00:02 (topic) eli: Racket -- http://racket-lang.org/ -- Matthew @ FLOSS Weekly on Racket, http://twit.tv/floss167/ -- Come to RacketCon: July 23-24 -- logs @ http://racket-lang.org/irc-logs/
00:02 eli: jonrafkind: ?
00:02 jonrafkind: Matthew Flatt
00:03 eli: Too long.
00:03 eli: Too long.
00:03 jonrafkind: rofl, fine
00:04 eli: The second one was a typo, I was about to add it.
00:04 jonrafkind: huh
00:04 jonrafkind: about to add Matthew Flatt to the topic?
00:04 eli: Yes.  But if it's fine then I'll leave it.
00:30 (quit) JoelMcCracken: Remote host closed the connection
00:30 (join) JoelMcCracken
01:01 (quit) ZachBeta: Ping timeout: 252 seconds
01:01 (quit) JoelMcCracken: Remote host closed the connection
01:02 (join) JoelMcCracken
01:03 (join) ghiu
01:03 (quit) ghiu: Client Quit
01:27 (join) dnolen
01:27 (quit) dnolen: Client Quit
01:33 (join) realitygrill_
01:34 (quit) realitygrill: Ping timeout: 258 seconds
01:34 (nick) realitygrill_ -> realitygrill
01:45 (quit) Nightwolf: Remote host closed the connection
02:14 (join) hkBst
02:28 (quit) realitygrill: Quit: realitygrill
02:55 (quit) jonrafkind: Ping timeout: 250 seconds
03:16 (join) masm
03:30 (join) Nightwolf
03:35 (quit) Demosthenes: Ping timeout: 240 seconds
03:37 (join) Demosthenes
03:54 (join) dnolen
03:56 eli: hkBst: What's the different spelling?
03:57 hkBst: eli: the dash in the url
03:57 hkBst: hashtables.html vs hash-tables.html
03:57 eli: "Meh"?
03:58 hkBst: eli: sure, just couldn't help noticing :)
03:58 eli: It's actually easy to change, but I'm not sure that it's worth it.
03:58 eli: (eg, old links etc)
03:59 (quit) JoelMcCracken: Ping timeout: 255 seconds
03:59 eli: hkBst: An re your question, what's wrong with `zero?' over the count?
04:01 hkBst: eli: nothing, except that it might be inefficient. The reference seems to say it's not constant time for weak hashes, though I don't know why that should be.
04:02 ohwow: hello
04:04 eli: hkBst: That sounds like a possible typo.
04:04 eli: The link to the caveats, that is.
04:04 eli: I'll reply on the list.
04:04 hkBst: thanks eli :)
04:12 eli: hkBst: See my reply --
04:12 (quit) dnolen: Quit: dnolen
04:12 eli: specifically, grep for `hash-count' in src/racket and you'll see the function,
04:12 eli: and in the weak case it counts the items that are still alive.
05:00 (join) PLT_Notify
05:00 PLT_Notify: racket: master Eli Barzilay * 9e9ad2f (1 files in 1 dirs): Improve the generation of bug form fields; add a captcha file for future ... - http://bit.ly/lVSfJu
05:00 (part) PLT_Notify
05:36 (join) mceier
05:38 (join) lucian
05:50 (join) MayDaniel
06:03 (quit) masm: Ping timeout: 241 seconds
06:08 (join) masm
06:08 (quit) MayDaniel: Read error: Connection reset by peer
06:10 (quit) lucian: Quit: Ex-Chat
06:10 (join) pistacchio
06:12 (join) lucian
06:17 pistacchio: hi
06:18 pistacchio: how can i explicitly state "the current input port is the keyboard"?
06:40 (quit) ckrailo: Quit: Computer has gone to sleep.
06:57 (quit) Checkie: Ping timeout: 260 seconds
07:00 em: what is an AST?
07:02 bremner_: usually abstract syntax tree
07:02 bremner_: but I don't know specicically in Scheme/Racket
07:06 em: what's an abstract syntax tree do?
