00:01 (join) Enoria
00:33 (join) Tyr42
00:39 (join) kvda
00:39 (quit) Fare: Ping timeout: 246 seconds
00:41 (quit) jrslepak: Quit: What happened to Systems A through E?
00:49 (quit) Tyr42: Read error: Connection reset by peer
00:49 (join) Tyr42
00:52 (quit) samth: Ping timeout: 246 seconds
01:05 (quit) Enoria: Ping timeout: 245 seconds
01:09 (join) Enoria
01:15 (quit) em: Ping timeout: 246 seconds
01:22 (quit) cdidd: Ping timeout: 245 seconds
01:35 (join) mrcarrot
01:48 (quit) dnolen: Remote host closed the connection
01:59 (join) dgs_
02:11 (quit) haffe: Ping timeout: 244 seconds
02:17 (quit) yoklov: Quit: computer sleeping
02:36 (quit) jeapostrophe: Ping timeout: 252 seconds
02:50 (join) hkBst
02:50 (quit) hkBst: Changing host
02:50 (join) hkBst
02:53 (join) em
02:53 (quit) Tyr42: Quit: Leaving.
03:12 (quit) mithos28: Quit: mithos28
03:23 (quit) ivan\: Ping timeout: 265 seconds
03:24 (quit) realitygrill: Quit: realitygrill
03:25 (join) kampsun
03:30 (join) ivan\
04:09 (join) Asphodelia
04:25 (join) bas_
04:29 (quit) bas_: Read error: Connection reset by peer
04:29 mrcarrot: is there any good guide for how to make text based programs using racket? if i am going to learn racket, i will most of the time do textbased programs over ssh
04:30 mrcarrot: htdp is pretty much focused on graphics and how to use drracket..
04:47 (join) bas_
04:47 (nick) bas_ -> Skola
05:00 (join) ahinki
05:02 (join) djcb
05:05 (quit) Asphodelia: Ping timeout: 244 seconds
05:25 (join) noelw
05:51 (join) tim-brown
05:51 (join) GeneralMaximus
05:57 (join) veer_
05:58 (join) rostayob
05:58 tim-brown: tewk: any progress with the places thing, can i do anything to help?
06:00 (join) Asphodelia
06:00 (quit) Asphodelia: Changing host
06:00 (join) Asphodelia
06:03 (nick) veer_ -> password
06:04 (nick) password -> veer_
06:10 (quit) Asphodelia: Ping timeout: 252 seconds
06:11 (join) mceier
06:16 (quit) dgs_: Read error: Connection reset by peer
06:24 (join) gciolli
06:28 (join) asdfhjkl
06:56 (quit) GeneralMaximus: Quit: Leaving
07:18 (quit) dous: Ping timeout: 246 seconds
07:29 (join) masm
07:32 (join) acarrico
07:42 (join) samth
08:09 (quit) kanak: Quit: Leaving.
08:17 (quit) Diarmid: Quit: Leaving.
08:20 (quit) rostayob: Quit: WeeChat 0.3.5
08:26 (join) Diarmid
08:26 (join) kanak
08:26 (quit) kanak: Client Quit
08:50 (nick) veer_ -> veer
08:52 mrcarrot: is racket having support for persistent datastructures like acid-state in haskell?
08:53 noelw: There is a data structure library on Planet
08:54 noelw: Oops. Not in Planet
08:55 mrcarrot: does it support persistence by writinh to disk?
08:56 noelw: Oh, I thought you meant persistence in the functional data structure sense.
08:56 noelw: Not as far as I know.
08:56 noelw: Racket has generic serialisation, which you might be able to use.
08:56 noelw: (See the 'data' package in Racket)
08:57 mrcarrot: okay
09:04 Diarmid: I might not be understanding what's being asked, but wouldn't just writing the s-exp's to a file and reading them back out, the way most Lisp programs store data, do what's needed?
09:05 Diarmid: If ACID in the database sense is what's needed, that won't have any notion of 'transactions', though, but I'd guess you'd be using a database for that.
09:07 (join) GeneralMaximus
09:11 (join) yoklov
09:32 (join) Fare
09:38 (join) jeapostrophe
09:44 (join) Tyr42
09:47 (quit) Technodrome: Read error: Connection reset by peer
09:48 (join) Technodrome
09:53 (join) EmmanuelOga
09:53 (join) rostayob
09:57 (quit) rostayob: Client Quit
09:57 (join) rostayob
10:03 (quit) yoklov: Quit: computer sleeping
10:03 mrcarrot: Diarmid: it was something similar to this one: http://acid-state.seize.it/
10:04 (quit) rostayob: Quit: WeeChat 0.3.5
10:05 (quit) GeneralMaximus: Ping timeout: 252 seconds
10:05 Diarmid: mrcarrot: I see, so it is essentially a database, without the database.  I'm curious what the use case is.
10:05 (join) rostayob
10:06 mrcarrot: Diarmid: acid-state takes care to automatically write things to disk, so if your program is crashing, nothing is lost. also, the whole database is all the time in RAM, meaning it is fast and that you can use any data structure for storing your data, a tree, a list, a hashtable... with anything you want
10:07 Diarmid: Couldn't you just use the normal s-exp serialization, wrapped in a function to validate the data (in case you're afraid your program will corrupt its state), and that wrapped in a function to write to disk?
10:07 mrcarrot: the normal case for most program is to use sql databases, meaning that you need to translate the tables and queries into datastructures i your program. this is always an extra burden...
10:07 Diarmid: mrcarrot: I think you'd have to do most of the same work anyway (e.g. explaining what 'valid' means).
10:08 mrcarrot: Diarmid: i have not been writing even one program with racket yet, i am just curious of racket and exploring what it is able to do :)
10:09 rostayob: it's hard to do what acid-state does in an impure language. acid-state works because the change that can be make to every transaction can be serialized easily
10:09 rostayob: and then "replayed"
10:09 rostayob: you can just trust the user anyways
10:09 rostayob: (the user of the library)
10:09 mrcarrot: isn't scheme basically a "pure" language?
10:10 rostayob: mrcarrot: no
10:10 Diarmid: mrcarrot: I see.  Well, Lisp is designed to be able to dump most data structures to a file and read them back, with little to no effort on your part.  It doesn't pretend to be a database, but if your only concern is making sure the preserved state is sane, that's not very difficult to achieve.
10:10 mrcarrot: okay, somebody must have been writing wrong on one site then they said that scheme is the pure version of lisp... and that clisp is impure
10:11 rostayob: mrcarrot: do you know what "pure" mean in this context?
10:11 Diarmid: I/O becomes hellacious in every 'pure' language I've seen.  I'm quite thankful Scheme isn't pure.  It makes it easy to write imperative code to do I/O, and attach that to the functional bits.
10:11 rostayob: Diarmid: you haven't tried Haskell :(
10:11 rostayob: ops, it was meant to be a :) ehe
10:11 Diarmid: rostayob: Actually, Haskell is what I had in mind when I thought of 'hellacious I/O'.
10:12 rostayob: Diarmid: IO is not hellacious in Haskell. but I don't want to start a big discussion about this.
10:12 rostayob: let's say that Haskell IO model makes things like STM and acid-state possible.
