00:03 (join) pcavs
00:14 (quit) jao: Ping timeout: 246 seconds
00:24 (join) cdidd
00:28 (join) gciolli
00:32 (quit) jschuster: Quit: Coyote finally caught me
00:36 (quit) Fare: Ping timeout: 260 seconds
00:47 (join) jeapostrophe
00:51 (join) vu3rdd
00:51 (quit) vu3rdd: Changing host
00:51 (join) vu3rdd
01:07 (quit) pcavs: Quit: Leaving.
01:07 (join) perrth1
01:09 (quit) hash_table: Ping timeout: 246 seconds
01:09 (quit) getpwnam: Ping timeout: 246 seconds
01:10 (quit) perrth: Ping timeout: 264 seconds
01:11 (part) perrth1
01:17 (quit) gciolli: Quit: Leaving.
01:19 (quit) jeapostrophe: Ping timeout: 246 seconds
01:20 (quit) mceier: Quit: leaving
01:46 (join) gciolli
01:59 (quit) gciolli: Read error: Connection reset by peer
01:59 (join) gciolli
02:05 (quit) Nisstyre-laptop: Quit: Leaving
02:25 (join) hkBst
02:25 (quit) hkBst: Changing host
02:25 (join) hkBst
02:33 (quit) DT`: Read error: Connection reset by peer
02:34 (join) DT`
02:44 (join) djcoin
02:52 (quit) bjz: Quit: Leaving...
02:56 (quit) gciolli: Quit: Leaving.
02:57 (quit) ioa: Remote host closed the connection
03:14 (join) mceier
03:24 (quit) jonrafkind: Ping timeout: 246 seconds
03:56 (quit) mithos28: Quit: mithos28
04:11 (quit) Shviller: Read error: Connection reset by peer
04:12 (join) Shviller
04:16 (join) neilv
05:02 (join) noelw
05:26 (join) bjz
05:28 (join) masm
06:11 (join) bitonic
06:11 (quit) noelw: Quit: noelw
06:17 (quit) bitonic: Remote host closed the connection
06:18 (join) bitonic
06:21 (join) jesyspa
06:32 (join) noelw
06:35 (join) vkz_
06:38 (quit) vkz: Ping timeout: 260 seconds
06:38 (nick) vkz_ -> vkz
06:49 (join) MightyFoo
06:49 MightyFoo: will planet work through an http proxy serveR?
06:53 neilv: MightyFoo: not necessarily. it passes some info through custom http headers
06:54 neilv: the proxy might not propagate those headers
06:54 MightyFoo: oh
06:55 MightyFoo: if i could convince the proxy to pass the headers, then would planet use the proxy?
06:55 neilv: MightyFoo: i believe so
06:55 MightyFoo: spanky!
06:56 MightyFoo: should this be a matter of setting http_proxy (or HTTP_PROXY) in my environment?
06:56 MightyFoo: or is there something more involved to do?
06:59 neilv: hold on. i thought you were talking about transparent proxies. let me see if i can find the answer
07:01 neilv: you could try setting proxy in drracket, and see if planet's net/url use picks that up. http://www.mail-archive.com/users@racket-lang.org/msg00114.html
07:02 neilv: my next step would be to grep the source, or ask on email list
07:02 MightyFoo: i've seen that, but i'm not sure if (current-proxy-servers ...) is executed/set before the (require (planet ...)) is run.
07:02 MightyFoo: i'm never too sure of the "phases"
07:03 noelw: as i recall you need to modify the planet source to set the proxy
07:03 noelw: not sure if a better config experience was ever created
07:03 noelw: it has come up before
07:04 MightyFoo: i've not used a "transparent proxy" before... it might be time to learn about that
07:04 MightyFoo: neilv, noelw: thanks for the hints
07:04 noelw: no worries
07:05 neilv: another thing you could do is to make a tunnel, and tell it that's your planet server. but that's another kludge you shouldn't have to do
07:06 neilv: the right thing would be for planet to use http proxy when set in the platform-conventional way, and also to fix its protocol to not rely on custom headers
07:07 neilv: i wonder whether i'm going to regret trying to put gui and web-server in the same racket process
07:07 MightyFoo: iirc, subversion uses custom (or at least uncommon) headers... so the headers is not quite so worrisome to me
07:07 MightyFoo: neilv: was that a change of topic?
07:08 MightyFoo: interesting mix of concepts, though?
07:08 MightyFoo: s/?/.
