00:25 (join) jonrafkind
01:39 (join) Matrox
01:40 (quit) neilv: Quit: Leaving
01:40 (quit) Matrox: Client Quit
01:48 (quit) evhan: Read error: Connection reset by peer
02:06 (quit) shkk: 
02:10 (quit) jonrafkind: Ping timeout: 248 seconds
02:35 (quit) hellmage: Ping timeout: 260 seconds
02:35 (join) masm
02:42 (join) hellmage
03:04 (quit) Spewns: Quit: Leaving.
03:17 (join) evhan
04:17 (join) hanDerPeder
04:23 (quit) evhan: Ping timeout: 258 seconds
04:59 (quit) jao: Ping timeout: 245 seconds
05:39 (join) jao
05:46 Lajla: chandler, explain to me how continuation passing style may help a Scheme compiler.
05:46 Lajla subscribes to chandler's teachings
05:49 Lajla: From what I understand you get TCO and continuations for free and you seem to have no call stack / heap, but methinks to implement it in some intermediate language, that language needs to have some form of TCO on itself or it overflows.
05:55 (join) Spewns
06:06 (quit) Spewns: Quit: Leaving.
07:35 (part) hellmage
09:50 (join) b-man
09:53 (join) shkk
09:58 samth: Lajla, read "compiling with continuations", by appel
10:03 Lajla: samth, does this channel consider it unethical to leech it rather than pay for it?
10:04 Lajla: samth, I see it is intended for an SML compiler, that doesn't matter that much I take?
10:06 samth: Lajla, I wouldn't steal from Andrew Appel
10:07 samth: the relevant issues are mostly the same for SML and Scheme
10:07 Lajla: samth, why him in perticular?
10:09 samth: no, i don't mean that i wouldn't steal from him in particular
10:09 samth: i'm just pointing out that it's a particular person, who's done impressive research, who you would be stealing from
10:14 Lajla: Well, to be honest, I don't see that as very relevant to pay a man, more if that man has a reasonable living on its own. For instance, I would sooner give money to the murderer of my sister if he has little to eat rather than to Nelson Mandela, who seems to have enough, but this is for ##politics I guess.
10:14 samth: Lajla, you're a very strange person
10:16 Lajla: samth, you get no argument from me here I guess, I've found that my ideas and praeferences seldom collide with other people I meet.
10:21 stamourv: Lajla: as a starting point, if you're opposed to buying books, would be the paper "continuation passing style, closure passing style", also by Appel
10:21 stamourv: it is freely available
10:22 Lajla: stamourv, I'll check that too I guess, though I never said that I am opposed to buying books per se.
10:23 stamourv: but the book is definitely worth it
10:24 Lajla: stamourv, hmm, maybe, but I don't like reading on paper, it misses the essential property of controlling font size I guess.
10:25 Lajla: So maybe I'll just download it and pay him anyway via some mechanism.
10:28 Lajla: samth: "It has higher-order functions, meaning that a function can be passed as an
10:28 Lajla: argument and returned as the result of another function—as in Scheme, C,
10:28 Lajla: and Haskell, but not Pascal and Lisp."
10:28 Lajla: Oh fuck, sorry, it put endlines in there.
10:31 (quit) masm: Read error: Connection timed out
10:33 (join) masm
10:44 (join) Gwyth
10:58 (quit) masm: Read error: Connection timed out
11:12 (quit) hanDerPeder: Ping timeout: 276 seconds
11:17 (join) Spewns
11:45 (join) jonrafkind
12:02 (quit) jao: Ping timeout: 265 seconds
12:16 (join) hanDerPeder
12:45 (quit) shkk: 
12:49 (quit) hanDerPeder: Quit: hanDerPeder
12:59 (join) hanDerPeder
13:16 (quit) Spewns: Quit: Leaving.
13:43 (join) carleastlund
14:18 (join) anRch
14:26 (join) Spewns
14:51 Lajla: samth, reading this though seems to imply that the tactics might have some problem with a call-by-need/name model, is that correct?
15:13 (quit) Spewns: Quit: Leaving.
15:14 samth: Lajla, call-by-need would require a different compilation strategy
15:19 Lajla: samth, well, 'tis an interesting read regardless
15:28 chandler: Lajla: The classic explanation of how CPS is useful as a program representation for Scheme compilers is David Kranz's thesis describing the ORBIT compiler: http://repository.readscheme.org/ftp/papers/orbit-thesis.pdf
15:28 chandler: It's well-written and very readable.