10:12 mrcarrot: IO is not that difficult to do in haskell... and even "imperative" code with do notation is quite easy
10:14 Diarmid: The "do" notation gives you sequence without most of the pain of monadic I/O, but I/O interacts poorly with mandatory lazy evaluation (you have to be sure to 'force' the I/O code to be evaluated as it's needed).  It's possible to work around all this, but I'm glad Scheme took a different path.
10:14 bremner: sure, on the surface it is easy. But really understanding it is a challenge for most people.
10:14 mrcarrot: personally i think the static typing (while it is good) makes the coding more hard to do.. much more than IO
10:15 rostayob: Diarmid: yeah lazy IO is bad, and the Haskell community agrees on that now, and great ways of doing IO efficiently in Haskell have surfaced
10:15 Diarmid: I don't mind static typing.  The second language I learned (C; first was BASIC with a smattering of assembly I didn't fully understand) was statically typed, and it never gave me any difficulty.
10:16 rostayob: anyways, without talking about Haskell, mrcarrot: you can definitely do something like acid-state in scheme but without the same guarantees
10:16 mrcarrot: okay
10:16 Diarmid: I just wish type systems did a better job catching real bugs rather than getting in the way.  Real bugs tend to have to do with the /value/ being wrong (read/write array with index out of bounds, divide by zero), rather than the /kind of thing/ (integer, in this case) being wrong.
10:17 rostayob: Diarmid: 1) Haskell catches loads of errors for me 2) you can encode those things at the type level if you really want to, but it's not that pratical ATM :)
10:17 Diarmid: Maybe what I'd really want is something like Qi/Shen, but from what little I've seen of that, it can't express those kinds of things very well.
10:17 (join) GeneralMaximus
10:18 rostayob: for example, if you have 'y / x' you can force the "kind" of x to be /= 0 at the type level with dependent types
10:18 rostayob: those are still types
10:19 Diarmid: rostayob: But how would you express this such that it could be evaluated at compile time?
10:19 rostayob: Diarmid: of course
10:19 rostayob: check out Coq or Agda
10:19 rostayob: Diarmid: IO and non-determinism don't play well in this picture for obvious reason
10:20 rostayob: so you have to have an "entry point" in which you assume that some properties hold
10:20 rostayob: long story short, right now programming with dependent types is quite an ordeal :P
10:20 Diarmid: rostayob: Assuring that a arbitrary function does not return 0 is non-trivial.  I'm aware of Coq and ACL2, and a handfull of less popular things.  Sadly, other than Coq and ACL2, most don't seem to export executable code. They're more designed with mathematicians than programmers in mind.
10:21 rostayob: Diarmid: Agda and Idris are designed with programmers in mind as well
10:21 rostayob: also, ATS
10:21 rostayob: again, I never claimed that I'd program with those. I'm just saying it's possible.
10:21 (quit) Tyr42: Quit: Leaving.
10:21 (quit) asdfhjkl: Read error: Connection reset by peer
10:21 rostayob: Agda in particular has very nice interoperability with Haskell
10:23 Diarmid: rostayob: I'd be happy to find something I could stand to use that made programming relatively painless, and verification at least largely automatic.  I'm not expecting it to tell me whether there's a odd perfect number without me expending any effort, but something that 'automatically' checked things like array indexes an division with interval arithmetic, and recognized I/O as a (imperative and) fact of life would help a awful
10:25 rostayob: Diarmid: Agda and Coq don't help in terms of lemma discovery (they don't prove stuff for you, but they give you a framework to prove stuff in). And even those are a pain for practical programming. So I think that the "largely automatic verification" is quite hard. However verifying little things like out of bounds indices or division by zero is quite OK
10:25 rostayob: still when doing IO you have to postulate things, because you don't know what you're going to get :P
10:26 rostayob: Diarmid: also, Haskell does recognise IO as imperative, and as a fact of life
10:27 Diarmid: rostayob: Mostly agreed.  I don't think I/O is as hard as theorists want to make it.  it bears some resemblance to the belief in the 'impossibility' of including GOTO in Hoare Logic (it was ultimately handled by just 'forgetting' anything except properties known to hold at the 'to' label).
10:28 rostayob: yeah... I'll be waiting for a practical dependently typed language
10:28 Diarmid: rostayob: Handling things programmers actually care about (e.g. that array index thing; about 60% of security holes in native code are that), without a unreasonable amount of errors, is difficult.  For instance, I can't imagine how pointer/reference use could be verified in the general case, and those are something I consider vital.
10:29 Diarmid: I don't write code that depends on odd perfect numbers, or the factors of large prime numbers.  As Dijkstra said, most program proofs are 'uninteresting'.  I like it that way.  I just want something that automates that uninteresting part.
10:30 Diarmid: Until a type system can do that, though, I don't intend on using a language with types except for generating code that runs fast (currently C/C++'s excuse for existing).
10:30 rostayob: I think that that can be done, but it still require a lot more care. Ideally you'd write "safe" modules and then some entry points for the IO in which you assert some properties, as I said before. That might work (or at least it does in my head)
10:31 rostayob: Diarmid: I still prefer types. Not only because they catch errors, but because they force me to think about the high level design of my program, and they provide great documentation
10:32 rostayob: it's very easy for me to browse around an haskell module and see what does what. In Erlang (the other language I work in atm) is much trickier
10:32 Diarmid: rostayob: I'm not sure how it would work.  I *need* data structures.  I don't need a incredible variety of them.  I guess if you gave me a vast number of data structures with 'assumed'/manually-proven correct functions, and pre/post-conditions, it wouldn't be that hard to work with, especially if no recursion or loops were provided other than for/foreach, and it was possible to 'hang them' off interrupts.
10:32 (join) agam
10:34 rostayob: Diarmid: well for example in Agda you get all the IO from Haskell functions
10:34 Diarmid: I mostly end up  (ab)using (nested) lists and arrays and strings heavily.  I sometimes end up making my own types to go into lists to provide some sort of functionality, but I think those are the least I could stand having in a modern language.
10:35 rostayob: lets say that you want a list of a certain length, which in Agda might be represented by a pair, composed by the list and a proof that the list is within those boundaries. You'd simply get the list in Haskell and then have some function that proves/disproves that property
10:36 rostayob: afterwards you'll carry around the list together with the proof, so that you can do operations on that list safely
10:36 rostayob: you could provide some combinators that automate these procedures for common properties of common data structures
10:36 Diarmid: rostayob: Interesting that they didn't just say 'no I/O' like most (semi)automated theorem provers did.  I'm still not sure I could tolerate Haskell.  I tried it.  I really didn't like it, and that's got nothing to do with functional programming.  It had everything to do with the monadic I/O, error messages (Helium apparently fixes that now), and syntax (too much Perl influence for my tastes, but cudos for indentation-based bloc
10:37 rostayob: Diarmid: I'm not sure if Helium is still alive. Once you've understood how the type system works, the error messages are clear
10:37 (join) agam_
10:37 rostayob: you have to take a week to understand the type system
10:38 (quit) agam: Ping timeout: 260 seconds
10:38 rostayob: for what concerns monadic IO, I think it's just habit really... it makes much sense to me after 2 years of haskell
10:38 Diarmid: I'm perfectly okay with 'odd' evaluation methods.  I actually quite like them.  I don't like anything that makes I/O harder than it is in imperative languages (again, 'do' gets you part of the way there), or a type system that isn't capable of mostly-automated full verification that requires more of me than C.
10:38 rostayob: and it allows to separate certain side effects from others
10:40 (join) Tyr42
10:40 Diarmid: rostayob: My take-away from my experience with monadic I/O was it was obtuse enough to make most programmers believe the language was 'pure' (whatever that means in a language with side effects; i.e. a practical one), but it was really just a way of saying what functions must be evaluated first so the side effects appear in the right order, much like (baz (bar (foo))) for sequence, or (bar (foo-x) (foo-y)) for "don't care".  I w
10:41 (quit) Tyr42: Remote host closed the connection
10:43 Diarmid: It's also worth noting that the only bug I've ever experienced that involved I/O was forgetting to close file handles.  I don't feel the need to be 'protected' from I/O, sort of like type (kind of thing) errors vs value errors.  I would very much like to be protected from value errors.
10:44 (quit) rostayob: Ping timeout: 246 seconds
10:45 (quit) Fare: Ping timeout: 246 seconds
10:47 (join) yoklov
10:51 (join) realitygrill
10:51 (join) vkz
10:52 stamourv: mrcarrot: https://github.com/takikawa/tr-pfds
10:52 stamourv: An older version is available on PLaneT.
10:57 (join) rostayob
11:01 rostayob: Diarmid: ops, crap connection. did you get my last messages?
11:02 Diarmid: rostayob: I can't say, since if I didn't, I wouldn't know one was missing.  The last I saw was "and it allows to separate certain side effects from others".
11:03 rostayob: Diarmid: http://pastebin.com/d7patv5e
11:05 Diarmid: rostayob: I saw some of that, and was aware of most of the rest (but haven't read that paper).  I think message passing concurrency and tree dataflow are better models than software transactional memory, but anything is better than shared state.
11:06 (join) dzhus
11:06 rostayob: Diarmid: my point is: you *can't* do STM in a language with unrestricted side effects.
11:06 Diarmid: rostayob: I'm also painfully aware that all I/O is imperative, and 'pure' claims are lies.  The most common one, the one you see in older languages, are "we're passing the state of the world" which is a claim that's trivial to make in any language, including blatantly imperative ones.
11:07 rostayob: Diarmid: that doesn't add much to my statement that Haskell makes it easier to separate different side effects from one another
11:08 Diarmid: rostayob: Perhaps.  Again, I don't personally think STM is that great a thing.  The main 'practical' application I can imagine for a language that forces you to use monadic I/O would be some sort of "optimizing I/O compiler".  I imagine something a little like how current async I/O frameworks plan reads for speed.
11:08 Diarmid: rostayob: So where the "don't care" ordering is present, you'd be free to do it whatever way is fastest.
11:08 rostayob: Diarmid: also, the fact that you don't like STM is not really the point here. The point is that once you go for unrestricted IO you can never go back
11:09 rostayob: with Haskell you can have both the "do whatever you want world" (IO) and the "you can do some things" worlds (STM, ST, acid-state, etc) and the "purity" world
11:10 Diarmid: rostayob: That could be, but I haven't really seen any benefits to the restricted I/O, and do find it exceedingly unpleasant to work with.  Articulating exactly why is hard.  Ignoring surface things like syntax I suppose it's because I feel like it's too "low level".  It's a bit like combing one's hair by moving strands individually with tweezers (though that is an exaggeration).
11:11 rostayob: mh. unrestricted IO definitely has a more "natural" feel to it, in some sense. but I suspect is mostly habit.
11:11 Diarmid: rostayob: I think I could have handled it better in a eager-by-default language.  My addiction to 'generators' shows I don't hate lazy evaluation.  I just hate it when I/O is involved. :P
11:11 (quit) realitygrill: Quit: realitygrill
11:12 (join) dous
11:12 rostayob: I love lazyness too much :P. anyways, gotta go
11:12 rostayob: I like a lot of languages anyways eheh
11:12 Diarmid: I hope you have a good day.  Thanks for chatting.
11:12 rostayob: (apart from C++. and Perl.)
11:12 Diarmid: Agreed on C++/Perl, and lots of languages. :)
11:13 rostayob: yeah :). good day to you too
11:13 rostayob goes to have some lunch
11:16 (quit) EmmanuelOga: Ping timeout: 272 seconds
11:21 (quit) Skola: Quit: Lost terminal
11:23 (join) bas_
11:24 (nick) bas_ -> Skola
11:28 (quit) samth: Ping timeout: 260 seconds
11:32 (quit) jeapostrophe: Ping timeout: 265 seconds
11:42 (join) EmmanuelOga
11:50 (quit) ahinki: Quit: ChatZilla 0.9.88.1 [Firefox 11.0/20120310173008]
11:59 (quit) m4burns: Ping timeout: 252 seconds
11:59 (join) m4burns
12:02 (quit) veer: Remote host closed the connection
12:11 (join) jeapostrophe
12:12 (quit) EmmanuelOga: Read error: Connection reset by peer
12:17 (quit) Shviller: Ping timeout: 255 seconds
12:17 (join) Shviller
12:19 (join) jonrafkind
12:19 (quit) Skola: Ping timeout: 252 seconds
12:28 (quit) Technodrome: Read error: Connection reset by peer
12:28 (join) Technodrome
12:29 (join) EmmanuelOga
12:34 (join) Fare
12:34 (quit) hkBst: Quit: Konversation terminated!
13:09 (quit) vkz: Quit: vkz
13:12 (join) anRch
13:18 (join) Skola
13:22 (join) MayDaniel
13:23 tewk: tim-brown: Its fixed in a patch that is awaiting final code review.  http://www.cs.utah.edu/~tewk/sync_fix.diff
13:23 (nick) noelw -> noelw_away
13:23 tewk: window 2
13:35 (quit) GeneralMaximus: Quit: Leaving
13:35 (join) GeneralMaximus
13:37 (quit) itasyno: Ping timeout: 265 seconds
13:37 (quit) MayDaniel: 
13:43 (join) bmp
13:52 (nick) samth_away -> samth
14:08 (quit) jeapostrophe: Ping timeout: 245 seconds
14:09 (part) gciolli
14:17 (quit) anRch: Quit: anRch
14:17 (join) vkz
14:17 (quit) yoklov: Quit: computer sleeping
14:22 (nick) GeneralMaximus -> ankur
14:22 (nick) ankur -> ankursethi
14:27 (join) yoklov
14:27 (quit) jonrafkind: Ping timeout: 252 seconds
14:27 (quit) aidy: Ping timeout: 248 seconds
14:28 (nick) ankursethi -> Ankur
14:28 (nick) Ankur -> GeneralMaximus
14:33 (nick) GeneralMaximus -> ankursethi
14:41 (join) jeapostrophe
14:41 (quit) ankursethi: Quit: Leaving
14:44 (quit) yoklov: Quit: computer sleeping
14:53 (join) GeneralMaximus
14:58 (quit) Skola: Quit: Lost terminal
14:58 (join) jonrafkind
15:00 (quit) Fare: Ping timeout: 246 seconds
15:04 (join) RacketCommitBot
15:04 RacketCommitBot: [racket] plt pushed 2 new commits to master: http://git.io/SKMdfg
15:04 RacketCommitBot: [racket/master] Renaming slice usage - Jay McCarthy
15:04 RacketCommitBot: [racket/master] Fixing recent errors - Jay McCarthy
15:04 (part) RacketCommitBot
15:13 (join) Fare
15:18 (join) yoklov
15:51 (part) Technodrome
15:58 (quit) dzhus: Remote host closed the connection
16:00 mrcarrot: rostayob: as you have been using both haskell and racket, may i ask if you still use both, and what did get you into racket?