07:08 neilv: yes, change of topic
07:19 neilv: gui seems to work if web-server is put in own thread. (i'm only doing "dispatch-rules", though, which shouldn't involve any black magic for continuations)
07:23 vu3rdd: anyone here doing Shriram's CS173 online course?
07:24 vu3rdd: Greatly enjoying it so far.
07:41 (join) HetQa9YjkK
07:47 (quit) bitonic: Ping timeout: 240 seconds
07:53 (quit) hkBst: Ping timeout: 240 seconds
07:54 (join) hkBst
08:05 (quit) hkBst: Ping timeout: 260 seconds
08:08 (quit) DGASAU: Remote host closed the connection
08:14 (join) hkBst
08:14 (quit) hkBst: Changing host
08:14 (join) hkBst
08:18 (quit) noelw: Read error: Operation timed out
08:31 (join) pcavs
08:32 (join) noelw
08:33 (quit) neilv: Ping timeout: 246 seconds
08:35 (quit) bjz: Quit: Bye!
08:35 (join) bjz
08:43 (join) jao
08:43 (quit) jao: Changing host
08:43 (join) jao
08:44 (quit) noelw: Quit: noelw
08:45 (join) Znudzon
08:48 (join) bitonic
08:50 (join) dnolen
08:51 (quit) masm: Read error: Connection reset by peer
08:57 (join) getpwnam
08:57 (join) hash_table
08:58 (quit) pcavs: Quit: Leaving.
09:01 (join) jeapostrophe
09:01 (quit) jeapostrophe: Changing host
09:01 (join) jeapostrophe
09:06 (quit) acarrico: Ping timeout: 256 seconds
09:08 (join) masm
09:16 (join) Fare
09:20 (join) noelw
09:22 (quit) noelw: Client Quit
09:23 (join) acarrico
09:27 (quit) hash_table: Ping timeout: 256 seconds
09:27 (quit) getpwnam: Ping timeout: 256 seconds
09:38 (join) noelw
09:38 (quit) noelw: Client Quit
09:43 (quit) hkBst: Read error: Connection reset by peer
09:43 (quit) Fare: Ping timeout: 246 seconds
09:44 (join) Nisstyre
09:46 (join) noelw
09:46 (join) jrslepak
09:49 (join) hkBst_
09:53 (quit) hkBst_: Read error: Connection reset by peer
09:54 (join) hkBst_
09:56 (join) Fare
09:57 (quit) jao: Remote host closed the connection
10:06 (quit) hkBst_: Ping timeout: 246 seconds
10:06 (join) RacketCommitBot
10:06 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/wTf9QA
10:06 RacketCommitBot: [racket/master] adjust define-judgment-form's typesetting so that it - Robby Findler
10:06 (part) RacketCommitBot
10:11 (join) hkBst_
10:12 (join) pcavs
10:13 (quit) jeapostrophe: Ping timeout: 246 seconds
10:15 (quit) hkBst_: Read error: Connection reset by peer
10:16 (join) hkBst_
10:21 (quit) hkBst_: Remote host closed the connection
10:25 (join) DT``
10:26 (join) hkBst__
10:28 (quit) DT`: Ping timeout: 252 seconds
10:30 (quit) hkBst__: Read error: Connection reset by peer
10:31 (quit) Nisstyre: Ping timeout: 260 seconds
10:31 (join) hkBst__
10:34 (join) dzhus
10:35 (join) chturne
10:35 (quit) vu3rdd: Ping timeout: 260 seconds
10:37 (quit) Znudzon: Remote host closed the connection
10:37 (nick) hkBst__ -> hkBst
10:40 (join) getpwnam
10:41 (join) hash_table
10:42 (join) jschuster
10:44 (join) Nisstyre
10:50 (join) hkBst
10:51 (join) neilv
10:51 (quit) neilv: Changing host
10:51 (join) neilv
10:55 (quit) noelw: Quit: noelw
10:58 (quit) hkBst: Ping timeout: 246 seconds
10:59 (quit) mceier: Quit: leaving
11:02 (join) noelw
11:04 (quit) lusory: Ping timeout: 246 seconds
11:05 (quit) noelw: Client Quit
11:09 (quit) dnolen: Ping timeout: 245 seconds
11:09 (join) hkBst
11:16 (join) RacketCommitBot
11:16 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/uj2viA
11:16 RacketCommitBot: [racket/master] Adjust linux version of Redex bitmap tests - Robby Findler
11:16 (part) RacketCommitBot
11:18 (join) yoklov
11:24 yoklov: are f32-vectors and flvectors the same?
11:25 yoklov: i notice that there are unsafe versions of flvector-* and f64-vector-*, but not f32-vector-*
11:25 (join) RPR
11:26 (quit) dzhus: Remote host closed the connection
11:26 stamourv: yoklov: flvectors hold doubles.
11:26 yoklov: oh
11:26 stamourv: Also, f64vectors have extra indirection.
11:26 stamourv: flvectors are usually faster.
11:26 stamourv: rudybot: doc flvector?
11:26 rudybot: stamourv: your sandbox is ready
11:26 rudybot: stamourv: no docs for a current binding, but provided by: racket/flonum
11:27 stamourv: rudybot: (require racket/flvector)
11:27 rudybot: stamourv: error: default-load-handler: cannot open input file: "/mnt/racket-5.2.900.1/collects/racket/flvector.rkt" (No such file or directory; errno=2)
11:27 stamourv: rudybot: (require racket/flonum)
11:27 rudybot: stamourv: Done.
11:27 stamourv: rudybot: doc flvector?
11:27 rudybot: stamourv: http://docs.racket-lang.org/reference/flonums.html#(def._((lib._racket%2Fflonum..rkt)._flvector~3f))
11:27 stamourv: yoklov: ^ Scroll up a bit from `flvector?', there's a comparison between the 2.
11:27 yoklov: i see
11:30 yoklov: do, for example, gl-double-vectors have the same indirection?
11:32 yoklov: it seems like they wouldnt, given that they are cvectors, but also they are always safe...
11:33 stamourv: Not sure. I haven't worked with them.
11:33 stamourv: What are you trying to do?
11:33 (join) ssbr_
11:35 (join) ssbr__
11:35 (quit) hkBst: Quit: Konversation terminated!
11:38 (quit) ssbr_: Ping timeout: 240 seconds
11:41 (join) dalob
11:42 dalob: is there anyway to directly create a duration for use with add-duration (http://docs.racket-lang.org/srfi-std/srfi-19.html?q=date#)? i can't seem to figure it out
11:43 yoklov: oh, i'm trying to make a small vector/matrix math library for graphics (but not really only for opengl, though compatibility would be nice)
11:44 stamourv: yoklov: Have you seen Neil Toronto's upcoming math library?
11:44 stamourv: https://github.com/ntoronto/racket
11:45 yoklov: i have not
11:45 stamourv: It may do what you want.
11:45 stamourv: (and, IIRC, it uses flvectors.)
11:47 stamourv: dalob: I don't know.
11:47 stamourv: racket/date may have what you want.
11:47 stamourv: rudybot: doc racket/date
11:47 rudybot: stamourv: not found in any library's documentation: racket/date
11:47 stamourv: rudybot: doc date
11:47 rudybot: stamourv: http://docs.racket-lang.org/reference/time.html#(def._((lib._racket%2Fprivate%2Fbase..rkt)._date))
11:48 dalob: i don't think it does actually. i'm trying to add some number of hours to a timestamp without having to do a date->seconds/seconds->date but i think that may be the best option
11:48 (join) soegaard
11:51 (quit) MightyFoo: Quit: WeeChat 0.3.8
11:52 soegaard: yoklov: I'd love some feed on the matrix library. The library provides matries over general Racket numbers and not over Reals.
11:52 (quit) dalob: Quit: Page closed
11:52 soegaard: If/when Typed Racket supports units, it will be easy to provide specialized versions.
11:54 (quit) neilv: Ping timeout: 260 seconds
11:54 (join) lusory
11:55 yoklov: it seems nice, but it looks like wouldn't be all that much use for graphics, primarially because of most of them being arbitrary dimensions (except for matrix-2d?) and using the generic numeric ops, both of which have a non-trivial overhead
11:56 yoklov: also, from what i can see it doesn't let you provide a destination matrix to avoid allocations, which is ugly and non-functional, but avoiding allocations when possible is… definitely useful sometimes.
11:57 soegaard: True.
11:57 yoklov: it's definitely very cool for generic matrix/vector math though, and maybe i'm underestimating the perf. increase that typed racket provides
11:58 soegaard: If you are after an old-school library check Noel Welsh's bindings for … what was it.
11:58 soegaard: https://github.com/noelwelsh/mzgsl
11:58 (join) ssbr_
11:59 yoklov: oh, i will
11:59 yoklov: thanks
11:59 yoklov: gotta run now though
11:59 (quit) yoklov: Quit: bye!