15:31 tewk: /whois chandler
15:32 samth: tewk, i think you have an extra space there :)
15:32 Lajla: tewk, he is my lord and master since I subscribed to his tecahings
15:32 Lajla: chandler, this article starts with the bold claim that it's better than most of the other compilers already. =)
15:33 chandler: At the time that was written, that was a justifiable claim.
15:34 Lajla: chandler, hmm, Scheme was compiled worse than it could for a long time?
15:34 Lajla: chandler, so the CPS is really a major asset to compiling functional anguages?
15:35 chandler: I'm not sure how to parse your first question. Compilers in the wild are hardly as good as they theoretically could be.
15:37 chandler: CPS is a useful representation on which to define program optimizations. There are other representations that are widely used, such as the closely-related ANF and the SSA form.
15:37 Lajla: Well, my quaestions in this channel do have to traverse the nigh impenetrable sea of lajla -> chandler it seems.
15:38 Lajla: How efficient is Pascall by the way, in terms of C?
15:38 Lajla: Let's C C as c and express speed in terms of C as they do in physics.
15:41 chandler: Well, ISO Pascal is 98 pages long, and ISO C99 is 550 pages. Weight is the enemy of speed (as Colin Chapman said, "add lightness"). Therefore, Pascal is over 5 times faster than C.
15:43 Lajla: I suppose I sort of defeat my own point yeah when I always say that fundamentally speed is a property of c ompilers and not of languages.
15:43 Lajla: But that would mean that lambda calculus is pretty fast.
15:43 Lajla: And brainfuck
15:58 carleastlund: The lambda calculus is really fast if all you want to do is beta reductions.  Encoding other computations in it is where it gets painful.
15:58 (join) jao
16:00 clklein: But, come on, who does those other things?
16:01 Lajla: clklein, I thought you designed trendy men's wear in lambda calculus
16:01 carleastlund: Who knows, people do all kinds of crazy things.
16:02 clklein: like eta reductions?
16:02 Lajla: Clickwheelcalculus is even more minimal, you assign every turing computable function a natural number, all your programs are just a few billion clicks away!
16:03 clklein: Do you try to assign the more useful programs shorter numbers?
16:04 Lajla: clklein, sure, Mac OS X 10.5 is 0, HEllo World is 1.
16:05 (quit) anRch: Quit: anRch
16:06 clklein: Can we make Racket 6.0 be 2 so that we're almost done?
16:12 carleastlund: Let me slip in a request for "infinite Turing machine oracle" before all the natural numbers are taken.
16:14 Lajla: carleastlund, yes, natural numbers do have a tendency to run out and be finite after all.
16:15 Lajla: Especially in computers
16:15 carleastlund: Yeah, I wouldn't want to click so many times I wrapped around and got "negative Hello World".
16:16 (quit) paul_stansifer: Quit: Page closed
16:17 (join) paul_stansifer
16:19 Gwyth: "negative Hello World" is "Goodbye World", I presume...
16:30 carleastlund: a.k.a. "global thermonuclear war"
16:31 clklein: Let's give that program a big number, just to be safe.
16:33 Gwyth: hmm... a transfinite number might do...
16:34 clklein: Yea, you'd have to be pretty bent on world destruction to clickwheel all the way to omega.
16:41 Lajla: Pffah, you philanthropes
16:41 Lajla: Nothing wrong with a little misanthrope and evil schieming I always say.
17:34 (quit) hanDerPeder: Quit: hanDerPeder
17:39 (join) hanDerPeder
18:16 (quit) Gwyth: Ping timeout: 240 seconds
19:02 (join) shkk
19:44 (join) Gwyth
20:12 (quit) hanDerPeder: Quit: hanDerPeder
20:20 (quit) carleastlund: Quit: carleastlund
20:56 (join) hellmage
21:07 (quit) b-man: Remote host closed the connection
21:12 (join) Spewns
21:27 (quit) jonrafkind: Read error: Operation timed out
21:57 (quit) Spewns: Remote host closed the connection
22:02 (join) jonrafkind
22:25 (quit) jonrafkind: Ping timeout: 240 seconds
22:37 (join) jonrafkind