16:05 (join) djcb`
16:37 (join) AmyChua
16:38 AmyChua: hallo, community! what's up?
16:39 offby1: rain, snow, clouds
16:39 offby1: :-|
16:39 rostayob: mrcarrot: I haven't been using racket a lot
16:40 AmyChua: offby1, decf yourself hehe
16:40 chandler: AmyChua: What's with the rotating nicks? Do you keep getting banned in other channels?
16:40 rostayob: mrcarrot: while I spent *a lot* of time with Haskell (and since it was free time, you can guess that I like it)
16:40 AmyChua: no, chandler
16:40 rostayob: mrcarrot: so I'm kind of biased. and I wasn't a big lisp/scheme fan before I dug a bit into racket
16:40 AmyChua: i like being always different as wel as always the same
16:40 AmyChua: :)
16:42 AmyChua: chandler, what is your favorite (alcoholic) drink?
16:44 AmyChua: Oh, so sad... No one wants to talk with me... :(
16:49 (quit) GeneralMaximus: Quit: Leaving
16:49 (quit) jeapostrophe: Ping timeout: 244 seconds
16:54 (join) itasyno
16:56 (join) fftb
16:59 AmyChua: rostayob: if you like Haskell, maybe you would like to check the Shen language too. It's really beautiful!
17:04 (join) Kaylin
17:04 AmyChua: hallo kaylin
17:05 Kaylin: hello!
17:05 AmyChua: what's up?
17:05 AmyChua: homework, etc?
17:05 AmyChua: alcoholic drink?
17:05 Kaylin: not even etc, just homework =/
17:05 AmyChua: hehehe, what a shame
17:06 AmyChua: maybe I could help you?
17:06 AmyChua: :)
17:07 Kaylin: depends, if you have a way to motivate me into doing things I don't want to be doing from far away, you should probably market that.
17:07 AmyChua: oh... neither I can do that... have motivational problems too :(
17:09 AmyChua: kaylin, who is your worst PLT-team member? :)
17:10 Kaylin doesn't understand the question.
17:10 samth: AmyChua: insulting people is not tolerated on this channel
17:10 samth: although we may make an exception for rudybot
17:10 AmyChua: i'm not insulting enyone, samth
17:10 samth: good
17:11 AmyChua: samth, calm down please, i'm totally benign
17:18 (quit) EmmanuelOga: Quit: WeeChat 0.3.7-dev
17:18 AmyChua: Mélodie d'amour chantait le cœur d'Emmanuelle
17:25 (join) dzhus
17:36 (quit) SeanTAllen: Remote host closed the connection
17:37 (quit) sid0: Remote host closed the connection
17:38 (quit) shaatar: Remote host closed the connection
17:38 (join) sid0
17:42 rostayob: I *think* I've already asked this here, but I still haven't found a nice solution
17:42 rostayob: what would be the "idiomatic" way to write a simple parser in racket?
17:45 AmyChua: most idiomatic way would be to pluck racket bad documentation an suffer
17:46 rostayob: AmyChua: oh god, you again? you used some other nick last time
17:47 AmyChua: :)
17:47 rostayob: what's the point?
17:47 AmyChua: joy of life!
17:47 rostayob: ok.
17:47 AmyChua: rostayob: if you like Haskell, maybe you would like to check the Shen language too. It's really beautiful!
17:48 rostayob: AmyChua: I understand you like Shen. I gave a quick look to it, and I didn't find it inspiring
17:48 AmyChua: weird. How so?
17:49 AmyChua: it's similar with 1) lisp 2) haskell
17:50 rostayob: I'm a bit wary of languages that try to do too much. It's very unsupported, and I'd have to be very enthusiastic to risk investing time in it
17:50 AmyChua: it's new and only one man is working on it
17:50 rostayob: exactly.
17:51 rostayob: in general I don't like "multi paradigm" languages. Usually languages with a clear vision are better to me
17:51 AmyChua: but this man has done more than the entire PLT-team!
17:51 rostayob: why are you such a great fan?
17:52 AmyChua: because i can recognise ingenuity
17:52 AmyChua: and Shen is pure ingenuity
17:52 AmyChua: :)
17:52 rostayob: AmyChua: that's very vague.
17:53 AmyChua: luk only at beautiful tutorial of shen. It's comparable to SICP
17:53 AmyChua: luk = look
17:53 rostayob: where?
17:53 AmyChua: here:
17:53 AmyChua: http://www.shenlanguage.org/Documentation/Reference/FPQi/page000.htm
17:54 AmyChua: this is official web site
17:54 AmyChua: http://www.shenlanguage.org/
17:54 rostayob: ah that again. I really don't have time for this.
17:54 AmyChua: now exist Shen version for javascript and Clojure too
17:55 AmyChua: pitty
17:55 rostayob: ok.
17:56 rostayob: also the fact that it's composed of images it's really annoying. tell the author if you know him.
17:56 AmyChua: try to ignore this fact. Soon will come a new release and will be ok
17:57 AmyChua: it's an unimportant detail that has nothing to do with the beauty of language
17:57 rostayob: it's important for me. I can barely read that.
17:57 AmyChua: me too
17:58 AmyChua: nut you can buy the book
17:58 AmyChua: :)
17:58 AmyChua: but
17:59 samth: rostayob: the most idiomatic way would probably be to use the parser-tools collection
17:59 AmyChua: (if you can understand the doc :)
17:59 rostayob: samth: the thing is that I feel like using lexer/parser generators would be detrimental for a small parser
17:59 samth: rostayob: danny yoo recently created a cool planet package for writing grammars: http://planet.racket-lang.org/display.ss?package=autogrammar.plt&owner=dyoo
18:00 (quit) itasyno: Ping timeout: 245 seconds
18:00 rudybot: http://tinyurl.com/7v5fwvp
18:00 rostayob: samth: ok, I'll check that out.
18:00 samth: rostayob: another option is to just hand-write a recursive descenet parser
18:01 samth: AmyChua: i don't feel like you're being helpful here
18:01 (join) jeapostrophe
18:01 rostayob: samth: yes that's what I'd do usually. but there doesn't seem to be a standard set of combinators for racket
18:01 rostayob: samth: also, what's the best way to handle the input? a port?
18:01 samth: rostayob: yes
18:02 (quit) sid0: Remote host closed the connection
18:02 rostayob: ok so I'd gradually consume the port. isn't backtracking inconvenient in that case?
18:02 AmyChua: i have wrote one for "the elements of computing systems" Jack
18:03 AmyChua: for Jack Language
18:03 rostayob: AmyChua: actually, my professor of type systems in mentioned in the "Acknowledgements" section of the book aha
18:03 rostayob: *in type systems
18:03 AmyChua: nice
18:04 AmyChua: and Shen is nice too :)
18:04 (join) agam
18:04 (quit) agam_: Read error: Connection reset by peer
18:04 samth: rostayob: you can put things back on ports if you need to
18:05 (join) shaatar
18:05 (join) rudybot_
18:05 (quit) rudybot_: Remote host closed the connection
18:05 rostayob: samth: yeah but that seems a bit awkward compared to what I usually do (working with immutable lists)
18:05 rostayob: anyways, I'll try to do that, thanks.
18:06 AmyChua: samt, are you married?
18:06 (quit) fftb: Remote host closed the connection
18:07 AmyChua: samth, are you married?
18:09 AmyChua: I have the impression, folks, that you're a little jealous of Tarver and Shen. Nobody said that he liked Shen, surprisingly
18:11 AmyChua: hahaha whole team of prominent PLT scientist cannot beat one lost desoriented man: Mark Tarver!
18:11 AmyChua: hahahahaž
18:11 AmyChua: desoriented but genius!
18:14 Diarmid: I'm assuming that, for some reason, you're trying to drive people away from using Shen.  It's a pity.  Maybe if more people work on things like dependent type systems we'll get to the point of having our code fully verified without much pain.  In any case, you do Shen no favors (probably the idea), and will probably end up banned.
18:14 AmyChua: I'm here to promote http://www.shenlanguage.org/ the best I can and to "insult" PLT clique, hahaha!
18:15 Diarmid: For the record, I find Qi, and derived languages 'interesting', but think the approach may still be not quite right, though probably closer than other attempts, so it's not exactly true that 'nobody likes' them.
18:15 AmyChua: it's nice to hear that, Diarmid
18:15 (join) sid0
18:17 AmyChua: So, people, do you really think Racket is better?
18:18 AmyChua: Please, don't be ignorants: take a slow walt through http://www.shenlanguage.org/ and enjoy!
18:18 AmyChua: walk
18:19 Diarmid: AmyChua: The languages are more similar than most you could have chosen for comparison, but it's rather like asking whether a tree is better than a muffler.  That said, Racket actually has a excellent standard library (GUI, 2-D graphics, OpenGL, sound, networking), which Shen doesn't seem to have.
18:20 AmyChua: it's true
18:20 AmyChua: but shen is so new
18:20 jonrafkind: shen macros don't look hygienic to me
18:20 jonrafkind: they look like clisp macros
18:20 AmyChua: there will be libraries because shen is for integration with other languages: common lisp, scheme, javascript, clojure...
18:21 AmyChua: see this: https://groups.google.com/group/qilang/browse_thread/thread/2568939713a1eba1?hl=
18:21 rudybot: http://tinyurl.com/6txex8p
18:21 Diarmid: For my purposes 'better' involves having good I/O facilities already provided.  I'm probably going to spend most of my time with Racket doing sysadmin-type stuff, with the comfort of knowing that if I ever /need/ to grow something into a application the language will let me do that.
18:22 AmyChua: or this: https://groups.google.com/group/qilang/browse_thread/thread/8cf846d189ae263b?hl=en#
18:22 rudybot: http://tinyurl.com/7fwhtjb
18:22 AmyChua: that you can do in sh shell
18:22 (join) SeanTAllen
18:23 AmyChua: for that there is no need for real programming language
18:23 rostayob: AmyChua: Shen as you say is very new. Most people - like me - don't want to rink spending time in a language that will never be useful (because a language isn't useful without a community that writes libraries and builds an environment). Unless that programming language is very close to my interests. Shen isn't. But I shouldn't spend time writing this since you're trolling
18:23 AmyChua: wrong. You can call every function from host language in shen
18:24 Diarmid: And languages aren't trivial to compare.  I'd rather write some sort of weird custom database, or a planning system, in Prolog than Lisp.  I'd rather write GUI code in Python than Prolog.
18:24 AmyChua: similar to clojure (clojure can call any java lib)
18:24 AmyChua: eh, prolog is integrated in Shen
18:24 AmyChua: :)
18:25 Diarmid: Shell scripting sucks, frankly.  Trying to control string expansion is horrible, making it, IMO, unsuitable even for its original purpose.
18:25 AmyChua: did you try powershell?
18:25 AmyChua: MS invention
18:25 AmyChua: it's great!
18:26 Diarmid: I quit using Bash where at all possible years ago, moving to IPython.  Since then I've actually *needed* to use lambdas (Python's sufficed at the time) and deep recursion (Python does not handle this well) a few times, so I ended up here when I looked for something equivalent but more powerful.
18:26 Diarmid: PowerShell isn't a bad idea, but IPython was already doing what it did, with prettier syntax, at the time.
18:26 AmyChua: I like Ruby more than Python
18:26 AmyChua: but Shen is my favorite... i fell in love immediately
18:26 AmyChua: :)
18:27 Kaylin: it has a prettier name, snakes scare me, gemstones are shiny.
18:27 Diarmid: While I appreciate the blocks as lambdas idea, I have an aversion to complex syntax.  Ruby was too Perl-like to me.
18:27 Diarmid: I'd probably be using Forth if I didn't hate debugging it so much.
18:27 (join) tomodo
18:28 AmyChua: uhhh forth. I once wrote some ilustrations in postscript by hand :)
18:29 AmyChua: postscript is essentially a forth
18:29 (join) siracusa
18:29 Diarmid: I love that the syntax is about as minimal as it gets.  Unfortunately, when something goes wrong, the crash or misbehavior usually happens far from what corrupted your stack, and all you get to play detective with is the stack state at that point.  Trying to use assertions at the start/end of everything is possible, but painful, since almost everything has stack side effects.
18:30 (join) nilyaK
18:30 AmyChua: than you maybe should use Visual Basic hehehe
18:30 AmyChua: :)
18:30 (part) siracusa
18:30 Diarmid: Yes, there's a lot of Forth derivatives.  Joy, Factor, MUF, Postscript...
18:31 Diarmid: I don't see the connection.
18:31 AmyChua: Visual Basic is just for you: no painful stack traces, planty of libraries... :)
18:32 Diarmid: It's C# with syntax designed to resemble VB6.
18:32 AmyChua: I'm talking abuot VB6 actually :)
18:32 AmyChua: and C# is ok, actually
18:32 Diarmid: And VB never exactly had the prettiest syntax or semantics itself.
18:33 AmyChua: better than yava: it has lambdas, LINQ, etc
18:33 AmyChua: yava = java
18:33 Diarmid: Yes, and now contracts, and support for continuations in the CLR.
18:33 AmyChua: continuations? really?
18:33 Diarmid: In the CLR, I don't think it's exposed in C#.
18:33 AmyChua: wow, i didn't know tjat
18:33 (quit) djcb`: Remote host closed the connection
18:34 Diarmid: It was intended to make compiling other languages to the CLR easier.  Happened when they improved the support for dynamic typing, for the same reasons.
18:34 AmyChua: It turns out that Microsoft is great!
18:35 AmyChua: their Language division, at least
18:35 Diarmid: The CLR is pretty nice now.  It's made it possible to use 'weird' languages without persecution, or people even being aware you're using something other than C#.
18:35 AmyChua: (not Sharepoint, sure)
18:35 AmyChua: :)
18:36 AmyChua: and now, there is F#, too!
18:37 Diarmid: Yeah.  I'm not a big fan of that sort of type system, but some people like it, and it seems the design is inspired by OCaml, which I have to respect for its performance (garbage collected, but often matches the performance of C and C++) if nothing else.
18:37 Diarmid: F# is probably responsible for much of the CLR improvements that favor functional languages.
18:38 AmyChua: i heard the rumors that Microsoft will make Shen as default language for programming new Metro system!
18:38 AmyChua: in Visual studio 2015
18:38 Diarmid: More trying to make people hate Shen, huh? ;)
18:38 AmyChua: no, i'm serious!
18:38 AmyChua: :)
18:39 rostayob: Diarmid: but F# throws away the nicest thing in SML/OCaml (the module system)
18:39 rostayob: well the nicest thing in SML. I haven't had much experience in OCaml.
18:39 AmyChua: unfortunately, i don't know much about F# and OCaml
18:39 Diarmid: Shen doesn't need to be associated with a UI designed to only function well on point-and-smudge systems.  Worst advertising ever. ;)
18:40 AmyChua: hehehe, nice try
18:40 AmyChua: :)
18:40 (quit) Fare: Ping timeout: 246 seconds
18:40 Diarmid: rostayob: I wasn't aware of that.  I didn't spend enough time with ML and Haskell (I gave ML a longer go at it though) to become very familiar with it, but coming from Python I do appreciate having things split into lots of small, nested, namespaces.
18:41 tomodo: what's up
18:41 (join) ben_m
18:41 rostayob: Diarmid: what's really cool about SML and OCaml is that modules are first class
18:41 AmyChua: tomodo: nice new language: http://shenlanguage.org/
18:42 rostayob: so you can have, for example, module-level functions that take a module with a certain signature and return some other module using types and values from the given module
18:42 Diarmid: rostayob: I'm guessing this gets used for things like 'mocking', for unit testing?
18:43 tomodo: what's good about it?
18:43 AmyChua: i don't like racket because of: 1) poor documentation, 2) every time I wanna write a macro, i begin with syntax-case, then i give-up and at the end I write macro in classic common lisp unhygienic way
18:43 tomodo: doesn't look at all interesting to me
18:43 Diarmid: rostayob: Or perhaps swapping out interfaces, rather like streams/ports in lisp, so say, the code for working with one kind of database can be replaced with another?
18:43 rostayob: Diarmid: I'm not sure about unit testing, but it's tremendously useful. for example, you can have a functor (module level function) that takes a signature with a type and an ordering on that type and returns a module that lets you create a binary tree on that type
18:44 AmyChua: tomodo, look better
18:44 tomodo: AmyChua: I think you're just mistaken
18:44 rostayob: Diarmid: yes, you can do stuff like that.
18:44 Diarmid: rostayob: I have to admit that's really cool and useful.
18:45 rostayob: it is, that's why I'm angry at F#! ehe
18:45 tomodo: AmyChua: it's easy to get all excited about some new shiny gadgets, but it's just another wacky combination of a bunch of cool sounding - yet useless - trinkets
18:46 Diarmid is rather mystified by the harping on bad documentation.  So far everything I've asked for has been documented, or not supplied, and the only things that didn't come with the standard library that I wanted were the "fmt" library for text formatting, and irregex for doing sick things to regular expressions.
18:46 AmyChua: just like a racket, tomodo
18:46 AmyChua: :)
18:46 rostayob: Diarmid: yeah, the docs that come with racket are excellent...
18:47 AmyChua: i don't agree at all
18:47 AmyChua: documentation for Shen is a way better!
18:47 tomodo: AmyChua: racket is irrelevant
18:47 tomodo: AmyChua: shen is a joke
18:48 AmyChua: no, it's not. It's veri thoughtful product
18:48 rostayob: aren't there any ops here?
18:48 AmyChua: made by genius Mark Tarver
18:48 Diarmid: tomodo: All languages suck, or blow, the important thing is linking them together to achieve balance? :P
18:48 tomodo: heh, no
18:49 tomodo: scheme is good, racket implements it welll
18:49 tomodo: this is irrelevant to whether or not shen is worth anything
18:49 tomodo: AmyChua is just dodging the critism by trying to move attention to racket
18:49 Diarmid: rostayob: I gather they're waiting for AmyChua to get bored, or finish making people not want to use Shen, due to the trolling in this channel.
18:50 AmyChua: i'm surprised i'm not banned yet
18:50 Diarmid: AmyChua: I'm surprised you're trying so hard.
18:50 ben_m: I'm impressed by this channel's patience and maturity :D
18:50 Diarmid: AmyChua: Don't you have anything better to do?
18:50 AmyChua: unfortunately, no
18:50 ben_m: AmyChua, who I presume is shen from yesterday, keeps trying to bash Racket and advocate his own language
18:51 ben_m: Trying to bait people into arguments
18:51 Diarmid: AmyChua: You could be adding good I/O support to Shen, so people like me might actually use it?
18:51 ben_m: And nobody took the bait yet :)
18:51 AmyChua: I suffer from deep dark depression
18:51 tomodo: why would you ever use this frankenstein?
18:51 Diarmid: AmyChua: Perhaps it's from feeling like everybody hates you, due to your trolling?
18:51 (quit) jonrafkind: Ping timeout: 245 seconds
18:51 AmyChua: maybe, diarmid
18:52 tomodo: shen is a waste of time
18:52 rostayob: btw, tomodo is a troll from another channel, I made the mistake of mentioning that someone was spamming links here and he came here.
18:52 rostayob: to put fuel on the flames I suppose
18:53 AmyChua: nice, i'm a troll too
18:53 AmyChua: we are brothers in arms
18:53 AmyChua: :)
18:53 tomodo: what are you talking about
18:53 rostayob: *there and he came here
18:53 Diarmid: tomodo: I actually like what people are trying to achieve with defendant type systems.  They still don't seem to offer a way to state the kinds of preconditions/postconditions/invariants/assertions that I actually want to use (contracts do that better), but the automatic checking is nice, and maybe if they keep at it they'll get there.
18:54 samth: rostayob: tomodo has been in this channel for a while, and hasn't seemed like a troll
18:54 AmyChua: and me?
18:54 AmyChua: :)
18:54 tomodo: Diarmid: shen does not satisfy the de bruijn criterion. It is useless if you want to prove things about your programs
18:54 rostayob: oh well. my bad. he was banned on #haskell-blah, and he started talking to AmyChua after I mentioned what was happening here there.
18:54 tomodo: its' inconsistent too
18:54 tomodo: it's basically a huge nebulous mess of rubbish
18:55 AmyChua: shen is great!
18:55 rostayob: samth: they unbanned him recently there
18:55 tomodo: false for the reasons I've just given
18:55 AmyChua: shen is great, mark tarvedr is fucking genius!
18:55 samth: ok, everyone, discussion of shen is now over
18:55 AmyChua: :(
18:55 AmyChua: samth, why you don't like shen?
18:56 AmyChua: i'm really interested in your reasons
18:56 samth: AmyChua: i have no problems with shen -- i've never used it
18:56 samth: but it is *not* the topic of this channel
18:56 AmyChua: i know, but it's just stronger then me... :)
18:57 AmyChua: beside shen, it seems that the main topic of this channel is silence
18:57 Diarmid: tomodo: In what way is it inconsistent?  I assume you mean a logical paradox?
18:57 tomodo: Diarmid: yeah
18:57 AmyChua: so beter shen, then silence, no?
18:57 AmyChua: sound of silence
18:57 Diarmid: tomodo: Hmm, I'm curious.  I've not heard this about Qi, the language it was supposedly based on.  Can you explain a bit more?
18:57 AmyChua: hello darknes my old friend...
18:57 tomodo: let's move to #shen as to not upset samth
18:57 rostayob: Diarmid: tomodo didn't know what shen was 15 mins ago.
18:58 tomodo: same for qi
18:58 rostayob: *what shen is
18:58 tomodo: it's all rubbish
18:58 AmyChua: i come to talk with you again...
18:59 Diarmid: tomodo: Perhaps this is naive, but since it allows you to specify the type of everything, and how the types are checked, it seems like if there /was/ a inconsistency, it would have to be due to the type specifications rather than the language itself.
18:59 (join) debrujin
19:01 Diarmid: tomodo: Anything with a Turing-complete type system can certainly have that problem, along with non-termination.
19:01 tomodo: yeah, Turing-complete is a mistake
19:02 Diarmid: tomodo: It's the only way to 'slay the beast' though.  If you want to be able to automatically verify things, you have to be able to add rules to the system OR restrict yourself to a set of already-verified components.  Most people aren't willing to give up recursion and while loops, so you pretty much have to give them a 'programmable type system'.
19:02 rostayob: Diarmid: anything turing complete language leads to an inconsistent logic at the type level
19:03 tomodo: no that's not true
19:03 rostayob: if you have general recursion, every type is inhabited
19:03 Diarmid: tomodo: I'd be very interested in alternatives.
19:03 rostayob: in other words you can prove everything
19:03 tomodo: let's talk about in #shen
19:05 (quit) agam: Read error: Connection reset by peer
19:05 (join) fizzbang
19:05 (join) agam
19:06 (quit) Kaylin: Read error: Connection reset by peer
19:08 (quit) mceier: Quit: leaving
19:13 debrujin: sound of silence, again???
19:13 debrujin: :(
19:13 debrujin: so sad! :((((((
19:13 debrujin: pity, people went to #shen
19:13 debrujin: :)
19:14 ben_m: You might want to consider why you were kicked and avoid it happening again
19:14 debrujin: i'm so sorry that samt don't understand me and think that i'm destructive
19:14 debrujin: i'm so nice person, totally benign
19:15 ben_m: You've said yourself that you're a troll, that's pretty much the opposite of what nice, benign and non destructive.
19:15 (quit) abbe: Quit: Aah IRC, where men are men, women are men
19:15 debrujin: i was just kidding
19:15 ben_m: Anyways, don't want to fuel this stuff, I'll be coding some Racket :D
19:16 debrujin: i try to animate people to talk, all the time
19:16 debrujin: but it's just silence here
19:16 debrujin: why is that so?
19:16 debrujin: is it really the case that nobody have noting to say?
19:17 (join) ashish
19:17 offby1 is hiding from the flamage
19:18 rostayob: debrujin: you missed an "i"
19:18 debrujin: ?
19:18 rostayob: it's de bruijin
19:18 debrujin: ah, thanks
19:18 rostayob: no problem.
19:18 rostayob goes to study
19:30 (quit) debrujin: Quit: Page closed
19:37 ben_m: Is there a nice way to pass a variable to the next on-paint call of a canvas%? ie. I want to animate something using no state if possible
19:38 (join) realitygrill
19:40 (join) dnolen
19:52 tomodo: it's just like, come on man.. im trying to give you useful advice you can put into practice.. this isn't a battle me vs you
19:52 Diarmid: ben_m: I'm not sure this is considered clean, but perhaps you could, upon calling it, return a lambda that would call it with the next value?
19:53 ben_m: I've thought about that, that would still need a bit of state though, right?
19:54 ben_m: To store the lambda for the next on-paint call
19:54 Diarmid: ben_m: I may not be the best person to ask about this, because I consider the whole 'there is no state in functional programming' to be about as much bunk as the 'pure functional I/O' stuff, but it's my understanding that returning a curried function would be considered 'pure' by people that care about such things.
19:55 Diarmid: ben_m: I guess that depends on what the rest of your code looks like.  You could tail-recurse to call yourself to avoid that, or something.
19:56 ben_m: Well the problem is I'm using racket/gui, so I don't have control over the event loop. The best I can do is override on-paint, but I can't control how it gets called.
19:56 (quit) kampsun: Ping timeout: 246 seconds
19:56 ben_m: As far as I understand racket/gui
20:04 eli: samth, stamourv: ping
20:11 Diarmid: ben_m: Do I understand correctly that you want to, basically, pass the event handler data 'from outside', like variables used by a closure?
20:12 (quit) shaatar: Ping timeout: 245 seconds
20:13 (quit) SeanTAllen: Ping timeout: 245 seconds
20:13 (quit) sid0: Ping timeout: 244 seconds
20:13 (quit) masm: Quit: Leaving.
20:14 ben_m: What I want, optimally, would be that the callback gets passed a state variable and the return value of the callback will be this state the next time the callback gets called. Sorry if that's confusingly explained.
20:14 ben_m: However, I just realize that on-paint isn't called periodically, so that won't work anyways.
20:15 Diarmid: ben_m: It's only called as needed, when the area is 'dirty?
20:15 ben_m: yeah
20:16 Diarmid: ben_m: I'm sure there's a way to hang things off a timer interrupt somehow.
20:16 ben_m: mhm
20:16 ben_m: 2htdp/universe does it somehow
20:16 ben_m: So there must be a way :D
20:17 Diarmid: there is a "timer%".
20:18 Diarmid: http://docs.racket-lang.org/gui/timer_.html
20:18 ben_m: Reading :)
20:18 (join) asurai
20:21 (quit) realitygrill: Read error: Connection reset by peer
20:22 (join) realitygrill
20:27 (quit) fizzbang: Quit: leaving
20:29 (join) shaatar
20:30 rostayob: samth: you mentioned that "you can put things back into ports if you want to". if I open a port from a mutable string, how can I do that?
20:30 rostayob: also, is there an easy way to "seek" ports, but skipping characters and not bytes?
20:31 (quit) yoklov: Quit: computer sleeping
20:34 rostayob: more than seek arbitrarly, simply go forward or backwards (so seek with SEEK_CUR)
20:36 (quit) dzhus: Remote host closed the connection
20:39 (join) SeanTAllen
20:40 (quit) dnolen: Ping timeout: 252 seconds
20:41 (join) yoklov
20:44 Diarmid: rostayob: http://docs.racket-lang.org/reference/port-lib.html#%28def._%28%28lib._racket/port..rkt%29._read-string-evt%29%29
20:44 rudybot: http://tinyurl.com/6p5ko5t
20:44 Diarmid: Err, sorry, lemme try that again
20:45 Diarmid: rostayob: http://docs.racket-lang.org/reference/port-lib.html#%28part._.Creating_.Ports%29
20:45 rudybot: http://tinyurl.com/76g7fcj
20:46 Diarmid: rostayob: It looks like you can append and seek string ports with some of the functions given there.
20:49 (quit) vkz: Quit: vkz
20:49 rostayob: Diarmid: thanks, I'll take a look.
20:50 rostayob: Diarmid: were you referring to peeking-input-port? because that provides half of what I need (going forward, while I also want to go backwards)
20:51 rostayob: also, it has the annoying downside that if the original port is read, the peeking port will skip the input
20:51 Diarmid: rostayob: Is there a reason to not use an array?
20:52 Diarmid: rostayob: http://docs.racket-lang.org/guide/vectors.html
20:53 rostayob: Diarmid: ports seem to provide a really nice interface if I need to write a parser.
20:53 Diarmid: rostayob: That might be what you want if you want true, fast, painless random access.
20:53 rostayob: I don't want random access, I just want to consume a stream with the ability to backtrack
20:53 Diarmid: rostayob: I suppose that's true, but I'm not sure why you'd need to 'read backwards' in a file.  Lexers generally go forward, at least.
20:54 rostayob: Diarmid: my plan was to write some parser combinators in the style of parsec, so no need of explicit lexing
20:54 nilyaK: language isn't LLK, doesn't want to store data they might not need and can't know ahead of time? idk =/
20:54 Diarmid: rostayob: Depends on what kind of language you're trying to parse, but you can often back up by using a recursive descent parser.  The lexer itself should be able to get by with a simple string port.
20:55 rostayob: Diarmid: it's not really because of the grammar, but for ease of programming
20:56 rostayob: I don't know if you've ever use parsec, but it's really nice
20:57 Diarmid: rostayob: I probably can't help then.  I'm not familiar with parsec, and learned to think of things in terms of lexing and parsing (as separate events).  The lispy way to do things is usually write a reader macro or just use a regular macro to embed a domain specific language, but if you need to parse C++ or Perl that won't do you much good.  I'm of the unpopular opinion that once you go past LL(1) you've messed up.
20:58 (quit) nilyaK: Quit: Leaving.
20:59 rostayob: Diarmid: I'm used to writing quick parsers easily with parsec now, it's really great (and for big parsers parsec is great since it can generate lexers internally)
20:59 Diarmid has a look.  Finds it hard to imagine mixing lexing in with parsing as making anything easier.
21:00 rostayob: Diarmid: http://hackage.haskell.org/packages/archive/parsec/3.1.2/doc/html/Text-Parsec-Token.html
21:00 rudybot: http://tinyurl.com/7fgoxaa
21:00 rostayob: trust me, I haven't met a person who doesn't like parsec (and parsec is not even that great)
21:00 rostayob: still, it seems very strange that I can't do fseek-style operations on ports.
21:01 (join) kampsun
21:02 (quit) tomodo: Quit: leaving
21:02 (quit) em: Ping timeout: 260 seconds
21:05 Diarmid: rostayob: streams/ports remind me very stroingly of dataflow language streams.  They're very nice if you plan your program around streaming data in and out, and it can be memory efficient by not needing to hold the whole thing in memory.
21:05 (join) jrslepak
21:06 Diarmid: rostayob: I'd normally think of a string port being fed into a lexer generator that returns lexemes, being called repeatedly by a parser that builds/decorates a tree representing the source.  None of that should require doing anything but reading forward in the port.
21:08 rostayob: Diarmid: ok, that sounds reasonable :)
21:08 Diarmid: rostayob: If you're parsing something hideous the parser might need to go through a lot of contortions, remembering things and 'looking ahead', but you should still be able to get by without moving back and forth in the stream.  Heck, a lot of systems do the lexing all in one pass, then do parsing next, though it means you often end up lexing the whole file unnecessarily if there's a syntax or (detectable) semantics  error.
21:09 rostayob: Diarmid: no, I'm actually parsing something really easy to parse. I'm just used to parsec, which works great for this kind of things (parsing small languages)
21:11 Diarmid: rostayob: At first glance it looks like parsec is a set of little functions often found in parsers.  I'm sure it's useful (sounds like sort of the opposite of the "fmt" library), but I don't think it'd be all that hard to write something similar in Scheme.
21:14 rostayob: Diarmid: oh it's not hard at all. I was wondering on how to write "try" when working on ports.
21:15 rostayob: try is a function that takes a parser, and if the parser fails puts back the input that the parser has consumed
21:15 (join) em
21:16 Diarmid: rostayob: The page I showed you on making 'custom' port types has a example of appending ports.  I'd think you could basically append "oops I couldn't use this" to "rest of the string" to put it back.
21:17 rostayob: Diarmid: yeah but what I want is to put it back to the beginning of the string
21:17 Diarmid: rostayob: Though you might consider whether that's the best way to go about it.  Nested cond (like switch-case) so you always 'go forward' or fail to match at all, could avoid the need to do such things.
21:18 Diarmid: rostayob: Can't you just 'append' things in that order then?  If you decide you really want to stick it back on?
21:20 rostayob: Diarmid: oh right, I could. It's late :)
21:20 Diarmid: rostayob: Don't worry about it. :)
21:21 rostayob: ok, going to bed
21:21 rostayob: Diarmid: thanks for the help, I'll be back!
21:21 Diarmid: rostayob: Sleep well.
21:21 rostayob: thanks
21:21 (quit) rostayob: Quit: WeeChat 0.3.5
21:26 (join) vkz
21:26 (quit) em: Ping timeout: 252 seconds
21:32 (quit) kampsun: Read error: Connection reset by peer
21:34 (join) em
21:34 asumu: Diarmid: there is a parsec-esque library for Racket called parseq on PLaneT.
21:34 asumu: I can't attest to how good it is, but it's there.
21:35 asumu: (It needs better docs... been meaning to ask YC if he'd accept patches)
21:36 Diarmid: asumu: I'll give it a look.  I kind of like the idea, especially given the symmetry with fmt, which I really like.
21:41 asumu: Seen on Twitter: https://twitter.com/#!/dlouhy/status/180046188660269057
21:41 asumu: Diarmid: It has docs, but as a series of blog posts. Good blog posts, but not scribble docs nonetheless.
21:47 (quit) em: Ping timeout: 240 seconds
21:52 Diarmid: asumu: I'm curious how people would feel if someone used this to still split things into separate lexing and parsing.
21:53 asumu: Diarmid: Hmm, I thought that you could still do that in parsec.
21:54 asumu hasn't used it in a while though
21:54 Diarmid: asumu: I think it'd be doable, but it looks like people usually mingle it together.  I wondered if people would find that 'ugly' since I'm a bit more comfortable with the old fashioned way of thinking about it.
21:59 asumu: Diarmid: http://legacy.cs.uu.nl/daan/download/parsec/parsec.html#SeperateScanners
22:01 Diarmid: asumu: Ok.  So I wouldn't get yelled at for that. :)
22:02 asumu: Yeah, probably not. :)
22:08 (quit) yoklov: Quit: computer sleeping
22:17 (quit) bmp: Disconnected by services
22:17 (nick) asurai -> bmp
22:18 (join) yoklov
22:29 (quit) bmp: Quit: Bye!
22:31 (join) veer
22:32 (join) mithos28
22:39 (quit) vkz: Quit: vkz
22:51 (join) em
22:57 (quit) em: Ping timeout: 240 seconds
23:05 (part) ben_m: "Leaving"
23:07 (join) jonrafkind
23:08 (quit) agam: Quit: Computer has gone to sleep
23:23 (join) Fare
23:27 (join) AlbireoX
23:42 (join) em