00:04 mithos28: is there a way to write a function ctype where the output ctype depends on the actual arguments when the function is called?
00:05 mithos28: i.e. (_fun (a b) :: _int _int -> (if a _int _uint))? (note this does not work since a is not defined in the output-type expression
00:06 (join) realitygrill_
00:06 (quit) neilv: Ping timeout: 260 seconds
00:07 (quit) realitygrill: Ping timeout: 260 seconds
00:07 (nick) realitygrill_ -> realitygrill
00:12 (join) jeapostrophe
00:14 eli: mithos28: No, if you can, the easiest thing would be to use some generic type that fits all return values, then use a second `->' and convert it to the desired one there.
00:14 eli: mithos28: Um, s/No,/No./
00:21 mithos28: thanks, That will work, just will be obnoxious.
00:26 (quit) jeapostrophe: Quit: jeapostrophe
00:30 eli: mithos28: Well, the thing is that libffi requires known types when the glue function is made up, and not at the time that the function is called.
00:35 (join) jonrafkind
00:38 mithos28: ok. I think I am just going to use a macro to examine the type and output the right wrapper after the second ->, thus ensuring that I don't mismatch the type and wrapper.
00:40 mithos28: Is there any documentation on all of the necessary measures to ensure safety when using the ffi?
00:41 (quit) flying_rhino: Ping timeout: 276 seconds
00:43 (join) flying_rhino
01:01 (join) realitygrill_
01:02 (quit) realitygrill: Ping timeout: 260 seconds
01:02 (nick) realitygrill_ -> realitygrill
01:09 eli: mithos28: Re macro -- I'm not sure that this will be the right way to deal with it since FFI types are runtime values, so you might fall into weird traps.
01:10 eli: And re safety, I'm not following that -- the ffi is inherently unsafe...  So all documentation for writing C code would apply...
01:11 mithos28: for the macro, in my use case it will just be identifiers not arbitrary expressions so I can add the extra information on to the identifier
01:14 mithos28: as with regards to the ffi, disciplined use will make it safe, and i was wondering if there was documentation on the common pitfalls (gc, breaks, killing threads)
01:16 (join) realitygrill_
01:17 (quit) otzi_: Ping timeout: 240 seconds
01:19 (quit) realitygrill: Ping timeout: 258 seconds
01:19 (nick) realitygrill_ -> realitygrill
01:21 (join) realitygrill_
01:24 eli: mithos28: Yes, those are the pitfalls...  IOW, I think it's "everything you'd suspect".  But feel free to improve the foreign docs -- they need a lot of care.
01:24 (quit) realitygrill: Ping timeout: 260 seconds
01:24 (nick) realitygrill_ -> realitygrill
01:28 (part) haruki_zaemon: "Laterz"
01:46 (quit) realitygrill: Ping timeout: 276 seconds
01:46 (join) realitygrill
01:56 (join) neilv
02:01 (quit) jonrafkind: Ping timeout: 258 seconds
02:17 (quit) torche: Quit: torche
02:24 (join) hkBst
02:38 (quit) dnolen: Quit: dnolen
02:41 (quit) realitygrill: Quit: realitygrill
03:00 (join) Burlingk
03:16 (quit) flying_rhino: 
03:16 (join) flying_rhino
03:16 (quit) flying_rhino: Client Quit
03:17 (join) flying_rhino
03:17 flying_rhino: hello
03:22 (quit) mithos28: Quit: mithos28
03:24 (quit) dmac: Ping timeout: 240 seconds
03:26 (quit) neilv: Ping timeout: 260 seconds
03:30 flying_rhino: someone here dudes?
03:35 Burlingk: :P I am, but I am a newb. :)
03:35 flying_rhino: heh
03:36 flying_rhino: I am newb too
03:36 flying_rhino: what do you plan to do with racket, fellow newb ?
03:38 Burlingk: Play around mostly.
03:39 Burlingk: I Would like to learn Lisp, so this is as good a place as any to start. :)
03:39 flying_rhino: Lisp is pretty cool
03:40 (join) dmac
03:41 flying_rhino: I know you are newb, but have you figured out how to create equivalents of java references or C pointers? I am trying to implement double linked list here
03:43 Burlingk: :P I am starting my way through a tutorial, and still get confused on handling text at times. :)
03:43 Burlingk: C or C++, I can tell you a lot more about :) hehe
03:44 flying_rhino: I know how to implement blasted thing there :P
03:52 (join) masm
03:52 flying_rhino: hi
04:50 (join) noelw
05:13 flying_rhino: guys?
05:21 noelw: duuude
05:29 flying_rhino: hi
05:30 flying_rhino: i am trying to implement doubly linked list (just to see how it works)
05:30 flying_rhino: and I am stuck
05:31 noelw: ok
05:31 flying_rhino: so node of list should be (struct: dl-node ([prev : dl-node] [next : dl-node] [data : Integer]))
05:31 flying_rhino: this compiles okay, but how to create instance
05:32 noelw: How does a normal list handle this case?
05:32 flying_rhino: like root that has some data, and prev-next references set to nill
05:32 flying_rhino: dunno
05:32 flying_rhino: I am creating my own to learn and see how would that work
05:33 noelw: Think about the single linked (i.e. normal) list
05:33 noelw: What is the datatype definition?
05:33 noelw: I.e. list = ???
05:34 flying_rhino: I am something of a lisp noob (can grok c/c++ just fine) , so you will have to use smaller words here :(
05:34 noelw: Ok.
05:34 noelw: The list (list 1 2 3) =
05:34 noelw: (cons 1 (cons 2 (cons 3 nil)))
05:35 noelw: So there are two different data structures used in a list
05:35 noelw: A list of integers
05:35 noelw: (list integer) = null
05:35 noelw: or (cons integer (list integer))
05:36 flying_rhino: basically it should work like list of integers (hence data : Integer) but it should be easier to traverse backwards
05:36 noelw: Sure.
05:36 noelw: But look at your data definition
05:36 noelw: You have only defined one struct
05:36 flying_rhino: I just thought to create some add-node and delete-node functions
05:36 noelw: Don't worry about those functions for now.
05:36 (join) mceier
05:37 noelw: Your data definition is wrong.
05:37 noelw: You won't be able to build a list till you fix it.
05:37 noelw: In C how would you create a double-linked list with one element?
05:37 noelw: What would you set prev and next to?
05:38 flying_rhino: well youl'l have struct that defines node. Prev and next would be pointers to previos and next node
05:38 noelw: And if there is no previous or next node?
05:38 flying_rhino: it would be null
05:38 flying_rhino: you'll have to check if it is null
05:39 noelw: Right.
05:39 noelw: So a (double) linked list is build from two types of data
05:39 noelw: nodes and null
05:40 flying_rhino: so prev and next should be node or null ?
05:40 noelw: Yes, just like a single linked list (which only has a next pointer, not a prev pointer)
05:41 noelw: For a single linked list, (list integer) = null or (cons integer (list integer))
05:41 noelw: A list is either null or a cons of an integer and a list
05:41 noelw: Now extend that to a double linked list
05:42 flying_rhino: something like (struct: dl-node ([prev : (U dl-node null)] [next : (U dl-node null)] [data : Integer]))
05:42 noelw: You probably want
05:43 noelw: (define-type dlist (U null dnode))
05:43 noelw: Then you can write prev and next as just dlist
05:43 noelw: (I mean as having type dlist)
05:44 flying_rhino: null means empty in Racket, right?
05:45 flying_rhino: I think I get it
05:45 noelw: groovy
05:46 flying_rhino: sp i do  (define-type dlist (U null dnode)) and then (struct: dl-node ([prev : dl-node] [next : dl-node] [data : Integer])) , and then I do functions for manipulating list
05:46 flying_rhino: *so I do
05:47 noelw: Yeah, though there is a typo in those type definitions.
05:47 noelw: What I called dnode you call dl-node
05:47 noelw: and prev and next should have type dlist
05:48 flying_rhino: okay... this will sound dumb but how to create instance of dl-node?
05:48 noelw: You mean the syntax?
05:48 flying_rhino: yeah
05:48 noelw: (dl-node ...)
05:48 noelw: (dl-node null null 1)
05:48 flying_rhino: yeah obvious, really
05:51 flying_rhino: so first node would be (define: first-node : dl-node (dl-node null null 1))
05:51 noelw: Note that's not a node but a list
05:51 noelw: The empty list is just null
05:52 noelw: The list of one element is what you've given
05:53 flying_rhino: so it that's not a node how to create first node?
05:54 noelw: When you create a node you have to create the rest of the list
05:54 noelw: Just like you can't create a cons without filling out the definition to create a list
05:54 noelw: You can't say, for example, (cons 1)
05:55 noelw: You have to say (cons 1 null)
05:55 noelw: and there you have a list...
05:55 flying_rhino: waitaminute, what I want here is mutable douby linked list. One where you can add new nodes at the end or in a middle.
05:56 noelw: Sure.
05:56 noelw: You have to mark the structure as mutable
05:56 noelw: (struct: … #:mutable)
05:56 flying_rhino: okay
05:56 noelw: (define dlist1 (dl-node null null 1))
05:57 flying_rhino: okay
05:57 noelw: (set-dl-node-next! dlist1 (dl-node dlist1 null 2))
05:59 flying_rhino: not sure it should work like that
06:01 flying_rhino: when adding new node to a list, there should be three arguments: (1) data you want node to have (2)existing node where you want to attach new node (3) bollean to se if you want node to be after or before existing node
06:02 flying_rhino: *three arguments
06:02 flying_rhino: I'll have to think about this more
06:02 noelw: Well you can implement that function if you want
06:02 flying_rhino: thank you for trying to help :)
06:03 noelw: np
06:05 flying_rhino: basically I am trying to develop a small game in Racket so I need some mutable data structures. I tried to start with doubly linked lists because they are easiest to implement (generally speaking). I know immutability is default with you guys, but when developing real time game it is unpractical when state changes 30 times a second.
06:06 noelw: Sure
06:06 flying_rhino: I hope you understand
06:06 noelw: There was a thread on double-linked lists on the Racket mailing list
06:06 noelw: Was last week, I think.
06:06 noelw: Maybe you started it? :)
06:06 noelw: Anyway, there were some interesting answers
06:07 flying_rhino: nope haven't started it
06:08 flying_rhino: I'll try to figure this out
06:31 flying_rhino: one more thing: when something has type 'any', that works just like in untyped Racket, right?
06:38 noelw: yeah
06:39 flying_rhino: but is immutable?
06:40 flying_rhino: I mean it is immutable in both cases
06:41 noelw: any is anything
06:41 noelw: immutable or not
06:41 flying_rhino: say I define (define: x : Positive-Integer 775)
06:41 flying_rhino: x is mutable ?
06:42 flying_rhino: I can chane it with (set! x 55)
06:42 flying_rhino: *change
06:42 noelw: Sure. Then it is
06:42 noelw: You've answered your own question :)
06:44 flying_rhino: and how would I define immutable one (not that I need it) ?
06:44 noelw: I don't think there is any syntax for that
06:44 flying_rhino: immutable positive-integer
06:44 noelw: Bindings are only mutable within a module
06:45 flying_rhino: so only lists, hashes and that fluf are immutable?
06:45 noelw: So if you don't export a set-x! function users of your module won't be able to mutate it
06:45 flying_rhino: didn't know that
06:45 flying_rhino: this racket is weird
06:45 noelw: Heh.
06:45 noelw: There is a difference between mutating a binding and mutating a value
06:46 noelw: Bindings not being mutable is an optimisation
06:46 noelw: You can change it if you so desire
06:47 noelw: I can't remember the name of the parameter, but it's in the docs somewhere
06:47 flying_rhino: okay
06:50 flying_rhino: I'll need more time to digest this stuff. I was drawn to Lisp because I like idea of customizing language with macros. I also like lack of syntax (makes easier to auto generate code). And posibility of extending code at runtime. But for imperative guy like me there is a lot of stuff to warp my mind around.
06:50 noelw: There sure is
06:50 noelw: First time I encountered functional programming my brain exploded
06:50 noelw: It took a long time to get it.
06:52 (join) neilv
06:52 (quit) neilv: Changing host
06:52 (join) neilv
06:52 flying_rhino: basically functional programming is to me lest atractive part of this thing. Nothing wrong with it but I jsut think imperative is better for a lot of stuff
06:52 flying_rhino: *the least
06:53 noelw: You'll learn a lot from the functional way of coding
06:53 flying_rhino: I hope so
06:53 noelw: There are many situations where this is useful
06:53 noelw: Concurreny and parallelism are the obvious cases
06:53 flying_rhino: ofcourse
06:54 flying_rhino: the problem with the thing I wanted to use it for (hobby game development) is that you need to create new state like 30 times a second... clearly unpractical
06:55 flying_rhino: can't be used for it
06:55 noelw: I generally agree with what you're saying
06:55 noelw: I'll note that Jay McCarthy at BYU is working on a different approach
06:55 noelw: Basically, if you accept that the performance critical parts of a game are moving to the GPU
06:55 noelw: and on the GPU sharing data is really bad
06:56 noelw: then you can get a lot of mileage from taking a functional approach and optimising the heck out of the resultant code to run fast on that GPU
06:56 noelw: Since FP code is much easier to analyse this can be a win
06:56 noelw: For the hobbyist, this isn't really viable. :)
06:57 flying_rhino: I doubt racket supports CUDA
06:57 flying_rhino: or wathever you need for GPU programming
06:57 bremner: you can certainly have state in racket. Heck you can have state in purist languages like Haskell
06:57 noelw: Racket support OpenCL
06:58 noelw: Sure. In Haskell you have to be explicit about all your data dependencies
07:00 flying_rhino: well since it has good C++ interop (or so I heard) so I can use 3d engine like ogre or irrlicht if I want. Not that I am going to do that right away
07:07 (quit) neilv: Ping timeout: 260 seconds
07:07 (quit) DT``: Read error: Operation timed out
07:22 (join) jeapostrophe
07:23 (join) DT``
07:29 flying_rhino: I have one more question (or more precisely, rant) if no one minds
07:35 noelw: rant on. it's irc
07:44 flying_rhino: why everyone (except lisp guys) hate idea of macroes? To program at all, you need turing complete programming language (stepping, branching, iterration), right? It therefore stands to reason that any abstraction facility worth spit will have to be turing-complete also. Or there will be plenty of cases when it is no help at all. OOP is basically fancy branching which everyone hyped to high heaven
07:44 flying_rhino: and which produces bloated code. C++ has pretrpocessor is cool, but lacks looping construct. And now we got generic, templates, interfaces and kitchen sink. And to the best of my knowledge, none of that is turing-complete nad therefore can't really reduce LOC much. It seems to me that programming language designers would rather give you blow job than tools that actually work (ie macroes). When
07:44 flying_rhino: are mainstream languages stop inventing excuses for why they don't give you macroes and start implementing macroes.
07:45 flying_rhino: *sorry for awfull grammar there
07:48 flying_rhino: *I accidentally hit enter before proofreading it
08:04 bremner: I think macros are pretty hard to do right in languages with non-trivial built-in syntax. Or if not hard to do right, hard to use the result (see template-haskell)
08:10 flying_rhino: Even language with non-trivial syntax is phrased to some kind of abstract tree. And you should be able to manipulate that tree like XML document, to add, remove or rename nodes. Results are not hard to see if IDE knows everything about macros. You should be able to expand every macro untill you get basic code. So I think the problem is cultural. They won't give us macros because they believe
08:10 flying_rhino: programmers are too incompetent to have it. So all you got are LEGO bricks (oop and stuff) that you can use to play around in Eclipse.
08:11 bremner: flying_rhino: there is no they.
08:12 flying_rhino: what do you mean? I don't design languages. For me they are they :)
08:13 bremner: I think your position is inconsistent. You claim to know better, but also not to know enough.
08:15 flying_rhino: Well I don't claim to be language genius. And even if I where one, it would take years to design language. And even if I where to desing one no one would use it because (1) lack of tools (IDE and so on) (2) lack of source code.
08:16 bremner: more or less one of the main themes of racket is making it easy to design your own language
08:16 flying_rhino: and it doesn't take genius to figure out that non turring complete abstraction facilities can be useful in some cases, but are awful in other cases.
08:17 flying_rhino: well, I am not complaining about racket
08:17 flying_rhino: I came to it because of macros (and a few other stuff)
08:17 flying_rhino: and no one says design a few domain specific langs isn't somewhere on my to-do list.
08:19 flying_rhino: okay, I went too far (I DID ask if it is okay to rant but I went to far), but I think I have a point somewhere there. I apologize
08:22 flying_rhino: basically I was wondering why most languages avoid macros at all cost (and create bunch of other stuff instead) instead of figuring out how to properly utilize them.
08:22 flying_rhino: I think that is fair question
08:28 (join) haruki_zaemon
08:29 flying_rhino: *pharsed
08:44 (join) dnolen
08:50 noelw: I think macros are great
08:50 noelw: I believe C++ templates are turing complete (but they are restricted to 16 levels of recursion IIRC)
08:51 noelw: The reason most languages don't have macros:
08:51 noelw: 1) Because they don't. People haven't used languages with macros before and so don't miss them
08:52 noelw: 2) Macros in infix languages are substantially harder (for the language designer). Manipulating AST is also harder (for the programmer)
08:52 noelw: That's my opinion
08:52 (part) haruki_zaemon: "Laterz"
09:03 flying_rhino: Hm... it's been awhile since I did anything serous in C++. If templates can do that, that's good news! In my time I was frustrated about alck of turring completeness in preprocessor glad to hear it is better now. Thanks, noelw.
09:03 (join) torche
09:27 (quit) jeapostrophe: Quit: jeapostrophe
09:29 (quit) dmac: Quit: WeeChat 0.3.5
10:26 (nick) samth_away -> samth
10:26 (join) jeapostrophe
10:29 (join) anRch
10:30 offby1: I'd heard the same about C++ templates (that they're Turing-complete) but to me that's an argument _against_ using them :)
10:30 offby1: mind-bendingly complex
10:30 offby1: but maybe C++ people feel the same way about Lisp macros
10:30 offby1: *shrug*
10:30 (quit) dnolen: Quit: dnolen
10:32 samth: gf3: take a look at `evcase'
10:32 samth: rudybot: doc evcase
10:32 rudybot: samth: your racket/init sandbox is ready
10:32 rudybot: samth: no docs for a current binding, but provided by: mzlib/etc
10:33 samth: rudybot: (require mzlib/etc)
10:33 rudybot: samth: Done.
10:33 samth: rudybot: doc evcase
10:33 rudybot: samth: http://docs.racket-lang.org/mzlib/mzlib_etc.html#(form._((lib._mzlib%2Fetc..rkt)._evcase))
10:33 samth: offby1: the "not really sent by jay.mcarthy@gmail.com" is b/c jay is using a mail server that isn't gmail, i think
10:33 offby1: how annoying
10:35 offby1: perhaps the real issue is: gmail's SMTP front end adds some X-I-Am-Not-A-Phishing-Message-Honest header, whereas those other SMTP servers don't
10:35 samth: but perhaps eli's explanation is the right one
10:35 samth: or maybe some combination
10:37 offby1: ah, hadn't read that until now.
10:37 offby1: It's odd; I'd have thought there'd be plenty of lists in the same situation
10:38 offby1: and that therefore google would have done something to make it easy for them to fix whatever the problem is
10:42 (quit) jeapostrophe: Quit: jeapostrophe
10:48 (join) realitygrill
10:50 (join) edjka
10:52 (quit) realitygrill: Client Quit
11:02 (quit) noelw: Ping timeout: 260 seconds
11:10 samth: offby1: just use google groups? :P
11:13 (join) noelw
11:16 eli: offby1: AFAICT, that header is the DKIM thing, which is some encryption-based fanciness to prove your honesty.  But I can't make our mailing list vouch for a gmail-originated email...
11:22 samth: eli: the instructions here: http://mail.google.com/support/bin/answer.py?hl=en&ctx=mail&answer=185812 make it seem like you just need the SPF record and an "envelope sender" (whatever that is)
11:23 (part) edjka: "ERC Version 5.3 (IRC client for Emacs)"
11:24 (quit) hkBst: Remote host closed the connection
11:24 (quit) Burlingk: Quit: Leaving
11:35 (join) dnolen
11:38 (join) jonrafkind
11:42 (join) RacketCommitBot
11:42 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/HM-7ZA
11:42 RacketCommitBot: [racket/master] share table of shared code pointers - Matthew Flatt
11:42 (part) RacketCommitBot
11:52 (join) RacketCommitBot
11:52 RacketCommitBot: [racket] plt pushed 4 new commits to master: http://git.io/Pz2mVw
11:52 RacketCommitBot: [racket/master] add a better link from the snip class docs to the relevant section of the - Robby Findler
11:52 RacketCommitBot: [racket/master] Rackety - Robby Findler
11:52 RacketCommitBot: [racket/master] remove unnecessary thread creation & remove commented out code - Robby Findler
11:52 (part) RacketCommitBot
11:54 (join) mithos28
11:59 (join) jeapostrophe
12:06 (join) jeapostrophe_
12:06 (quit) jeapostrophe: Read error: Connection reset by peer
12:06 (nick) jeapostrophe_ -> jeapostrophe
12:07 samth: jeapostrophe: tragically, drdr loses timing history on moved files
12:07 (part) jeapostrophe
12:07 (join) jeapostrophe
12:07 jeapostrophe: yup
12:07 samth: for example: http://drdr.racket-lang.org/23441/collects/tests/typed-racket/run.rkt
12:08 jeapostrophe: it doesn't know anything about what git is doing
12:11 (join) realitygrill
12:16 samth: jeapostrophe: can I ask you to manually copy the timing data for that file?
12:20 (join) RacketCommitBot
12:20 RacketCommitBot: [racket] plt pushed 3 new commits to master: http://git.io/AlgOOw
12:20 RacketCommitBot: [racket/master] Make the type of `syntax-local-module-defined-identifiers' more precise. - Sam Tobin-Hochstadt
12:20 RacketCommitBot: [racket/master] Move environment initialization after local expansion in Typed Racket. - Sam Tobin-Hochstadt
12:20 RacketCommitBot: [racket/master] Add test of behavior like the "module" language in DrRacket. - Sam Tobin-Hochstadt
12:20 (part) RacketCommitBot
12:21 jeapostrophe: samth: done
12:23 samth: jeapostrophe: there's something wrong there: http://drdr.racket-lang.org/23441/collects/tests/typed-racket/run.rkt
12:24 jeapostrophe: you are assuming the data points come in order, aren't you?
12:24 samth: i am at the moment -- should i change that?
12:25 samth: it's easy enough to change
12:25 jeapostrophe: they normally do, but I can just sort the file
12:25 samth: i can change it to sort in JS
12:25 samth: it'll be easy
12:26 jeapostrophe: i just sorted it
12:29 (quit) anRch: Quit: anRch
12:29 samth: i just changed it to sort, so we won't have to worry
12:30 (join) RacketCommitBot
12:30 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/XSZrlQ
12:30 RacketCommitBot: [racket/master] Sort chart data. - Sam Tobin-Hochstadt
12:30 (part) RacketCommitBot
12:38 (join) anRch_
12:38 (join) MayDaniel
12:53 samth: stamourv: this chart makes me happy http://drdr.racket-lang.org/23513/collects/tests/typed-racket/run.rkt
13:03 stamourv: samth: I'm happy too.
13:04 jonrafkind: when I mouse over and get a popup it looks like the 'push 123' is a link but i can't possibly click that becuase if i move the mouse the popup goes away
13:05 (quit) anRch_: Quit: anRch_
13:05 samth: jonrafkind:  you click on the point in the graph, the popup stays
13:06 jonrafkind: oh.. the page should say that somewhere
13:06 samth: yeah, i have to write the help text
13:06 jonrafkind: ok
13:06 (join) danking
13:08 (join) superjudge
13:11 samth: btw, stamourv, i had the following idea for optimization
13:13 samth: if you care comparing two numeric union types for subtype, then that can be done faster by basically checking that all the elements of the first are elements of the second, which is faster b/c they're in order
13:13 samth: (this only works for numeric types b/c other types have non-trivial subtype relationships)
13:14 (join) machuga
13:33 (join) GeneralMaximus
13:44 stamourv: samth: Wouldn't that work for all union types? The contents are always sorted.
13:53 (quit) superjudge: Quit: superjudge
14:10 (quit) noelw: Quit: noelw
14:22 (join) anRch
14:34 (quit) dnolen: Ping timeout: 252 seconds
14:38 (quit) realitygrill: Quit: realitygrill
14:41 (join) dmac
14:55 samth: stamourv: other types that might be components of unions can have non-trival subtype relationships (eg, structs)
14:57 (quit) MayDaniel: Read error: Connection reset by peer
14:59 stamourv: Oh, right.
14:59 stamourv: Ah, but to first know if a type is a numeric type, you need to ask if it's a subtype of number, so I don't know how much we win here.
14:59 stamourv: Unless we want to keep a list of types that are numeric.
15:05 samth: we should make it possible to have the numeric types all have Type-key 'number
15:16 (join) realitygrill
15:17 eli: samth: That link-on-a-popup-that-disappears is very bad UI.
15:17 eli: Just have it disappear after a timeout, and there won't be any need for clicks.
15:18 samth: eli: i disagree -- i'd find it very hard to navigate without the tooltip
15:19 eli: samth: I'm not objecting to the tooltip -- just to it disappearing as soon as you move the mouse.  If you let it linger for a second, then things become easy without any need for some help text.  (Which I promise you nobody will read.)
15:20 samth: eli: that's how tooltips work -- they appear only while your mouse is over things
15:21 eli: samth: But a tooltip that has a *link* should not disappear immediately.
15:21 samth: would you want a stack of tooltips to pile up?
15:21 samth: that's why you can click
15:22 samth: would you prefer if the hover-tooltip didn't have the link?
15:22 eli: (1) no, when a new one appears you immediately remove the old one; (2) even havig a bunch of them appearing and disappearing is way better than a counter-intuitive click.
15:22 eli: s/havig/having/
15:23 eli: Again, the link is fine, just don't make the tooltip disappear.
15:23 samth: having a new one replace the old would defeat the purpose of your suggestion
15:23 samth: the charts are too dense for that to work w/o the click
15:24 eli: That's the way it is, people will just carefully move the mouse away to a place where no new one appears.
15:24 (quit) Twey: *.net *.split
15:24 (quit) tewk: *.net *.split
15:24 (quit) zakwilson: *.net *.split
15:24 (quit) adzuci: *.net *.split
15:24 (quit) janne: *.net *.split
15:24 (quit) rapacity: *.net *.split
15:24 (quit) ozzloy: *.net *.split
15:24 (join) janne
15:24 (join) zakwilson
15:24 (join) rapacity
15:24 (join) adzuci
15:24 stamourv: samth: Isn't that type key suposed to be unique?
15:24 eli: If you really want to make things fine, then make it jump to the page on a click.
15:24 eli: But earlier you said that there was some problem with that.
15:24 (join) ozzloy
15:24 (quit) si14: Remote host closed the connection
15:24 stamourv: Or am I confusing it with something else.
15:24 (join) tewk
15:24 (quit) ozzloy: Changing host
15:24 (join) ozzloy
15:25 samth: eli: but that isn't the way to get to where the link is
15:25 samth: stamourv: no, the type key is a coarse grouping
15:25 samth: if two types both have non-#f type keys, then they must overlap (or be the same) for there to be a subtyping relationship between them
15:25 stamourv: samth: Ah, ok. Yeah, we can do that.
15:26 stamourv: I'll implement that.
15:26 eli: samth: What I'm saying is that either a click takes you to the page, and then the tooltip can say "click to get to that push", or the tooltip has the link and in that case it should linger for a while.
15:26 eli: That's very basic UI.
15:26 eli: A tooltip is providing extra information.
15:26 eli: That's why it's called a "tip".
15:26 eli: Your tooltips are more than tips now, which is why it's bad for them to disappear like tooltips.
15:26 samth: stamourv: we might need to change how the key information is speced in type-rep
15:27 stamourv: samth: How so?
15:27 (join) Twey
15:28 samth: stamourv: right now, the key for a base type is computed very simplistically
15:28 samth: see the `def-type Base' in type-rep.rkt
15:29 samth: eli: i disagree -- i find the interface nicer than either of your suggestions
15:30 stamourv: samth: Right. Could we add a kw argument for it?
15:31 stamourv: (To make-Base.)
15:31 eli: samth: Whatever.
15:31 stamourv: samth: And we'd need it for U too.
15:31 eli: I just wonder how many more people need to complain before you change your mind.
15:31 (join) si14
15:33 eli: samth: Oh, BTW, if you really want: find precedence for the current behavior.  I bet that I can find 10 times more for what I suggested; and for the simple click-to-jump it's trivially 1000x.
15:39 (quit) si14: Remote host closed the connection
15:43 (join) superjudge
15:44 ohwow: Hello
15:46 (quit) jeapostrophe: Quit: jeapostrophe
15:47 (quit) anRch: Quit: anRch
16:07 (quit) realitygrill: Quit: realitygrill
16:08 clklein: I have a few unpushed commits that I've decided should be on a branch. