#racket log 2013-05-04

02:50 philix_: how can I call a function right after the user hits Ctrl+C?

02:50 mithos28: Ctrl+c raises a break exception, you need to catch that
02:51           http://docs.racket-lang.org/reference/breakhandler.html?q=break#%28tech._break%29

02:51 rudybot_: http://tinyurl.com/clglh8x

03:00 philix_: Thank you!  (with-handlers ([exn:break? (lambda (x) (quit-server))]) (do-not-return))  WORKED really well!

14:27 eg0: packages seem way more complicated than they need to be, and the guide is almost completely unhelpful in explaining it
14:30      s/packages/modules

14:33 bremner: ah, modules.  Well, you don't need to use all of the features.

14:34 eg0: even if i wanted to, i couldnt because the guide explains them so poorly
14:35      i spent a few hours yesterday and a few more this morning wasting time on guide's ch 6 and 7, and now i've elected to just skip them entirey

14:41 soegaard: eg0: That'd difficult :-)   A file that begins with    #lang racket    defines a module.

14:42 eg0: why does racket not use namespaces like CL?
14:42      just out of curiosity

14:45 soegaard: I am not familar with CL namespaces. However the module and macro system in Racket solves a problem: The languages macros are written in and the file using the macros are different. The Racket module system use phases to solve this.
14:47           See for example Eli's answer on stackoverflow: http://stackoverflow.com/a/7073230/23567

14:47 eg0: cl namespaces are packagename:functionname
14:47      its that simple
14:47      no ambiguity, no problems

14:49 eli: CL packages are a half-assed hack.

14:49 soegaard: If you like the naming scheme, then you can use (require (prefix-in math: math)) to import the functions in math. Then use e.g. math:divides?

14:49 eli: They provide some convenience, but not enough to get actual guarantees from it.

14:49 eg0: cl packages take 30 seconds to explain and learn forever

14:49 eli: And they're simple name management tools, rather than Racket modules that also serve as tools for encapsulation, compilation, and linking.

14:50 eg0: none of which is explainer in the guide of course
14:50      explained*

14:51 eli: If you're looking for a comparison between CL namespaces and Racket modules, then obviously it doesn't belong in the guide.

14:51 eg0: im not
14:51      you've only been participating for half the exchange here, so i can understand your sentiment
14:52      but ive wasted a fair bit of time trying to comprehend something that was not explained well, and im asking why it is that way

14:52 eli: I'm not talking about any exchange, I'm just responding to the point of romanticizing CL packages because "they're simple".

14:52 eg0: cl's namespaces do what I expect them to, which is leaps and bounds ahead of racket's modules, as the guide is horridly vague
14:52      what do you care? thats my opinion

14:52 eli: It's something that often goes hand-in-hand with `defmacro' which is also touted as a good option because "it's simple".

14:53 eg0: racket isnt better or worse by my preference of namespaces to modules
14:53      and i could reiterate all of my points again

14:53 eli: To be clear, namespaces are not comparable to modules.

14:53 eg0: although hygienic macros are great

14:53 eli: ThereYouGo™.

14:53 eg0: thats apples to oranges though, and offtopic as well
14:54      you havent been listening
14:54      im not asking for comparisons
14:54      i do not understand modules
14:54      the guide poorly explains them
14:54      wat do

14:55 soegaard: Here is the minimal to know:

14:55 eli: If you have a technical question about modules, then ask it, but don't go raving about CL namespaces since it's not the same tool at all.

14:55 soegaard: 1. A file that begins with #lang racket defines a module
14:55           2. You can import functions into your module with (require modulename)
14:56           3. If two modules export two different values with the same name, you must use  (require (except-in modulename name)) in order not to import name (but import everything else)
14:56           4. Use (provide name) to export name from your module.
14:57           That's it.

14:57 eg0: alright
14:57      thanks soegaard i appreciate it

14:58 soegaard: 5. If you have a macro that uses an imported name use (require (for-syntax modulename)) to import it.

14:58 eli: An alternative for #3 is (require (prefix-in foo: some-module))
14:58      (One that has a more namespace-ish feel to it.)