12:02 (quit) ssbr__: Ping timeout: 246 seconds
12:08 soegaard: LAPACK bindings would be *really* nice to have. Maybe an idea for the hackaton?
12:08 soegaard: http://www.netlib.org/lapack/
12:10 (join) mithos28
12:12 (quit) ssbr_: Ping timeout: 264 seconds
12:12 stamourv: soegaard: Sounds good. Can you add it to the wiki page?
12:12 soegaard: Done.
12:12 stamourv: Great, thanks!
12:13 stamourv: Are you coming to Racketcon?
12:13 soegaard: Sadly no.
12:14 soegaard: Will there be videos this year?
12:14 (join) MayDaniel
12:14 stamourv: I think so. But I'm not sure. samth would know.
12:15 (join) jeapostrophe
12:15 (quit) jeapostrophe: Changing host
12:15 (join) jeapostrophe
12:15 samth: soegaard: we are planning to do something, video-wise
12:15 soegaard: great
12:15 mithos28: Is there going to be a live feed?
12:15 samth: mithos28: no
12:15 samth: although maybe i'll try the google hangout trick
12:15 mithos28: That was going to be my suggestion
12:16 mithos28: Hangouts on air should make it about as painless as doing normally VCing
12:16 soegaard: That would be ideal. Another option is justin.tv
12:22 (quit) djcoin: Quit: WeeChat 0.3.7
12:28 (join) anRch
12:29 (quit) pcavs: Quit: Leaving.
12:29 (join) pcavs
12:50 soegaard: samth: Btw - If you decide to record both speaker and slides, then I can recommend ScreenFlow for the editing. It is really easy to use.
12:51 (quit) mithos28: Quit: mithos28
12:55 (join) jonrafkind
13:01 (quit) bitonic: Ping timeout: 260 seconds
13:05 asumu: Woo, Redex's new judgment form rendering is nice.
13:18 (join) nilyaK
13:24 (join) dyoo
13:30 (quit) anRch: Quit: anRch
13:36 (join) mceier
13:42 (quit) nilyaK: Quit: Leaving.
13:43 (join) RacketCommitBot
13:43 RacketCommitBot: [racket] plt pushed 2 new commits to master: http://git.io/InSDMw
13:43 RacketCommitBot: [racket/master] Handle contract generation for recursive class types. - Sam Tobin-Hochstadt
13:43 RacketCommitBot: [racket/master] Add `Frame%` and `message-box` types. - Sam Tobin-Hochstadt
13:43 (part) RacketCommitBot
13:47 (join) jao
13:47 (quit) jao: Changing host
13:47 (join) jao
13:49 SamB_MacG5: Sigh.  Isn't anyone going to merge https://github.com/plt/racket/pull/148 ?
13:49 asumu: SamB_MacG5: I'll remind eli of it if I see him around.
13:50 asumu: But it generally takes us time to merge things.
13:50 asumu: Since we need to actually review the commits.
13:51 (quit) jonrafkind: Ping timeout: 260 seconds
13:52 SamB_MacG5: point
13:52 SamB_MacG5: any feedback would be appreciated, really
13:53 (join) danl-ndi
13:53 danl-ndi: can racket do anything like a .jar for it's bytecode files?
13:54 danl-ndi: (out app 'stat's and loads thousands of .zo files. it makes for terrible file i/o performance)
13:54 danl-ndi: (out => our)
13:54 SamB_MacG5: hmm, not a bad idea
13:55 danl-ndi: Mr Flatt setup up separate dirs for the .zos recently. It seems like this'd be a natural next step.
14:00 samth: danl-ndi: there isn't something like that now
14:00 samth: but i imagine that it would be reasonable to add
14:00 samth: you should mention this on the mailing list
14:01 samth: since matthew (the one who'd implement this) doesn't read irc
14:01 danl-ndi: samth: thanks, I just did...
14:01 (join) RacketCommitBot
14:01 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/IAVucA
14:01 RacketCommitBot: [racket/master] Small typo fix - Danny Yoo
14:01 (part) RacketCommitBot
14:08 (quit) jamessan: Ping timeout: 245 seconds
14:10 (join) jonrafkind
14:12 (join) nilyaK
14:12 (join) jamessan
14:12 stamourv: danl-ndi: Do .plt files do what you want?
14:19 danl-ndi: stamourv: seems like they should!  is there any way to ".plt" up standard modules in the stock distribution?  more or less transparently?