07:06 bremner_: it's just a datastructure, usually output from a parser, representing come code/other-structured-text
07:06 ohwow: Well, when you parse a language, you get an abstraction syntax tree which you can evaluate or compile
07:07 bremner_: if you're really curious, you can look at the first few chapters of "Programming Languages, Application and Interpretation"; it's online and in racket
07:07 ohwow: For example, "x = 3*(y-z)" yeilds this tree: http://www.redhat.com/magazine/002dec04/features/gcc/figs/ast.png
07:08 ohwow: Lisp code looks like AST itself
07:08 ohwow: since (+ x y) is just a tree with '+' on top and two branches: 'x' and 'y'
07:36 (quit) pistacchio: Ping timeout: 252 seconds
07:52 (join) jeapostrophe
07:53 (join) samth
07:59 (nick) mario-go` -> mario-goulart
08:06 (quit) aoh: Remote host closed the connection
08:15 (quit) lucian: Remote host closed the connection
08:15 (join) lucian
08:19 (join) Lajla
08:23 (quit) blomqvist: Quit: Heroes die once, Cowards live longer!
08:23 (quit) lucian: Remote host closed the connection
08:24 (join) blomqvist
08:24 (join) aoh
08:36 (join) lucian
09:08 (join) DT``
09:14 (quit) aoh: Ping timeout: 240 seconds
09:16 (quit) samth: Ping timeout: 255 seconds
09:19 (join) MayDaniel
09:28 (quit) jeapostrophe: Quit: jeapostrophe
09:34 (join) aoh
09:35 hkBst: what's the difference between a sequence and a stream? According to "3.14.2 Streams" in the reference a stream is a kind of sequence that supports functional iteration. But why is the interface for making a struct look like a sequence so different from making it look like a stream?
09:58 (join) PLT_Notify
09:58 PLT_Notify: racket: master Matthew Flatt * 9f1c214 (1 files in 1 dirs): refine comment that is possibly misleading ...
09:58 PLT_Notify: racket: master Matthew Flatt * f3e1cf1 (2 files in 2 dirs): change `hash-count' to take the lock; fix docs
09:58 PLT_Notify: racket: master Matthew Flatt * bbfeb85 (3 files in 3 dirs): mak argument to `make-immutable-hash' optional
09:58 PLT_Notify: racket: master Jose Lopes * 84415dd (1 files in 1 dirs): Fixed bug that prevented from accessing ActiveX components that did not expose type information. ...
09:58 PLT_Notify: racket: master commits 9e9ad2f...84415dd - http://bit.ly/kgLzGS
09:58 (part) PLT_Notify
10:02 (nick) samth_ -> samth
10:03 (join) jeapostrophe
10:16 samth: hkBst, sequences are more general than streams
10:18 samth: in particular, iterating through a sequence may produce side effects
10:19 samth: for example, reading from a network port
10:19 samth: which you can't do "stream-rest" of
10:20 hkBst: samth: but you can make a stream out of each sequence, no?
10:21 (topic) samth: Racket -- http://racket-lang.org/ -- mflatt interview: http://twit.tv/floss167/ -- Come to RacketCon: July 23-24 -- logs @ http://racket-lang.org/irc-logs/
10:21 (topic) samth: Racket: http://racket-lang.org/ -- mflatt interview: http://twit.tv/floss167/ -- Come to RacketCon: July 23-24 -- logs @ http://racket-lang.org/irc-logs/
10:21 (topic) samth: Racket: http://racket-lang.org/ -- mflatt interview: http://twit.tv/floss167/ -- Come to RacketCon: July 23-24 -- logs: http://racket-lang.org/irc-logs/
10:22 samth: hkBst, no, you can't
10:22 samth: well, you could iterate through the whole sequence, build a list, and use that as a stream
10:22 samth: but that isn't usually what you want
10:25 hkBst: samth: so each sequence that can be iterated without side-effects is a stream?
10:29 samth: well, streams are really things created with the racket/stream library
10:29 samth: but that's the idea
10:34 hkBst: samth: if so, I dont get why making a struct be a sequence requires 6 values passed to make-do-sequence and making it be a stream requires instead a vector of 3 procedures.