14:58 eg0: right up my alley
14:58      apparently
15:00      heres a question, whats the consensus on the chapter about modules in the guide? i think it could be improved a great deal, and id like to know what others feel

15:01 eli: If you're just learning the language, then it's probably better to focus on just doing that, rather than trying to improve the guide.
15:01      (And that's not saying that the guide is perfect in any way.)

15:02 eg0: im trying to learn the language
15:02      i spend 4-5 hours every single day working through the guide, and working on a few hobby projects in racket
15:02      im a slow learner, admittedly, but im not exactly stupid

15:03 eli: So just ask questions (here or on the list) about things that are unclear.

15:04 soegaard: We are talking about this one?  http://docs.racket-lang.org/guide/module-basics.html?q=module

15:04 eg0: yes

15:06 soegaard: I see that the discussion jumps from modules to collections without explicitly motivate when to use (non-library) collections.

15:13 asumu: eg0: can you pinpoint anything in particular that is not explaiend well?

15:14 eg0: 6.2.1, that example doesnt work for me

15:14 asumu: Does it need more examples? Different prose?

15:14 eg0: working through the chapter
15:14      im going to make a list of all the stuff i dont understand, and then im gonna do something about it
15:16      when i have that example in my editor pane, and i try (require 'cake) in my repl pane, i get an unknown module error

15:17 asumu: Are you using DrRacket or some other editor?

15:18 eg0: drracket
15:18      i can do (require (submod "." cake)) and have access to that print-cake function
15:18      but not (require 'cake) like the guide says

15:19 soegaard: (Have to ask) Did you click run first?
15:19           Before you tried (require 'cake) in the repl.

15:19 eg0: yes
15:20      heres a question, is it working for you when you try it?

15:20 soegaard: Hmm. Doesn't work for me either.

15:20 eg0: ok
15:20      thats comforting
15:20      i am sane

15:20 soegaard: I expected it to do so.
15:20           If you pase (module cake …) into the repl it works.
15:21           pase -> paste

15:21 eg0: paste that whole example?

15:21 soegaard: yes

15:21 eg0: yeah, same here
15:21      it worked
15:22      but is it actually getting it from the module?

15:22 asumu: Oh, that's a submodule and enter!-related bug. I don't remember if it's been fixed. Submodules are a relatively new feature.

15:22 soegaard: Oh!
15:23           I got it wrong.
15:23           If you write (module cake racket …) without the #lang racket file it works fine.
15:23           And in the guide there is no #lang racket line.

15:23 eg0: negative

15:24 asumu: Oh, the 6.2.1 example. Yes, that is not supposed to have a #lang.

15:24 eg0: i just tried it without a #lang file, and its giving me the same unknown module error

15:24 soegaard: That works for me on 5.3.4.3

15:25 eg0: im on 5.3.3

15:26 soegaard: works in 5.3.2 as well
15:26           (I haven't got a 5.3.3 installed)

15:27 eg0: i mean, i dont know what to say
15:27      i have that exact example, with no #lang racket
15:27      and its the same error

15:28 soegaard: Try updating. http://pre.racket-lang.org/installers/

15:29 asumu: I'm pretty sure that should work in 5.3.3. I have a 5.3.3 installed and tried a similar example. So you have a module (module cake ...) at the top of the file?
15:29        And in the interaction pane doing (require 'cake)?

15:29 eg0: yes

15:30 asumu: Did you run the file before doing the `require` in the REPL?

15:30 eg0: yes
15:30      several different tries
15:31      upgrading...

15:35 mithos28: If I have an optional keyword in a macro what is the easiest way to match it, and then splice it/the empty sequence into another macro call

15:36 eg0: when its talking about new system links, where should I install them?

15:37 mithos28: I'm using syntax-parse, but cannot figure out how to go from (~optional #:kw) to something useful

15:37 asumu: mithos28: (~optional (~and #:kw kw-attr)) and then use the kw-attr attribute.

15:38 mithos28: asumu: but that is #f sometimes

15:38 soegaard: eg0: don't know, is there a suggestion?

15:38 mithos28: how to do I easily splice in the empty sequence in that case

15:38 asumu: mithos28: you can put a (if (attribute foo.kw-attr) #'(foo.kw-attr) #'()) and splice that

15:39 mithos28: Which is what I have, my question is how to do this easily

15:39 eg0: no, the default is to skip it, but i installed drracket through apt-get and it wont update from 5 3 3 to 5 3 4
15:39      if i dont install the links over the old ones
15:39      i just dont know where the old ones are
15:39      im gonna try usr/bin

15:40 asumu: mithos28: ?? or ?@ from syntax/parse/experimental/template might help, but it's experimental.

15:41 mithos28: asumu: Thanks i'll take a look. TR already has tons of deps on unstable so this shouldn't be an issue

15:41 soegaard: I would use http://pre.racket-lang.org/installers/   instead. The installer is very good.

15:42 eg0: it skipped them all, it says that non-links exist
15:42      i might have to remove drracket from apt get, ill do that some other time, i cant use that much data or my internet will get turned off
15:45      ok
15:45      i was able to start drracket from the new install directory
15:45      and that cake example still does not work
15:45      no #lang file
15:45      line*

15:49 soegaard: eg0: Can you make a screenshot and put it on imgur ?

15:49 asumu got struct's #:guard clause working in TR, but wonders how bad it is for the top-level typecheck pass to local-expand more stuff...

15:50 eg0: link inbound
15:51      http://imgur.com/m8zZMPs

15:53 asumu: Oh, ugh. That happens to me too. The difference is if you save the file somewhere.

15:53 eg0: it is saved

15:53 asumu: I wonder if that's a regression.

15:54 mithos28: Asumu, we should sync up as that is exactly what I'm working on

15:54 soegaard: eg0: I can't see you are doing anything wrong. Here is what I see on 5.3.4. http://imgur.com/OeC6FRC

15:56 asumu: eg0: if you do (require "cake.rkt") it should work. That is pretty annoying though.
15:57        mithos28: Oh, sorry I should've coordinated with you then. I'll put it up as a branch somewhere.

15:57 eg0: asumu, that works

15:58 mithos28: asumu: https://github.com/shekari/racket/tree/struct-properties, but literally a work in progress
15:58           Don't worry about it, I didn't tell anyone

16:00 eg0: the guide says that the example works in the repl because its not associated with a file name
16:01      so (require 'cake) and (require "cake.rkt") are two fundamentally different happenings

16:01 asumu: eg0: it should at least say something like "if you save it to a file, you need to use <blah>".

16:02 eg0: well

16:02 asumu: mithos28: https://github.com/takikawa/racket/commit/b6e7dfea1b3b1233210432760e19815b91fcdcd9

16:02 rudybot: http://tinyurl.com/czlwe7n

16:02 eg0: my mental gears are moving
16:03      you cant do (require 'cake) in a repl because racket doesnt know what cake is, only cake.rkt knows what cake is
16:03      right?

16:04 asumu: mithos28: ah, you handle struct properties though, that's nice. How do you type-check the self struct argument? (to say custom-write)

16:05 mithos28: By using the type name I know from the struct: form
16:06           Were you trying to support guards that change the arguments?

16:06 asumu: mithos28: no, are there such guards that ought to type-check?

16:06 mithos28: It would change the constructor type, which I thought was too complicated to support

16:07 asumu: I agree.

16:07 mithos28: Oh, I think I misread your original question. I changed struct: from ignore to ignore-some
16:07           which means that you can annotate experessions within its form to be typechecked
16:08           Thus no change from the typechecking standpoint

16:08 asumu: Actually you answered it, but I did want to know the answer to that question too. That would eliminate the need to do the local expansion I do.

16:09 mithos28: https://github.com/shekari/racket/commit/99203197e4805e27fe74b24b4e3a7068a5b2db07#L13R547

16:09 rudybot: http://tinyurl.com/cmc6d8m

16:09 asumu: I asked my question because I was thinking it'd be useful to have a (Struct-Property (Self) ...) type that bound a 'self'-type so you could support user-defined properties.

16:10 mithos28: Thats something I didn't even try to handle, I'm only handling the three cases I saw in the core code base
16:10           equal+hash, custom-write, and custom-print-quotable

16:11 asumu: Fair enough. The general extension could be done later anyway.
16:17        mithos28: yeah, the way you're doing the guard type-checking makes more sense. Less invasive.

17:27 eg0: so im trying to write to a file that does not exist and racket is giving me a hassle
17:28      jk was writing to path string and not the port, nvm

17:33 iceking: Does anyone know of an IRC like this where I can get some help with assembly? (sorry, googling didnt do me any good so I figure Id just ask)

17:34 offby1: iceking: not offhand, but you might try typing /list to get a list of channels on freenode.  Maybe one of 'em will be about assembly.

17:34 eg0: try #asm

17:34 offby1: You might also try looking for the _specific_ assembly language you're interested in -- x86, or ARM, or whatever.
17:34         or what eg0 said :)

17:34 iceking: thanks :]

17:34 eg0: im an irc wizard
17:34      its the one thing im good at

17:35 offby1: follow your bliss!

17:35 eg0: i am

17:35 rich: iceking: what level of assembly?

17:35 eg0: i am designing and building a successor to irc
17:36      rich, #asm handles all arches
17:36      they are quite friendly

17:37 rich: cool.
17:37       if it was is basic x86 assembly i might have been able to help him

17:37 eg0: oh

17:37 rich: s/was//

17:37 eg0: well hop in #asm, he's in there right now

17:38 rich: but either way, he's defniitely better asking the #asm guys

17:38 eg0: he has homework apparently
17:38      well yeah, its offtopic here
17:38      not unwelcome, just offtopic

17:38 rich: yeh

17:38 eg0: so how about those parens

17:39 rich: i assume read-line uses 'linefeed instead of 'any to remain compatible for old scripts ?
17:41       i had a list of numbers in a text file - apparently `man sum` is not what i thought it was :) anyway, i whipped up a racket script using in-lines, number->string and apply + - fantastic

17:43 eg0: why didnt you just store the list?
17:43      and then get the file's contents, and read-from-string
17:44      and then you have a lisp/scheme compatible list to use apply on

17:44 rich: the numbers were from an existing text file which i had to `cut` from

17:45 eg0: oh

17:45 rich: i don't know awk - and i'm trying to learn racket

17:45 eg0: sorry i misunderstood
17:45      i dont know either
17:45      i just pretend and hope nobody asks me to do anything

17:46 rich: eg0: do you know any lisp-like ?

17:46 eg0: i worked my way through most of gentle intro to symbolic computation before i switched to racket
17:47      and im currently working through chapter 8 of the racket guide

17:47 rich: if you like pattern matching, have you seen formica for use in racket?

17:48 eg0: ive heard of it, but havent tried it

17:48 rich: match is cool

17:48 eg0: and i have to go also :(

17:48 rich: match is racket

20:10 technomancy: is box-cas! new since 5.2.1?
20:11              the docs don't mention needing to require it first

20:31 zRecursive: rudybot: where is git repo of racket ?

20:31 rudybot: zRecursive: because I was free to commit & push to the work repo without worrying about having to keep a clean branch for merge.

20:32 asumu: zRecursive: https://github.com/plt/racket
20:33        technomancy: 5.3, see https://github.com/plt/racket/blob/master/doc/release-notes/racket/HISTORY.txt

20:33 rudybot: http://tinyurl.com/btonomq

20:35 zRecursive: asumu: thx

20:35 zRecursive cloning now ...

20:36 offby1: .oO("box-cas!"?)

20:37 bremner: next version has thai-kick-box-cas!

20:38 offby1: Hobos from Bahstan ride Bahx Cahs

20:56 zRecursive: build error: ../../../racket/gc2/../src/string.c:41:21: error: iconv.h: No such file or directory

21:06 offby1: zRecursive: you probably need to install "development libraries" for libiconv
21:06         if you're on Debian or Ubuntu, "sudo apt-get build-dep racket" might get you everything you need

21:06 zRecursive: offby1: i am using freebsd

21:07 offby1: alas, I don't know the formula for FreeBSD, but I bet it's roughly similar.

21:07 zRecursive: yeah, i am installing libiconv now

21:08 technomancy: asumu: gotcha; thanks
21:11              huh, I assumed that meant boxes themselves are fairly recent, but it sounds like that's not the case
21:12              I guess cas semantics aren't as important without OS-level threads, which are fairly new
21:13              oh huh; JIT on ARM is fairly new too; I guess I should upgrade

21:15 zRecursive: Anybody uses ARM desktop ?

21:16 offby1: *crickets*

21:16 zRecursive: rudybot: crickets

21:16 rudybot: zRecursive: error: crickets: undefined; <NEWLINE>  cannot reference an identifier before its definition <NEWLINE>   in module: 'program
21:18          zRecursive: that'll teach ya!

21:19 zRecursive: :)

21:19 zRecursive raco setup now ...

21:33 zRecursive: `raco setup ...` is so slow even i have `configure --with-doc=no ...` ?!

21:37 eg0: is there anything like multiple-values-list in racket? i have several values, and i would like to collect all of them in a list

21:38 bremner: I'm not sure I understand the question.

21:39 eg0: if i have a function that returns two values, how do I interact with both of them and not just the first one?

21:40 bremner: e.g. with let-values?

21:40 eg0: oh
21:40      oh no
21:40      thats assignment
21:41      like, say I had (values 4 5)
21:41      what can I use to get a list (4 5)?
21:41      (list (values 4 5)) doesnt work, and there doesnt seem to be a multiple-values-list like there is in CL

21:43 bremner: I see. I never use multiple values, but at least I understand the question now.
21:43          I guess you could make something with call-with-values

21:45 zRecursive: rudybot: (values 4 5)

21:45 rudybot: zRecursive: ; Value: 4
21:45          zRecursive: ; Value#2: 5

21:45 bremner: rudybot:  (call-with-values (lambda () (values 1 2)) list)

21:45 rudybot: bremner: your sandbox is ready
21:45          bremner: ; Value: '(1 2)

21:45 eg0: yeah
21:45      thats what I just tried
21:45      ok, that will be perfect
21:45      thanks bremner, i appreciate the help

21:45 bremner: today I learned ;)

21:46 eg0: me too
21:46      knowledge is power

21:51 Nisstyre: eg0: you can also do the reverse, ie..
21:51           rudybot: (apply values (list 1 2))

21:51 rudybot: Nisstyre: your sandbox is ready
21:51          Nisstyre: ; Value: 1
21:51          Nisstyre: ; Value#2: 2

21:51 eg0: i did know that one
21:51      apply is sweet

21:51 Nisstyre: apply takes a procedure and a list and applies the procedure to each element in the list as the arguments
21:51           yeah

21:52 eg0: rudybot: (apply values (call-with-values (lambda () (values 4 5)) list))

21:52 rudybot: eg0: your sandbox is ready
21:52          eg0: ; Value: 4
21:52          eg0: ; Value#2: 5

21:52 eg0: undoing
21:52      trollolo

21:52 Nisstyre: yeah

21:52 eg0: if i worked IT, id drop snippets like that all over the place

21:53 Nisstyre: that's generally considered bad form :P

21:53 eg0: my dad wrote cobol for 30 years and he says he did shit like that all the time so that he could secure his position
21:53      but thats only IT

21:53 Nisstyre: hahaha

21:53 eg0: youd get fired or slapped if you did that somewhere else

21:53 Nisstyre: eg0: http://thc.org/root/phun/unmaintain.html

21:54 eg0: omg this is gold

21:54 Nisstyre: that one is a classic

21:55 eg0: when I saw that go demo where they did hello world in japanese i thought about writing complete projects with foreign keyboards
21:57      this is the best thing ever, there's so many of these

22:19 asumu: technomancy: boxes are ancient. CAS was added primarily because it's useful for a concurrency library that's under development.
22:19        (ancient in that they've always been there, not that they're outdated or anything)