14:21 danl-ndi: note that I'm interested in the bytecode files. I'd prefer to avoid the timestamp comparisons and only go for them
14:24 danl-ndi: stamourv: and i'm interested in handling .plt files directly -- putting them under source control to distribute them in our production environment for example
14:25 bremner: I seem to recall .plt files containing source which is recompiled on install.
14:26 jonrafkind: im sure its possible to make a compile-handler that can load files from zip's
14:26 danl-ndi: jonrafkind: that'd be a great step.. might be enough in fact.
14:27 jonrafkind: well zip support in racket is not amazing..
14:28 jonrafkind: and judging from the zip C libraries I've seen its quite an effort to write a native racket library
14:29 (join) h11r
14:30 asumu: Actually we have a zip library on PLaneT... and Jens and I put some work into modernizing a bit to put it in the file collection, but have gotten stalled.
14:30 asumu: Probably could just make an FFI binding though.
14:30 jonrafkind: gunzip.rkt was directly translated from C to racket :p
14:31 jonrafkind: an FFI binding to what? not libz..
14:31 jonrafkind: i mean gunzip.rkt == libz
14:31 asumu: Oh, makes sense. And I have no idea, I don't know what zip libraries there are.
14:31 jonrafkind: what you want is a library that can open zip files and get a list of files and extract a single file from them
14:31 asumu: lib7zip?
14:32 jonrafkind: I use libminizip for zip files
14:32 danl-ndi: needs to extract stat info in order to give enough info to the compilation manager, right?
14:33 jonrafkind: yea, or you can just use the stat info from the zip itself I guess
14:33 jonrafkind: but anyway zip files store stat info for each file
14:34 h11r: Hi. I have a very quick question. I was reading the racket documentation of the build in class system because I wondered if it is possible to add methods to an existing class without changing the defining expression of the class and without inheriting from it? Like it is possible in CLOS like systems. As far as I understood the documentation this is not possible, or am I wrong?
14:34 (quit) ivan\: Quit: ERC Version 5.3 (IRC client for Emacs)
14:35 jonrafkind: h11r, its not possible
14:35 asumu: h11r: This is not possible in Racket directly as in CLOS. There are patterns using units & classes that get you the same thing though.
14:35 asumu: Also, this is by design.
14:36 h11r: ah, thanks for the answer. Where can I find more information about this patterns?
14:37 asumu: h11r: http://www.cs.utah.edu/plt/publications/icfp98-ff/paper.shtml
14:37 h11r: asumu: Do you know the reason for the design decision?
14:37 m4burns: eli: ping
14:37 asumu: Extending classes post-facto means you end up doing mutation or something non-local behind your back. It makes it harder to reason about your code in a modular way.
14:38 danl-ndi: procmon.exe shows 8504 file operations on paths in racket/collects during startup
14:38 h11r: asumu: Thanks a lot for the link. This is exactly what I was looking for.
14:38 asumu: NP. :)
14:39 (join) ivan\
14:40 h11r: I'll go ahead and read it before bothering you any further :) . Thanks again for the quick answers. Bye.
14:48 (join) bitonic
14:52 (quit) nilyaK: Ping timeout: 260 seconds
14:55 (quit) masm: Ping timeout: 240 seconds
15:02 (join) uselpa
15:10 (join) nilyaK
15:10 (quit) acarrico: Ping timeout: 245 seconds
15:11 (quit) uselpa: Remote host closed the connection
15:11 (join) neilv
15:11 (join) ssbr_
15:11 (quit) neilv: Changing host
15:11 (join) neilv
15:11 (quit) ssbr_: Changing host
15:11 (join) ssbr_
15:11 (quit) jamessan: Ping timeout: 245 seconds
15:15 (join) jamessan
15:15 (nick) nilyaK -> `Kaylin
15:18 (join) acarrico
15:21 SamB_MacG5: danl-ndi: no wonder you wanted this; you're on Windows!
15:26 (quit) bitonic: Ping timeout: 246 seconds
15:27 SamB_MacG5: (Opening/statting files is not well optimized on Windows, or at least not in any of the default filesystems; there was an NT kernel hacker trying to figure out how to keep it from bottlenecking his kernel compilation on #emacs a while back ...)
15:29 (join) masm
15:31 SamB_MacG5: (For some reason, "use Debian" didn't seem to satisfy him ;-)
15:39 bremner: cross compilation is a pain.
15:41 (quit) ssbr_: Ping timeout: 246 seconds
15:42 SamB_MacG5: I think the problem was more to do with policy frowning on that sort of thing; possibly he wished to come up with something he would be allowed to share with coworkers.  (It seems that suggesting open-source tools to others is a no-no at MS...)
15:44 (quit) chturne: Quit: Leaving
15:49 SamB_MacG5: (Obviously, the kernel is intended to be built with MSC (aka cl.exe), and it should come as no surprise that it effectively *must* be built with MSC, or that MSC is only available as win32 (and maybe win64?) binaries these days, even to kernel hackers ... so cross-compiling doesn't really come up as an issue, because the compiler doesn't need to know it's not running on Windows in the first place
15:49 SamB_MacG5: ;-)
15:57 (quit) cdidd: Read error: Connection reset by peer
15:58 SamB_MacG5 wonders how MzRacket can load startup.rktl on its own
16:00 SamB_MacG5 realizes that it's only slightly less hard to load whatever comes next, because startup.rktl doesn't exactly contain a compiler ...
16:06 samth: SamB_MacG5: startup.rktl is compiled ahead of time into startup.inc
16:06 samth: which is a static c declaration of a big array with the relevant bytecode in it
16:07 SamB_MacG5: er, I think you've forgotten something
16:07 samth: i don't think so
16:07 SamB_MacG5: startup.inc appears to contain a bunch of scheme/racket code
16:07 samth: yes
16:07 samth: that's correct
16:07 SamB_MacG5: perhaps you meant cstartup.inc?
16:08 samth: yes
16:08 samth: oh, actually startup.inc is C code too
16:08 SamB_MacG5: yes, I know
16:09 samth: then i don't understand your original question
16:12 SamB_MacG5: well, it appears to contain little more than the (presumed) macro EVAL_ONE_STR applied to literal C strings containing minified versions of each of the modules in startup.rktl ...
16:12 samth: right
16:13 (quit) acarrico: Ping timeout: 255 seconds
16:13 samth: which then get byte-compiled to produce the result seen in cstartup.inc
16:13 SamB_MacG5: I was just pondering how the core can actually understand scheme/racket code ;-)
16:13 samth: SamB_MacG5: there's an interpreter and jit compiler for the byte code
16:13 samth: and a compiler that takes source and produces byte code
16:14 (join) yoklov
16:15 SamB_MacG5: so what happens when you set USE_COMPILED_STARTUP to 0?
16:16 samth: it uses the source
16:16 samth: in startup.rktl
16:18 SamB_MacG5: hmm, looks like builtin.c includes either cstartup.inc or startup.inc, depending on whether or not USE_COMPILED_STARTUP is enabled
16:20 (quit) RPR: Quit: Leaving
16:25 (quit) pcavs: Quit: Leaving.
16:26 (join) pcavs
16:26 (quit) pcavs: Client Quit
16:29 (join) pcavs
16:29 (join) acarrico
16:30 (join) uselpa
16:32 (quit) pcavs: Client Quit
16:33 h11r: asumu: Finished the paper and it was very interesting, so thanks again for the tip. I wasn't aware of the possibility offered by the unit system to link the units externally. But this kind of dependency injection is a neat way to solve the problem (and certainly much more elegant than Spring ;) in the Java world )
16:33 (join) RacketCommitBot
16:33 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/62lCHQ
16:33 RacketCommitBot: [racket/master] [Distributed Places] use racket -lm to launch new racket nodes instead of -tm - Kevin Tew
16:33 (part) RacketCommitBot
16:34 (join) pcavs
16:43 (quit) bjz: Quit: Leaving...
16:49 (quit) h11r: Ping timeout: 246 seconds
16:54 (quit) jeapostrophe: Ping timeout: 264 seconds
17:05 (quit) hash_table: Ping timeout: 248 seconds
17:05 (quit) getpwnam: Ping timeout: 248 seconds
17:06 (quit) yoklov: Quit: bye!
17:10 (quit) uselpa: Remote host closed the connection
17:11 (join) jeapostrophe
17:11 (quit) jeapostrophe: Changing host
17:11 (join) jeapostrophe
17:13 (join) DanBurton
17:17 danl-ndi: so our startup his about 8000 files in collects and takes only 2 seconds on a local drive
17:17 danl-ndi: and it takes 30 seconds on a network filesystem
17:18 danl-ndi: his => hits
17:18 danl-ndi: windows seems to perform quite well at this, at least for the local drive option
17:18 danl-ndi: we also rely on linux for the same thing, but my linux-fu to measure this stuff isn't as good
17:20 SamB_MacG5: yeah, I suppose Windows doesn't do all *that* badly
17:21 SamB_MacG5: not like process startup in general ...
17:22 SamB_MacG5: (and I've seen much worse timing with NFS!)
17:23 samth: on my installation, racket -e '(displayln hi)' `stat`s 2000+ files in .26 seconds
17:24 danl-ndi: local drive?
17:24 samth: yes
17:24 SamB_MacG5: danl-ndi: how long does it take to run true.exe?
17:25 danl-ndi: that's roughly comparable to what I'm seeing.  I went over my numbers on the mailing list...
17:25 samth: i've long thought we need to reduce the IO we do at startup
17:26 samth: I think IO is the big reason that the racket repl start 10x slower than python an 100x (!) slower than v8
17:27 danl-ndi: samth: it seems like a lot of what is going on during require expansion belongs more to a compilation phase
17:27 danl-ndi: SamB_MacG5: I don't have that
17:27 samth: danl-ndi: what happens at startup is more just linking
17:28 SamB_MacG5: danl-ndi: or any other .exe that does essentially nothing
17:28 samth: finding the location of various libraries, and whether they're zos or source or .so etc
17:28 danl-ndi: samth: the `stat`ing and comparing of rkt and dll and zo files and their containing directories
17:29 samth: danl-ndi: i think that ld.so on a linux system does roughly the same thing
17:29 SamB_MacG5: Python has to do some of this too, but it doesn't have anything comparable to 'racket or 'racket/base ...
17:29 samth: it's just that there are many more files
17:29 samth: right
17:29 samth: that's why the python repl starts up in 20 msec
17:30 danl-ndi: does python consider py vs pyc files?
17:30 danl-ndi: (during repl startup?)
17:30 samth: i assume so
17:30 danl-ndi: in any case I think the model, somewhat embarassingly, is Java in this case
17:30 danl-ndi: (or C#)
17:31 danl-ndi: python is not a good model IMHO
17:31 SamB_MacG5: Python, in fact, typically has to stat many different filenames before it actually finds one for a given module name ...
17:31 SamB_MacG5: Python does support loading from .zip files, though
17:33 samth: yeah, if i look through the strace output for python startup of the repl, it looks for os.py/os.pyc in a few places
17:33 samth: but that's about it for source files
17:33 (quit) neilv: Quit: Leaving
17:33 (join) anRch
17:33 (join) bitonic
17:33 danl-ndi: racket modules incorporate paths, like in java
17:33 SamB_MacG5: samth: and for some .so/.dll files, right?
17:34 danl-ndi: python is not a model of performance. it's more of a warning.
17:34 SamB_MacG5: danl-ndi: Python module names pretty much do, too, though it can be a bit ambiguous whether they are meant as relative or not ...
17:35 danl-ndi: java isn't a whole lot better, but it does have a fairly good packaging & module distribution setup
17:35 danl-ndi: I think if my app were in python it'd be a wholesale disaster
17:35 samth: danl-ndi: i agree that python isn't much of a model in most things, but i am envious of the repl startup speed
17:37 SamB_MacG5: as long as you wouldn't have to actually write any Java code ...
17:38 danl-ndi: in our environment python startup is much worse than racket startup, for what we import
17:38 danl-ndi: mostly, now that I think about it, because our python install is on the network
17:38 SamB_MacG5: danl-ndi: probably it would be faster if you had stdlib in a zip
17:38 jonrafkind: dont you cache stuff?
17:39 danl-ndi: jonrafkind: nope. not even a little.
17:39 jonrafkind: rsync -avP all.the.things:/stuff  .; ./run, no?
17:39 danl-ndi: our python setup isn't very production oriented... it needs a lot of work
17:40 danl-ndi: jonrafkind: we do have a lot of awkward distribution problems. we use rsync, perforce, git, and network drives
17:41 (quit) bitonic: Ping timeout: 260 seconds
17:41 danl-ndi: jonrafkind: plus windows on every desktop and a set of linux compile servers. some stuff linux, some windows, some cygwin.
17:42 (quit) stamourv: Read error: Connection reset by peer
17:42 (join) stamourv
17:43 (quit) pcavs: Quit: Leaving.
17:46 danl-ndi: we have a super-fast file server and 1-10 gig network connections, so throughput is fantastic. but the stat operations just kill us.
17:46 (join) bitonic
17:50 (join) RacketCommitBot
17:50 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/7yCvNA
17:50 RacketCommitBot: [racket/master] Recommend float-specific operations in case of unexpected complex. - Vincent St-Amour
17:50 (part) RacketCommitBot
17:54 samth: danl-ndi: i wonder how much that would be improved by just eliminating the check for .so versions
17:55 (quit) HetQa9YjkK: Quit: Page closed
17:58 SamB_MacG5: danl-ndi: check out http://docs.python.org/whatsnew/2.3.html#pep-273-importing-modules-from-zip-archives
17:58 rudybot: http://tinyurl.com/8hpf6xe
17:58 SamB_MacG5: and look at your sys.path
17:59 (join) Guest20130
18:02 (join) getpwnam
18:02 danl-ndi: samth: looks like very little time is spent on dlls (under windows at least)
18:03 (join) hash_table
18:05 danl-ndi: SamB_MacG5: huh... that seems promising.
18:06 danl-ndi: samth: in my example, for each module there's a dll check, and it takes about 0.00001 seconds to determine the dll doesn't exist
18:10 (join) bjz
18:11 (quit) jeapostrophe: Ping timeout: 260 seconds
18:14 SamB_MacG5: danl-ndi: py2exe might also be of interest
18:18 (quit) anRch: Quit: anRch
18:19 (quit) MayDaniel: Read error: Connection reset by peer
18:24 (quit) jesyspa: Quit: leaving
18:29 (quit) DanBurton: Quit: leaving
18:43 (join) gciolli
19:12 (quit) jonrafkind: Ping timeout: 245 seconds
19:12 (quit) bitonic: Ping timeout: 260 seconds
19:13 (join) jeapostrophe
19:13 (quit) jeapostrophe: Changing host
19:13 (join) jeapostrophe
19:23 (quit) dyoo: Quit: dyoo
19:41 (quit) gciolli: Ping timeout: 240 seconds
19:57 (quit) bjz: Quit: Leaving...
19:58 (join) gciolli
20:14 (quit) masm: Quit: Leaving.
20:37 (join) bjz
20:44 (quit) hash_table: Ping timeout: 255 seconds
20:44 (quit) getpwnam: Ping timeout: 255 seconds
20:53 (join) hash_table
20:55 (join) jonathansizz
20:55 (join) getpwnam
21:18 SamB_MacG5: intriguing ... geiser shows quite good help for parameterize!
21:21 (quit) gciolli: Ping timeout: 246 seconds
21:22 SamB_MacG5: soo ... how come the debugger only supports stepping in one direction?
21:24 (join) jesyspa
21:25 (quit) jeapostrophe: Ping timeout: 260 seconds
21:30 (quit) Fare: Quit: Leaving
21:31 (join) ynniv
21:40 (join) kidd
21:40 kidd: hey can anyone help me with an animation im working on
21:41 kidd: is anyone online?
21:44 (join) pcavs
21:47 (quit) jonathansizz: Ping timeout: 256 seconds
21:47 (quit) hash_table: Ping timeout: 252 seconds
21:47 (quit) getpwnam: Ping timeout: 256 seconds
21:48 kidd: hello
21:48 kidd: anyone help me out wit racket
21:48 (quit) kidd: Quit: Page closed
21:49 (join) getpwnam
21:50 (join) hash_table
21:56 (quit) pcavs: Quit: Leaving.
22:06 (join) RacketCommitBot
22:06 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/DZUmRw
22:06 RacketCommitBot: [racket/master] win32: fix combo-field% click handling - Matthew Flatt
22:06 (part) RacketCommitBot
22:13 (join) pcavs
22:17 (join) dyoo
22:42 (quit) dyoo: Quit: dyoo
23:14 (join) RacketCommitBot
23:14 RacketCommitBot: [racket] plt pushed 2 new commits to master: http://git.io/c5JHRA
23:14 RacketCommitBot: [racket/master] fix `vector-append' error reporting - Matthew Flatt
23:14 RacketCommitBot: [racket/master] racket/gui win32: fix drag-and-drop onto frame - Matthew Flatt
23:14 (part) RacketCommitBot
23:14 (join) mithos28
23:19 (quit) Shviller: Ping timeout: 240 seconds
23:20 (join) Shviller
23:41 (join) yoklov
23:42 yoklov: does drracket scroll very slowly for anybody else?
23:46 (join) dnolen
23:51 (join) jonrafkind
23:54 (join) vu3rdd
23:54 (quit) vu3rdd: Changing host
23:54 (join) vu3rdd