10:35 hkBst: s/if so, //
10:45 (quit) lucian: Read error: Connection reset by peer
10:46 (join) lucian
10:46 samth: hkBst, streams have a limited api, which supports just `first' and `rest'
10:46 samth: sequences have a more general API
10:46 samth: probably, it could be organized differently, so that you could provide fewer of the procedures
10:58 (quit) hkBst: Read error: Connection reset by peer
10:59 (join) PLT_Notify
10:59 PLT_Notify: racket: master Matthew Flatt * a7ae3b8 (2 files in 2 dirs): fix docs for `multi-in' ...
10:59 PLT_Notify: racket: master Matthew Flatt * fbbb30f (1 files in 1 dirs): try to clarify intent of `#:contracts' in `defform'
10:59 PLT_Notify: racket: master commits 84415dd...fbbb30f - http://bit.ly/l8nWgM
10:59 (part) PLT_Notify
11:16 askhader: Man... I want to write some racket code for work for a change.
11:29 (join) jonrafkind
11:39 (join) asumu
11:51 offby1 laughs cruelly
11:51 offby1: askhader: your workplace allows you to use weird obscure languages?
11:53 askhader: No, just silly pythonic ones.
11:53 askhader: And when I try to exploit the functional features of python I'm not met with happy supervisors.
11:55 askhader: THe same goes for functional javascript.
11:59 em: wow im watching this video. Mathew Flatt is #@$%#@%$ awesome!
12:03 offby1: em: I keep saying: he looks so YOUNG
12:03 em: I assumed he is
12:03 offby1: 40-ish, I'm told
12:03 em: wow that's weird
12:03 em: he looks about 28
12:03 offby1: it's the hair gel.
12:03 em: hehe yeah
12:03 offby1: and the enthusiasm :)
12:15 ohwow: yeah hair gel
12:15 ohwow: i even wrote about his coolness to the mailing list
12:24 em: offby1: there was a plug for your bot in the interview! :)
12:31 (part) jao: "Killed buffer"
12:34 (quit) MayDaniel: Read error: Connection reset by peer
12:35 (join) bbowes
12:38 (part) bbowes
12:41 (join) bbowes
12:54 (quit) lucian: Quit: Ex-Chat
12:55 (join) lucian
13:14 (quit) masm: Read error: Operation timed out
13:23 offby1: I noticed!
13:26 (quit) jeapostrophe: Quit: jeapostrophe
13:28 ohwow: http://stackoverflow.com/questions/1878458/1130-host-localhost-is-not-allowed-to-connect-to-this-mysql-server lol
13:28 rudybot: http://tinyurl.com/422qhrw
13:31 (join) lucian_
13:33 (join) bugQ
13:33 (quit) lucian: Ping timeout: 252 seconds
13:34 bugQ: well despite minor gripes I have to say that learning racket/scheme has been the least painful of any language so far.
13:41 bugQ: granted I had researched lisp a bit already
13:41 (join) jeapostrophe
13:46 (nick) lucian_ -> lucian
13:47 (part) Lajla: "I am gay for you"
13:51 (join) masm
13:55 (quit) jeapostrophe: Quit: jeapostrophe
13:57 (join) MayDaniel
14:08 (quit) bbowes: Quit: ERC Version 5.3 (IRC client for Emacs)
14:22 em: Suppose a person would like to use racket to collect data in an on going way. Would the smart thing be to write it to a file?
14:22 bremner_: I think so.
14:23 em: in that case, what would be good is if there were ways to write to the top of the file and push everything else down sometimes, and write to the end of the file other times, without having to re-write the whole existing file
14:23 em: does that make sense?
14:25 bremner_: it makes sense, but I don't know a good answer using (non-database) files
14:25 em: i see so that's perhaps tricky with a flat file of text
14:25 bremner_: I think so, mainly the inserting at the beginning
14:26 em: how does a word processor do it?
14:26 bremner_: with a data-structure in memory
14:27 em: yeah i could keep it all in a data structure in memory and maybe just write a file for a backup
14:27 bremner_: that might be easiest.
14:28 em: but can the data structures in racket get very very large?
14:28 em: this could grow to be a lot of data
14:28 bremner_: sure
14:28 ohwow: em: well the stupid way would be to
14:28 ohwow: save data from a file
14:28 ohwow: delete everything from a file
14:28 ohwow: write your data
14:28 ohwow: write your save
14:28 bremner_: em: what is large for you? megabytes is OK, but gigabytes is tough
14:28 em: i want to analyse something that will have 8 different numbers associated with it, every 5 minutes, indefinitely.
14:29 offby1: em: funny that you're asking about collecting data and writing it to a file.  rudybot does just that, but I'm slowwwly working to get the data into Amazon's Web Services
14:29 em: actually 16 numbers and a time stamp so i guess 17 different integers every 5 minutes.
14:30 offby1: and this process will run for many days, or weeks, etc
14:30 bremner_: can you just write some kind of index to the front of the line, and sort the file later?
14:30 em: yeah probably.
14:30 offby1: database sounds like the right thing to me -- something that maintains an index for you, so that when you want to pull data out of this mess, it will be quick
14:30 em: offby1: yep
14:30 offby1: that's precisely the problem I have with rudybot!
14:31 em: yeah i want to run it for a couple months at least.
14:31 offby1: getting the data out, actually, is currently fairly easy
14:31 offby1: my real problem is: when the computer on which he runs goes wonky (which happens more often than you'd expect), I sometimes forget to copy the old logs over to the new box.  So I want those logs to live "in the cloud" instead of on the local disk, since that way they'll always be available.
14:32 em: makes sense
14:34 offby1: also, I just wanna play with cloud services, 'cuz they're cool :)
14:58 (quit) MayDaniel: Read error: Connection reset by peer
14:59 (quit) DT``: Ping timeout: 276 seconds
15:11 (quit) bugQ: Read error: Operation timed out
15:13 (join) jeapostrophe
15:14 (join) DT``
15:18 (join) bugQ
15:19 (quit) jeapostrophe: Quit: jeapostrophe
15:21 (join) ghiu
15:23 (join) anRch
15:45 offby1: given a structure, is there an easy way I can get the field names?  Ultimately I'd like to be able to write a function that converts the struct to a dict, where the keys are the field names and the values are (of course) the struct's slot values.
15:46 samth: offby1, it's pretty hard, actually
15:46 offby1: bah
15:46 samth: you probably need the static struct info
15:46 offby1: I've written my own (thing->dict), which wasn't hard, of course, but it just seems wrong that I had to write it at all.
15:46 offby1: given that it duplicates the list of field names.
15:47 samth: as you see here: http://docs.racket-lang.org/reference/inspectors.html?q=struct-info#(def._((quote._~23~25kernel)._struct-type-info))
15:47 samth: the struct type doesn't have field names
15:48 samth: i think this is something we should add to struct types
15:48 offby1: it'd be handy
15:48 offby1: https://github.com/offby1/rudybot/blob/userdb-on-aws/userinfo.rkt#L123 is the specific case I'm working on
15:49 offby1: hm, s/123/120/ for the relevant comment
15:49 samth: it would be pretty easy to write a macro that took static struct info and generated such a procedure
15:49 offby1: really?
15:49 offby1: what'd I have to pay you to write it? :)
15:49 samth: yeah, the field names are in the static info
15:50 offby1: ok, how do i get there here static info?
15:50 offby1: s/there/this/
15:50 samth: look here: http://docs.racket-lang.org/reference/structinfo.html
15:51 offby1: ah yes, I'd read that page before, but frankly didn't understand it.
15:52 offby1: I suspected it _might_ be relevant, but couldn't figure out what to do with the functions it mentions.
15:52 samth: take a look at this: https://github.com/plt/racket/blob/master/collects/racket/match/parse-helper.rkt#L71
15:52 rudybot: http://tinyurl.com/3wghco6
15:54 offby1: I'm afraid that's too complex for my tiny brain
15:57 (join) tauntaun
15:59 eli: samth: field names are a syntactic thing, putting in the struct types is leaking information in the same way as making lambdas remember their argument names.
16:00 offby1: I'd be happy if this thing only worked for prefab structs
16:01 offby1: eli: I can explain what I'm trying to do: I'd like to easily send a struct type to Amazon's "simple DB" service.  To do that, I need to send a sequence of attribute-name/attribute-value pairs.
16:01 offby1: the values can come from the struct's "slots", and I'd like the names to be the field names.
16:02 eli: offby1: prefab structs are even more problematic:
16:02 eli: rudybot: (struct foo (x y) #:prefab)
16:02 rudybot: eli: your sandbox is ready
16:02 rudybot: eli: Done.
16:02 eli: rudybot: (struct bar (xxx yyy) #:prefab)
16:02 rudybot: eli: Done.
16:02 eli: rudybot: (foo 1 2)
16:02 rudybot: eli: ; Value: #s(foo 1 2)
16:02 eli: rudybot: (bar 1 2)
16:02 rudybot: eli: ; Value: #s(bar 1 2)
16:02 eli: So there's no mention of field names at all.
16:02 offby1: true true
16:02 eli: I suspect that what you want is a hash table.
16:03 eli: Perhaps wrapped in some code that makes it look like structs.
16:03 offby1: hmph
16:03 offby1: yeah, that'd probably work.
16:03 eli: But that would be equivalent to writing some `struct->hash' thing.
16:03 offby1: yep
16:03 eli: ...which wouldn't be too hard actually.
16:03 offby1: that'd be ideal.
16:03 eli: You could even do it properly, with a struct property.
16:03 eli: (Similar to a custom printer.)
16:03 offby1: dunno what you mean by that
16:04 offby1: I've barely ever used struct properties
16:04 eli: A custom printer controls how some struct gets printed.
16:04 eli: Your custom thing would control how it gets converted to a hash.
16:04 DT``: wouldn't serialize help here? (just asking)
16:04 offby1: DT``: it might.
16:04 eli: Serialize could help too, of course.
16:04 offby1: I haven't looked into that
16:05 eli: And serializing is trivial if you're dealing with prefab.
16:05 eli: You can serialize with just format:
16:05 eli: rudybot: (format "~s" (foo 1 2))
16:05 rudybot: eli: ; Value: "#s(foo 1 2)"
16:05 offby1: yes, but as we established: we don't get field names with prefab structs.
16:05 eli: That's fine as long as you read it later with the right prefab definition.
16:05 DT``: eli, then what's the difference between a serializable-struct and a struct with prefab?
16:06 eli: DT``: prefabs are non generative structs.
16:06 (quit) jonrafkind: Read error: Operation timed out
16:06 eli: So, a (struct foo (x y) #:prefab) just creates bindings for `foo', `foo-x', etc.
16:07 eli: But (struct foo (x y)) returns a *new* struct type, always.
16:07 samth: eli, i strongly disagree that field names are syntactic
16:07 eli: So serializing with the latter is more difficult, since the name isn't meaningful; and with the former it's trivial since all you have is the name.
16:07 samth: or rather, that they are necessarily syntactic
16:08 eli: samth: That's a very odd thing to say.  I don't know of anything more syntactic than ... names.
16:08 samth: this is a choice the current implementation makes, and i think we should make a different choice
16:08 samth: eli, right now, we have the ability to get names from procedures and other objects, and we can get the name from a struct type
16:08 eli: I'm not saying anything about whether it should or should not be done.  at least not now.
16:09 eli: What I *am* saying, is that the decision to not include field names in the type information is completely in line with the decisions to keep argument names out of closure values.
16:09 samth: are you saying that `object-name' is a "leak" in some sense?
16:09 DT``: eli, oh, so if I define a (struct foo (x y) #:prefab) in two different files, it will be the same struct?
16:09 samth: DT``, yes
16:09 eli: samth: Yes, isn't that obvious?
16:09 DT``: makes sense.
16:10 samth: eli, no, it was not obvious
16:10 eli: samth: You can see the leak in action every time someone does (define foo* (letrec ([foo (lambda ...)]) foo)) -- *just* to get the name right.
16:10 samth: since you seemed to find my suggestion bizzare
16:10 samth: whereas it's something Racket does currently
16:11 eli: I don't see where I contradicted myself -- having access to names is leaking; and `object-name' is unsurprisingly an instance of this.
16:11 eli: It's a kind of a necessary evil when it comes to debugging etc.
16:12 eli: Which of course you can claim is true for argument names.  (And indeed many had claimed that.)
16:12 samth: you did not explicitly contradict yourself -- you expressed skepticism about a feature, and analogized it to a feature we don't have, making me think that there was a distinction in your mind
16:12 eli: No, I said that the two missing features are extremely similar.
16:12 samth: anyway, there are lots of useful non-debugging applications of having struct field names, and none that i know of for argument names
16:13 samth: however, i think we understand each other now
16:13 eli: In any case, the real danger in that is that it tempts people to rely more and more on these names being more than they are.  For example, if you see #<procedure:foo> you might be tempted to assume that `foo' is bound to it.
16:13 eli: That's "obviously" not going to work, but it's very clear that it won't.
16:14 eli: With struct names the danger is even higher (ie, the many people getting utterly confused about why it's so damn difficult to print struct values in a readable way).
16:14 eli: And with field names it's even more fun than that.
16:14 offby1: I'm thinking ... hash table :)
16:14 samth: personally, i'd like to have (ref v name)
16:15 samth: that works for objects, dicts, and structs
16:15 eli: ...that's a hash table...
16:15 (join) MayDaniel
16:15 eli: All of that is from experience BTW -- CLOS makes it damn tempting to use syntactic values at runtime.
16:15 samth: no, it isn't, any more than vectors are hash tables because they are `dict?'
16:16 eli: What will (ref v blah) do when v is a struct with no blah field?
16:16 (quit) anRch: Quit: anRch
16:17 samth: raise an error
16:17 eli: Which kind of error?
16:17 samth: the same thing (dict-ref #(1 2 3) 'foo) does
16:17 eli: You mean a runtime error?
16:17 samth: yes
16:18 eli: Then you want a hash table.
16:18 samth: no, really, i don't
16:18 samth: i want generic operations
16:20 (join) jonrafkind
16:21 eli: Those generic operations are equivalent in all senses to a hash table.
16:21 eli: The kind of checks and errors, and the runtime overhead.
16:22 eli: You just want the hashes to be limited in having a fixed set of keys, with some of them marked as read-only etc; but it is still the same as a hash table.
16:22 eli: (And even the implementation is likely going to have one; perhaps implemented as an alist.)
16:25 (join) rudybot_
16:25 samth: why would the implementation need to be a hash table?
16:26 samth: i am suggesting that struct type values should contain a list of symbols in them
16:26 (join) shachaf_
16:26 samth: that's it
16:26 (join) em_
16:29 (quit) shachaf: Disconnected by services
16:29 (join) ghiu_
16:29 (nick) shachaf_ -> shachaf
16:29 eli: You need to do a _lookup_, making it equivalent to a hash.
16:29 (quit) ghiu: Ping timeout: 260 seconds
16:29 (quit) asumu: Ping timeout: 260 seconds
16:29 (quit) em: Ping timeout: 260 seconds
16:29 (quit) rudybot: Ping timeout: 260 seconds
16:29 (nick) ghiu_ -> ghiu
16:29 samth: a lookup where?
16:30 samth: is your claim that objects (in the `racket/class' sense) == hashtables?
16:31 (join) asumu
16:31 eli: A lookup in your generalized `ref'.
16:31 eli: And yes, same deal.
16:32 (join) jeapostrophe
16:33 samth: ok, then you just have a very expansive notion of hash table, and we aren't actually arguing about anything
16:34 (quit) jeapostrophe: Client Quit
16:36 (quit) bugQ: Ping timeout: 240 seconds
16:52 offby1: Gentlemen, you can't fight here.  This is the War Room.
16:53 (quit) Demosthenes: Quit: leaving
16:57 (nick) rudybot_ -> rudybot
17:03 (nick) em_ -> em
17:06 (quit) DT``: Ping timeout: 260 seconds
17:09 samth: offby1, best movie ever
17:12 (quit) em: Ping timeout: 244 seconds
17:13 offby1: I don't avoid film references, samth, but I do deny them my ... uh
17:14 (join) em
17:16 samth: offby1 wins the internet for today
17:21 (join) DT``
17:23 em: I dont know why im bouncing right now. and im not at home so i can't fix it.
17:24 em: is the idea of 'contracts' to make complication easier or better?
17:24 em: oops i mean compliation.
17:25 em: erg..
17:25 em: Compilation
17:33 (join) ckrailo
17:34 samth: em, mostly no
17:34 samth: contracts are run at runtime
17:40 ghiu: how can i escape html characters?
17:41 samth: in what?
17:41 ghiu: like < -> &lt;
17:42 ghiu: is there any library to get this?
17:42 samth: as in, to automatically do this?
17:42 samth: you can use scribble
17:44 samth: you can also use the `xml' collection, which might be more what you want
17:44 ghiu: scribble
17:44 ghiu: thanks
17:45 ghiu: i'll look there :)
17:47 (join) jeapostrophe
17:51 (quit) mceier: Quit: leaving
17:53 (quit) ghiu: Quit: ghiu
17:53 (quit) jeapostrophe: Quit: jeapostrophe
17:53 offby1: eli: I'm thinking (as I occasionally do) about rewriting rudybot's parser.  Does parser-tools/lex seem like a reasonable way to parse the stuff I get from the IRC server?
17:53 offby1: i.e., if it's an awful idea, holler
17:54 (join) RandalSchwartz
17:54 (quit) RandalSchwartz: Changing host
17:54 (join) RandalSchwartz
18:03 samth: offby1, the parser tools are used pretty widely, so i'd expect them to work well
18:03 offby1: it's not a question of "do they work well" -- I assume they do -- but rather "are they the appropriate tool for this particular job".
18:04 offby1: I suspect they might be overkill
18:04 offby1: but am not sure, having never used lex or yacc in anger
18:05 (part) RandalSchwartz
18:08 samth: if it's bigger than a regexp, i'd use the parser tools
18:08 offby1: sounds reasonable.
18:38 (join) mithos28
18:40 asumu: I used the parser tools to implement a toy stack-based language. It was overkill in a way but worked fine.
18:40 asumu: And would've been easier than most other options.
18:41 bremner_: isn't the point of stack based languages that they don't need parsers, just lexers?
18:42 bremner_: not that it matters, of course.
18:45 asumu: Hmm, okay I must be misremembering why I used the parser-tools then. Looking at the code, I see that I just used s-expressions.
18:46 asumu: Okay no, it was for some simple lambda calculi stuff.
18:47 asumu: I did it during the same summer so it blended together in my head.
18:50 (quit) DT``: Ping timeout: 240 seconds
18:54 (join) DT``
19:01 (join) jeapostrophe
19:05 (quit) jeapostrophe: Client Quit
19:06 (quit) MayDaniel: Read error: Connection reset by peer
19:07 (quit) mithos28: Quit: mithos28
19:17 (quit) askhader: Ping timeout: 258 seconds
19:17 (join) askhader
19:23 (quit) DT``: Ping timeout: 260 seconds
19:36 (join) DT``
19:36 (join) mithos28
19:41 (quit) mithos28: Client Quit
19:42 eli: offby1: Sounds to me like a definite overkill.
20:09 (quit) lucian: Remote host closed the connection
20:26 (quit) masm: Quit: Leaving.
20:37 (nick) samth -> samth_away
20:43 offby1: aha
20:52 (join) jeapostrophe
20:59 (join) mithos28
21:25 (quit) jeapostrophe: Quit: jeapostrophe
21:26 (quit) mithos28: Quit: mithos28
21:26 (join) mithos28
21:27 (quit) mithos28: Client Quit
21:29 (join) samth
21:31 (join) mithos28
21:33 (quit) mithos28: Client Quit
21:38 (join) mithos28
21:41 (join) realitygrill
22:29 (join) jeapostrophe
22:29 (quit) jeapostrophe: Client Quit
22:34 (join) PLT_Notify
22:34 PLT_Notify: racket: master Robby Findler * b43e956 (4 files in 3 dirs): Redex: added to-lw/stx ... - http://bit.ly/kOefoO
22:34 (part) PLT_Notify
23:12 (join) jeapostrophe
23:18 (quit) jeapostrophe: Quit: jeapostrophe
23:23 (quit) samth: Ping timeout: 255 seconds
23:25 (quit) qha: Ping timeout: 240 seconds
23:29 (join) qha
23:43 (quit) mithos28: Quit: mithos28
23:50 (join) dnolen
23:54 (join) jeapostrophe