
15:09 samth: some of those people think of racket development as their main job (like matthew)
15:09 samth: but matthew does also teach, do research, write papers, advise students, that sort of thing
15:09 SummerWine: maybe will be better to hire Steven Sinofsky
15:09 samth: most of the people devote much less of their time to racket development than matthew
15:09 SummerWine: :)
15:10 mithos28: https://groups.google.com/forum/?fromgroups=#!searchin/racket-users/check$20syntax/racket-users/E4xmiGUVgPU/fS6C7Lxmbf8J
15:10 rudybot: http://tinyurl.com/azzlmqq
15:10 mithos28: is that the person you were talking about
15:10 mithos28: * killer bad: clicking on Check Syntax
15:10 mithos28: * changes the ui drastically and i have no idea how to get
15:10 mithos28: back to how it was before i clicked it
15:10 SummerWine: i think yes
15:12 mithos28: You did read the rest of the response where Eli asked him what he meant by that and didn't get an actionable answer back
15:12 mithos28: Also check syntax doesn't change the ui very much at all
15:14 SummerWine: but repl disappear when you click check syntax and it's just annoying
15:14 SummerWine: because of that i never click that button
15:15 mithos28: Ok, thats actually actionable. If you file a bug or send an email to dev@racket-lang you will actually get a discussion on why that decision was made to begin with and if it is worth changing
15:15 SummerWine: and one more thing: when i turn on the line numbering in editor, editor becomes much more sluggish
15:15 mithos28: there are issues with the editor speed, danny is working on reimplemting some of that code right now
15:16 SummerWine: Oh it's just danny, not Sinofsky :)
15:16 mithos28: Well I'm not sure about you, but I don't have the money to pay his salary
15:17 SummerWine: when danny solves the problem, DrRacket will take 1Gb of RAM, i suppose :)
15:17 mithos28: If you can point to actual items which are slow and hurt your workflow then people will listen
15:18 SummerWine: when i use Typed racket, all is incredibly slow
15:18 (join) mye_
15:18 (nick) mye_ -> mye
15:18 mithos28: SummerWine: Compilation or running?
15:18 SummerWine: slower then "regular" racket
15:18 SummerWine: i edit the code and then it's slow when click "Run"
15:19 mithos28: Yes, sam and I work on improving that
15:19 SummerWine: i have to wait too much to program "begin"
15:19 mithos28: but its a hard problem
15:19 SummerWine: maybe that problem is easy for Mark Tarver
15:19 SummerWine: :)
15:20 mithos28: does he have time that he is willing to donate?
15:20 SummerWine: Shen is astonishly fast! (and typed too)
15:20 SummerWine: you should ask him
15:20 mithos28: I don't know him
15:21 SummerWine: i have donated a certain amout to him
15:21 SummerWine: amount
15:21 mithos28: If you can point to TR programs that are extremely slow to compile, we can take a look at them
15:22 mithos28: But if you just say that TR is slow, then I have no idea if you are seeing a real issue or just the fact that TR is doing a lot of work and you would like it to be faster
15:22 SummerWine: i don't have such a program right now, but that happened when i have doing my homework for the Krishnamurthi's online class
15:23 SummerWine: look TR is not slow when running, it's slow to compile, i assume
15:24 noam: back to the memory use issue: i would feel better if i knew what the memory was actually used for, and which parts of the program allocate what amount of memory
15:25 mithos28: noam: do you know how other languages solve this issue? I have never had to debug such problems
15:25 SummerWine: noam: the picture that ignites this polemic is made by merely starting DrRacket and doing nothing else at all
15:26 mithos28: SummerWine: Do you have any ram pressure on your machine?
15:26 SummerWine: no, it's a laptop with 4gb of RAM and nothing else running except the web browser
15:26 mithos28: If not why would you expect drracket to use less ram?
15:26 SummerWine: because Visual studio does
15:27 SummerWine: :)
15:27 SummerWine: and you guys are beter then M$
15:27 SummerWine: are you?
15:27 SummerWine: :)
15:27 noam: mithos28: not the i have used any tools like that but haskell seems to have something: http://www.haskell.org/ghc/docs/7.0.1/html/users_guide/prof-heap.html
15:34 mithos28: noam: That looks cool, I don't think racket has anything of the sort
15:35 samth: mithos28: we have the infrastructre to do some things like that
15:35 samth: but nothing built yet
15:36 noam: i think that should get a place in the TODO list (wherever that is :P)
15:37 samth: noam: i was talking about this a few days a go, in fact
15:38 noam: good to know
15:38 (join) bitonic
15:38 noam: is there a plan?
15:39 SummerWine: a man, a plan, a RAM! :)
15:41 samth: noam: no, not at the moment
15:42 noam: alright. i can't work on this myself but good luck to whoever does (hopefully).
15:45 samth: basically, you'd want to use a combination of periodic calls to `dump-memory-stats` and `plot`
15:46 mithos28: oh wow, didn't know that existed
15:48 samth: it would be nicer to have a custodian-limited version of that
15:48 samth: but i don't know how hard that would be
15:50 (quit) untrusted: Remote host closed the connection
15:50 (quit) Blkt: Remote host closed the connection
15:53 (quit) jrslepak: Quit: What happened to Systems A through E?
15:57 aezx: noam, I'm way late
15:57 aezx: i've been doing other things
15:57 aezx: but not every small program needs 500mb
15:58 noam: no problem
15:58 aezx: i'm perfectly satisfied with drracket using that much
15:58 (join) jrslepak
15:58 aezx: and i could not care any less about an opposing opinion regarding my programming setup
15:58 aezx: it is quite easy to use racket without drracket
15:58 aezx: a great deal of text editors have paren matching and syntax highlighting
15:58 aezx: drracket is not essential
15:59 noam: what about the memory usage of racket programs is general?
15:59 aezx: im not sure what you're trying to ask
15:59 noam: is writing lightweight programs is possible using racket?
16:00 aezx: of course
16:00 mithos28: noam: How lightweight is lightweight
16:00 mithos28: embedded devices?
16:00 (part) pavelpenev: "http://quassel-irc.org - Chat comfortably. Anywhere."
16:01 samth: we've successfully run a racket-implemented version of ssh on a consumer-grade router with 256 MB of ram
16:01 noam: no, just some stuff that runs at the background for example like a torrent client
16:01 samth: but that's probably too small
16:01 (join) dsp
16:01 aezx: the other major flaw in the argument is that I have yet to see drracket use more than 200mb without me asking it to
16:01 mithos28: oh wow, last I heard embedded racket was a no-go
16:01 aezx: I'm using it right now, and its using about 185
16:02 bremner: aezx: on what architecture?
16:02 noam: when i start drracket it uses more then 400mb, aezx
16:02 aezx: crunchban linux x86
16:02 aezx: crunchbang*
16:02 bremner: aezx: so, 32bit?
16:02 noam: so it seems
16:02 aezx: thats what x86 generally means......
16:02 noam: i'm on 64bit
16:02 aezx: noam are you on windows?
16:02 noam: also
16:03 noam: but i have a linux box too
16:03 bremner: aezx: I don't think that tone is called for, but whatever.
16:03 aezx: does drracket behave differently on the two machines?
16:03 mithos28: aezx: x86 also means not arm sometimes
16:03 noam: i don't remember.. let me check that
16:03 aezx: I've never experienced that myself, but I'll take your word for it
16:04 aezx: the point im trying to establish is that drracket behaves differently on different operating systems and archs
16:04 samth: mithos28: "embedded" is a slippery term
16:04 samth: that machine is really a linux machine
16:04 aezx: ^
16:05 bremner: aezx: yes, it could really be that a 32bit version uses less memory
16:05 mithos28: samth: Last I had heard of putting racket on a machine with limited resources, things didn't work
16:05 samth: mithos28: that was probably the same machine
16:05 samth: we eventually got things to work
16:05 samth: but not work that well
16:05 (join) dnolen
16:06 aezx: i dont have a 64bit machine to test that theory
16:06 aezx: but to some degree, I'd expect it to be that way
16:06 mithos28: given a snip% how do I get it to display in a frame%?
16:07 asumu: mithos28: you need to put it in an editor<%>, in an editor-canvas%, in a frame%
16:07 mithos28: asumu: Thanks
16:07 aezx: i love all of you
16:07 (quit) noam: Read error: Connection reset by peer
16:08 (join) noam
16:08 SummerWine: aezx: it's not a theory, it's a reality: 64bit DrRacket on WIndows 7 takes nearly 500mb
16:09 SummerWine: without any program loaded in it!
16:09 SummerWine: that's incredible thing!
16:10 noam: aezx: on linux 64bit drracket uses 305MB and goes to 339MB after opening two very small files
16:10 mithos28: noam: How are you measuring that?
16:10 noam: 305mb is better i'd give you that.
16:11 noam: just kde's system monitor thingie
16:11 SummerWine: on the same time, Visual studio takes 10 times less
16:11 asumu: SummerWine: you'
16:12 asumu: Sorry, pressed enter too early.
16:12 asumu: SummerWine: you've made your point. Please keep the discussion constructive. :)
16:12 SummerWine: ok, sorry
16:15 asumu: FWIW, I see around 400-450MB on 64-bit Win7/Linux.
16:15 (quit) bitonic: Ping timeout: 240 seconds
16:15 SummerWine: and that doesn't make you worried?
16:16 mye: using seconds->date with something after the end of the unix epoch it only possible with 64-bit racket. But is there a workaround?
16:16 SummerWine: come on, it's a WAY too much!
16:16 mye: I was quite annoyed that racket didn't abstract this. Is there reason? Performance?
16:17 mithos28: mye: abstract what?
16:17 asumu: mye: IIRC, some of the time/date stuff is implemented in the runtime.
16:17 asumu: So it's likely an implementation issue.
16:17 mye: mithos28: allow dates after 2038 with a 32-bit build
16:17 mithos28: oh, I missed your first message.
16:18 aezx: summerwine
16:19 aezx: i can tell that you and i arent going to get along
16:19 aezx: very easily
16:19 mithos28: mye: because seconds->date is meant to be platform specific it looks like
16:19 asumu: aezx, SummerWine : please keep discussion on-topic. No feuds here. :)
16:20 mithos28: and 32bit linux cannot do past that
16:20 noam: aezx: i think i can say that too about summerwine if that is of any consolation :P
16:20 mithos28: mye: Also what are you trying to do that you need to go out that far?
16:21 mithos28: mye: especially since you don't know what leap seconds will look like or if someone will change DST on you
16:21 SummerWine: asumu: Is it true that the Japanese have a small dick???
16:21 mye: mithos28: I can accept that. But is there way to do it another way to handle dates? Is it worth trying? Probably not. But I just thought it would be .. nice
16:21 mye: ouch?
16:21 mye: what are the banning policies here?
16:22 offby1: yay, thanks
16:22 mithos28: mye: what is your goal?
16:23 samth: mye: the ban policies are (a) please keep it constructive and (b) inappropriate behavior like that is an immediate ban
16:23 (join) wlangstroth
16:26 mye: mithos28: I'm writing this url shortener that encodes dates in base 60, and was getting error with certain dates in the tests. So my goal after I found out that 32-bit can't handle it was to find out how to do the calculations in a way that isn't dependent on the runtime.
16:26 offby1: What do dates have to do with shortening URLs?
16:27 mithos28: why are you turning dates into seconds since the epoch? Like I pointed out you will have issues with leap seconds.
16:28 mye: offby1: every url has the form <base60><base60><base60><base60> which encodes the days since the epoch
16:28 mye: the last is a growing ID
16:28 offby1: oh I see
16:28 mithos28: What am I not doing here? I just get an empty frame. https://gist.github.com/4107614
16:28 offby1: I assumed your URL would be more or less a hash of the input URL; that's how the real ones work, I think
16:28 mye: I got the idea here: http://tantek.pbworks.com/w/page/21743973/Whistle
16:29 mye: the trick is, its algorithmically reversible
16:31 asumu: mithos28: it works if you define the canvas after the editor (and use the 'editor' init-arg for the canvas)
16:32 asumu: (might also work if you set-editor instead)
16:32 asumu: (yes, it does)
16:32 mithos28: thanks, that worked
16:36 (join) jao
16:36 (quit) jao: Changing host
16:36 (join) jao
16:39 (join) greghendershott
16:40 (quit) greghendershott: Client Quit
16:41 (quit) wlangstroth: Quit: wlangstroth
16:41 (join) bjz
16:41 (quit) bjz: Client Quit
16:41 mye: mithos28: re: going out that far: I probably got an overflow of unwarranted perfectionism (pun intended).
16:42 mye: It just bothered me that I couldn't write test cases with certain dates.
16:42 mithos28: I would submit a bug if I were in your position
16:43 mye: I mean, it's not like time based bugs didn't prevent me from using drracket one day  :P
16:44 mithos28: mye: Hush, that never happened.
16:54 samth: mye: for more time fun, there was a bug where seconds->date errord on some values b/c they don't exist in that time zone
16:55 asumu: samth: wasn't that a "not a bug"? :p
16:55 samth: yes
16:55 asumu: Or maybe "it's like a bug".
16:57 samth: the bug is that people have created a totally insane time systems
16:58 mithos28: https://gist.github.com/4107614, realtime plot of racket's memory usage
16:59 mye: is it a good idea then to, in general, generate date structs that use UTC only? Does it make the code more reliable?
17:01 mye: I get a shark tooth pattern, mithos28. Looks cool
17:02 mithos28: yeah thats what I get
17:02 mithos28: the problem is that is all from the chart generation
17:02 asumu: samth: this is why we should switch to kiloseconds and the International Fixed Calendar.
17:03 mithos28: asumu: first lets get rid of leap seconds
17:03 mithos28: start with the easy things
17:03 asumu has a kilosecond clock on some of his taskbars... it's not actually very intuitive.
17:04 mithos28: what is a kilosecond?
17:04 mithos28: just 1000 seconds
17:04 mithos28: or is it like swatch's beats
17:04 asumu: 1000 seconds. 86.4 of them is a day.
17:04 asumu: 0.5ks is about an hour.
17:05 asumu: Wait, not 4ks is about an hour.
17:05 samth: mithos28: i think a moving avg would improve the chart a bunch
17:05 samth: but it's really awesome
17:05 mithos28: samth: yeah it was just a proof of concept
17:06 mithos28: most of the time was figuring out how to do guis in racket
17:06 mithos28: I couldn't figure out how to collect the information from dump-memory-stats
17:08 (quit) masm1: Ping timeout: 246 seconds
17:09 (join) masm
17:09 samth: dump-memory-stats prints to stderr
17:09 samth: the low-level process stderr, too, i think
17:09 samth: not `current-error-port`
17:09 mithos28: right, thats what it seemed like
17:10 mithos28: probably due to issues of allocation during memory collection
17:10 mithos28: but it seems like it could put the information in some static place, and then allocate after the collection
17:14 samth: yeah
17:14 samth: but matthew would probably have more info on this
17:27 (join) ambrosebs
17:27 (join) bitonic
17:30 asumu: Anyone else see this on the latest git? raco setup:   UNKNOWN: : read (compiled): ill-formed code [../../../racket/gc2/../src/validate.c:1167]
17:30 asumu might just have some stale state
17:31 samth: ambrosebs: did you have slides for your clojure conj talk?
17:33 ambrosebs: samth: I'll send them to you, but I managed to get roughly 1/3 of REPL demos in
17:34 samth: 1/3 is a pretty good ratio for a demo
17:34 ambrosebs: samth: typing invocations of `map` got a round of applause.
17:34 samth: demos never cover as much as you think they will
17:34 samth: ambrosebs: awesome
17:35 ambrosebs: samth: I attributed Strickland, TH, Felleisen off the top of my head.. correct?
17:35 samth: yes, that's right
17:35 ambrosebs: great
17:35 ambrosebs: I also did a small session on a whiteboard covering occurrence typing and dotted tvars
17:36 ambrosebs: and I showed off my `filter` extensions
17:36 ambrosebs: in detail
17:40 (quit) jonrafkind: Ping timeout: 245 seconds
17:42 (quit) soegaard: Quit: soegaard
17:45 (join) Kaylin
17:46 samth: ambrosebs: sounds great
17:46 samth: wish i could have been there
17:54 (quit) mithos28: Quit: mithos28
17:58 (join) mithos28
18:07 ambrosebs: if anyone else is interested: Typed Clojure @ Clojure Conj 2012 slides https://github.com/downloads/frenchy64/papers/conj%20main%20talk.pdf
18:08 (quit) bitonic: Ping timeout: 240 seconds
18:10 (quit) MayDaniel: Read error: Connection reset by peer
18:11 mithos28: ambrosebs: Why the need for annotation in type aliases?
18:12 ambrosebs: methos28: Aliases are handy? What are you referring to exactly?
18:13 mithos28: the variance annotations
18:13 ambrosebs: methos28: ah, it lets me abstract over monads
18:14 ambrosebs: (I'm pretty sure) monads are covariant in their argument
18:14 mithos28: TR doesn't need them, so I'm wondering why TC needs the annotations
18:15 ambrosebs: https://github.com/frenchy64/typed-clojure/blob/master/test/typed/test/monads.clj#L481
18:15 rudybot: http://tinyurl.com/bfa9n72
18:15 ambrosebs: that's an example where they're handy
18:16 ambrosebs: TR polymorphic aliases are a bit of a hack, just wrap your type in a `All`
18:16 ambrosebs: the variance annotations take it a bit further so they're useful for monads
18:17 ambrosebs: of course, I could have a default variance, it's pretty annoying to have to explicitly say every time
18:17 mithos28: I don't see why the system cannot infer the variance
18:17 mithos28: rudybot: init typed/racket
18:18 rudybot: mithos28: your typed/racket sandbox is ready
18:18 ambrosebs: mithos28: that's probably true.
18:19 mithos28: So is the answer you hadn't looked into not requiring the annotations?
18:19 mithos28: I just want to make sure there isn't something missing from TR
18:20 ambrosebs: mithos28: Well what's missing with TR is higher-kinded types.
18:20 mithos28: ah, yeah I just realized thats what you have
18:20 ambrosebs: mithos28: I can probably infer the annotations, yes.
18:21 mithos28: right TR just has first order type functions
18:21 mithos28: which cover a lot of ground
18:21 mithos28: I like defining my types as the fix point of polymorphic structs
18:22 mithos28: thus I can reuse structure code across different tree types
18:22 mithos28: useful in code transformations
18:22 ambrosebs: what is the fix point of polymorphic structs?
18:23 mithos28: (struct: (e) app ((fn : e) (arg : e)))
18:23 mithos28: rudybot: (struct: (e) app ((fn : e) (arg : e)))
18:23 rudybot: mithos28: error: #<undefined>:1:13: dtsi*: expected struct name at: app in: (dtsi* (e) app ((fn : e) (arg : e)) #:maker make-app)
18:24 mithos28: oh right repl sucks
18:25 (quit) SamB_MacG5: Ping timeout: 260 seconds
18:26 (join) bitonic
18:26 (join) SamB_MacG5
18:27 (quit) ambrosebs: Ping timeout: 246 seconds
18:29 mithos28: https://gist.github.com/4108103
18:29 mithos28: oh well he left
18:36 (join) adamgolding
18:44 (join) anRch
18:48 (quit) Shviller: Ping timeout: 244 seconds
18:48 (join) Shviller
18:56 (quit) mithos28: Quit: mithos28
19:00 (join) mithos28
19:08 (quit) hash_table: Ping timeout: 276 seconds
19:15 (quit) anRch: Quit: anRch
19:53 (quit) bitonic: Ping timeout: 240 seconds
19:55 (quit) masm: Quit: Leaving.
19:57 (join) jonrafkind
20:04 (quit) tsion: Quit: Leaving
20:05 (quit) adamgolding: Quit: Page closed
20:07 (join) dedis21
20:13 (join) yoklov
20:15 (quit) dedis21: Remote host closed the connection
20:24 (join) ambrosebs
20:24 ambrosebs: mithos28: sorry about that, airplane wifi ran out
20:25 ambrosebs: so what did you mean by "fix point of polymorphic structs"?
20:36 mithos28: https://gist.github.com/4108103
20:43 (quit) dnolen: Quit: ERC Version 5.3 (IRC client for Emacs)
20:46 mithos28: ambrosebs: Just put some more examples in that gist
20:49 (join) cipher
20:49 ambrosebs: mithos28: Pretty cool idea. Is it related to the higher-kinded types?
20:50 mithos28: sorta, app and lam are both higher order types
20:50 ambrosebs: Typed Clojure also supports recursive types.
20:50 mithos28: sorry higher-kinded types
20:50 mithos28: but need to be used in first order settings
20:51 (join) greghendershott
20:52 ambrosebs: mithos28: I think I need clarification: LC and LC-with-int just look like normal types to me.
20:52 mithos28: app has kind * * -> *
20:52 mithos28: sorry * -> *
20:53 mithos28: and so does lam
20:53 ambrosebs: Right, yes I see.
20:53 mithos28: but you don't have lambda at the type level
20:53 ambrosebs: yes
20:54 mithos28: and all arguments to All must be of kind *
20:55 ambrosebs: mithos28: it gets kind of weird because I'm pretty sure you can pass a `All` to another `All` anyway in TR
20:55 ambrosebs: but definitely not a good idea I guess.
20:55 mithos28: example?
20:56 mithos28: thats not a contradiction
20:56 mithos28: All is different from lambda at the type level
20:57 ambrosebs: I see. They seem to play the same role in TR
20:57 ambrosebs: Whats the diff?
20:57 mithos28: a value can have the type (All (a) (Listof a))
20:57 mithos28: for example null
20:58 mithos28: so all (All (a) (Listof a)) has kind *
20:58 mithos28: while (lambda (a) (Listof a)) has kind * -> *
20:59 ambrosebs: ah! very important difference, I see.
20:59 mithos28: Usually All is written with capital lambda
20:59 mithos28: no thats not right
21:00 ambrosebs: (V (x) ...) right?
21:00 ambrosebs: forall
21:00 mithos28: yeah, capital lambda is the value with that type
21:01 ambrosebs: mithos28: I see
21:01 ambrosebs: Seems like a bit of an issue to have contract generation + higher-kinded types?
21:02 mithos28: why?
21:02 ambrosebs: I guess you always assign an expression a type *
21:02 ambrosebs: it doesn't make sense to give it any other kind.
21:02 ambrosebs: So I guess it's ok
21:02 mithos28: right all expressions have a type who's kind is *
21:03 ambrosebs: I thought about your variance question. I actually already do infer the variance, so it's completely redundant :)
21:04 ambrosebs: the interesting bit is when you want a type variable of a higher-kind
21:04 ambrosebs: and that always needs extra annotation
21:04 mithos28: that makes sense
21:04 ambrosebs: It's really handy in monad transformers
21:05 ambrosebs: https://github.com/frenchy64/typed-clojure/blob/master/test/typed/test/monads.clj#L826
21:05 rudybot: http://tinyurl.com/9w83rff
21:05 (join) tsion
21:05 ambrosebs: A bit hard to read, but basically `m` is the monad, and it has a covariant argument
21:05 (join) mizu_no_oto
21:06 ambrosebs: and :kind gives the kind of the type variable
21:06 ambrosebs: actually, does it make sense to have `m` as an argument to `All`?
21:07 ambrosebs: Seeing as m is * -> *
21:07 ambrosebs: Of course, you need to instantiate `m` every time you use maybe-t
21:07 mithos28: Yes it makes sense, TR doesn't support it but thats an implementation defect
21:08 ambrosebs: I wonder if there's a good use case for Racket.
21:08 (join) bbloom
21:08 mithos28: I have a question about you maybe-m
21:08 mithos28: your*
21:09 mithos28: how do you differentiate Just Nothing and Nothing
21:09 mithos28: syntax taken from haskell
21:09 ambrosebs: that's a defect of that particular monad implementation
21:09 ambrosebs: you can't
21:09 mithos28: pretty big defect
21:09 ambrosebs: yep. :)
21:10 ambrosebs: other monad impls use a unique keyword
21:10 ambrosebs: but it's uninteresting from a static type perspective
21:11 (quit) dzhus: Ping timeout: 248 seconds
21:12 ambrosebs: Many people ask that question, it amazes me how quickly you see the defect.
21:12 ambrosebs: I don't use monads :)
21:12 mithos28: I have a knack for seeing the edge cases
21:12 ambrosebs: haha
21:12 ambrosebs: nice
21:13 mithos28: It was fun looking at the TR code base and finding out how to subvert the typesystem
21:13 ambrosebs: type functions in TR would be lots of fun I imagine.
21:13 ambrosebs: needs to be implemented
21:13 ambrosebs: :)
21:13 mithos28: they go on the feature request list
21:14 mithos28: which should get looked at after the bug list
21:14 asumu: ambrosebs: I looked at your slides, very nice work!
21:14 ambrosebs: asumu: thanks!
21:14 (join) amalloy
21:15 asumu: ambrosebs: was there anything tricky you needed in particular to support protocols?
21:15 ambrosebs: asumu: well the tricky case is when protocols are extended at runtime. And I have no idea how to handle that except add a "trust me" annotation
21:15 ambrosebs: asumu: but outside that, no issue.
21:16 asumu: Ah, I see. This is the whole `extend-protocol` thing Clojure has?
21:16 asumu: Yeah, that's interesting. Does that see a lot of use in Clojure code?
21:17 ambrosebs: depends on the code I guess. I personally haven't seen it that often
21:18 (join) wlangstroth
21:18 ambrosebs: asumu: it seems like where runtime contracts could help
21:18 ambrosebs: right now, TC doesn't change runtime semantics at all.
21:20 mithos28: does anyone know if whalesong is close to being able to do all of racket/base yet?
21:36 offby1: I bet danny knows
21:36 offby1: rudybot: seen dyoo
21:36 rudybot: *offby1: dyoo was seen joining in #racket one day ago, and then dyoo was seen quitting in c-98-202-217-133.hsd1.ut.comcast.net one day ago, saying "Quit: dyoo"
21:40 (join) racketeer
21:42 racketeer: Hello, I'm currently working through HtDP and I'm trying to use the make-pson function. When I try and use it, I get an error: "make-pson: this function is not defined"
21:42 mithos28: are you spelling it make-pson?
21:42 mithos28: it should be make-posn
21:42 mithos28: you switched the s and the o
21:43 (join) neilv
21:43 racketeer: Gah! I just looked again. You're right.
21:43 mithos28: posn is short for position (and the o corresponds to the first o)
21:43 (quit) Kaylin: Read error: Connection reset by peer
21:43 racketeer: Thanks!
21:43 (join) shriphani__
21:43 (quit) shriphani_: Ping timeout: 240 seconds
21:44 jonrafkind: (match blah [(list x ...) (define x ...)]) I keep doing this argh! x ends up being #<undefined>
21:44 mithos28: racketeer: no problem. If you have any more questions, just ask away
21:44 mithos28: jonrafkind: just never reuse variable names
21:44 mithos28: ever
21:45 jonrafkind: good idea, then we won't need hygiene
21:46 mithos28: jonrafkind: no because if a macro is invoked twice it would put the same 'unique' variable in twice
21:46 jonrafkind: wait so you were serious about that suggestion?
21:46 mithos28: Its what I'm currently doing
21:46 jonrafkind: are you like, a crazy person
21:47 mithos28: I may be writing a compiler though, so doing it programatically
21:47 jonrafkind: ryan has on occasion said to me that it might be a good idea to disallow shadowing in general
21:47 jonrafkind: i still think its a crazy idea
21:48 jonrafkind: or to compromise that you can't shadow macro bindings
21:48 mithos28: why only macro bindings?
21:49 jonrafkind: well i forget the reasoning
21:51 mithos28: rudybot: (:print-type filter)
21:51 rudybot: mithos28: ; stdout: "(All (a b) (case-> ((a -> Boolean : ((b @ 0) | Top)) (Listof a) -> (Listof b)) ((a -> Any) (Listof a) -> (Listof a))))\n"
21:51 (quit) racketeer: Quit: Page closed
21:51 mithos28: why does filters first argument need to return a Boolean in the first case?
21:52 asumu: jonrafkind: that annoys me too about define.
21:56 (join) sw2wolf
21:58 sw2wolf: It is VERY hard for me to find an app. developed using racket. https://github.com/Metaxal/RWind/  But it is so experimental ...
21:58 mithos28: I want to use filter for a list of type (Listof (Option <complicated>)), <complicated> doesn't have simple predicate
22:00 asumu: sw2wolf: http://planet.racket-lang.org/
22:00 (join) hash_table
22:01 sw2wolf: <asumu> thanks
22:01 mithos28: sw2wolf: http://en.wikipedia.org/wiki/Uncharted_3:_Drake%27s_Deception
22:01 mithos28: sw2wolf: http://cufp.org/conference/sessions/2011/functional-mzscheme-dsls-game-development
22:01 rudybot: http://tinyurl.com/6ac6sxp
22:02 jonrafkind: sw2wolf, i wrote pong and asteroids in racket
22:04 sw2wolf: I am a racket newbie. I have CL experience by using stumpwm and i want to learn racket
22:05 sw2wolf: It seems racket is so big compared with guile and chicken, isnot it?
22:05 jonrafkind: ya
22:06 (quit) wlangstroth: Quit: wlangstroth
22:06 neilv: racket has a lot of stuff, and it's well-documented. start reading. http://docs.racket-lang.org/getting-started/
22:06 sw2wolf: ok
22:08 sw2wolf: It takes so long to install racket on my freebsd box as freebsd 9 doesnot supply rqacket binary package.
22:09 offby1: sw2wolf: you can save some time by passing an argument like --without-documentation to "configure"
22:10 neilv: some freebsd person could step up and make racket binary packages
22:10 sw2wolf: i hope so
22:10 sw2wolf: Both guile and chicken has binary package
22:10 neilv: i did know someone using racket on freebsd, but they switched to linux
22:11 sw2wolf: freebsd and stumpwm is my every day desktop
22:11 sw2wolf: now the racket can be installed from freebsd ports.
22:14 (join) cdidd
22:15 (quit) hash_table: Ping timeout: 252 seconds
22:18 (quit) walter|r: Quit: This computer has gone to sleep
22:25 (nick) sethalve_ -> sethalves
22:26 (quit) nathanpc: Quit: Computer has gone to sleep.
22:42 (quit) ambrosebs: Ping timeout: 245 seconds
22:43 (quit) mizu_no_oto: Quit: Computer has gone to sleep.
22:45 (part) amalloy
23:21 jonrafkind: I find I still need let loop a lot
23:28 (join) Kaylin
23:31 neilv: there's a reason they came up with something as funny-looking as named-let for rnrs: it's actually a good idea
23:35 (join) Nisstyre-laptop
23:37 jonrafkind: yea but in light of the for* stuff it seems like an anachronism to use named-let
23:38 mithos28: for* only works where the recursion scheme is known before looking at the data
23:38 neilv: it's not. lots of things don't fit the for-things
23:39 neilv: the for-things are nice conveniences that expose certain structures
23:39 neilv: i'm a little concerned that people will see the for-things and never get good at doing it a non-for-thing way
23:39 jonrafkind: what like map and foreach?
23:39 (quit) greghendershott: Quit: Leaving.
23:40 neilv: so when they have a problem that doesn't fit the for-things, they are stumped, or write pascal in racket
23:40 jonrafkind: id much rather write pascal than named-let actually
23:40 mithos28: jonrafkind: what kind of loops are you writing?
23:41 jonrafkind: its like parse 3 expressions from a list, then return those 3 expressions and the unparsed list
23:41 jonrafkind: and the 3 expressions might be separated by commas or whatever
23:41 mithos28: and an expression can be more than one element?
23:41 jonrafkind: yea it could be 1 + 1, 2 + 2
23:41 jonrafkind: the parse function returns a parsed thing and an unparsed list
23:42 neilv: hm. there isn't a scribble def-something or convention for the new define-logger thing in 5.3.1, is there
23:42 mithos28: this seems like it should be easy in monadic form
23:42 jonrafkind: lulz monads..
23:43 jonrafkind: etc etc if you use monads now you have 2 problems
23:43 mithos28: runParser (do x <- parseExpression; y <- parseExpression; z <-parseExpression; return (x,y,z)) inputList
23:44 mithos28: its just the state monad
23:44 jonrafkind: ok now parse expressions separated by commas until you hit a 'X delimiter
23:46 mithos28: def parseUntil pred f = if pred then return [] else do x <- f; fmap (cons x) (parseUntil pred f)
23:47 mithos28: parseUntil (isNext 'x) parseExpr
23:47 asumu will often write a let loop as a first cut and then translate to `for` if there's a good match
23:47 neilv: asumu is smart
23:49 jonrafkind: mithos28, oh ok it works because the current parse state is inside the monad eh
23:49 mithos28: yeah
23:49 mithos28: thats the whole monad
23:49 mithos28: just a state monad over the parse state
23:49 jonrafkind: ok i guess its fine, it doesnt seem much better than just using mutation, but i guess monads scale better
23:51 mithos28: right. I'm not sure what the structure of Honu is but it seems like the concept of a parser is likely core
23:51 mithos28: and you want an easy way to combine them
23:51 jonrafkind: well this is for python actually.. im just using honu-style expansion
23:51 jonrafkind: ill keep it in mind to refactor the honu parser using monads, maybe it iwll be cleaner
23:52 mithos28: you are writing a python parser in racket?
23:52 jonrafkind: right now its a huge mess of (match ...)
23:52 jonrafkind: yea
23:52 jonrafkind: for my thesis
23:52 jonrafkind: as a way to add macros to python
23:52 mithos28: ah cool
23:52 mithos28: too bad I won't get to use it at work anytime soon
23:52 jonrafkind: its pretty easy really, just write some expand thing for the core ast and enforest to deal with the concrete syntax
23:53 jonrafkind: so its just a tad more work than writing a regular parser
23:53 mithos28: enforest?
23:53 jonrafkind: to use it for real id have to write it in python and add in meta-levels and whatnot
23:53 jonrafkind: enforest is the parsing layer that deals with concrete syntax
23:53 jonrafkind: just the name we came up with
23:53 mithos28: ah
23:53 (join) mizu_no_oto
23:57 (join) ewemoa
23:58 (quit) mizu_no_oto: Ping timeout: 240 seconds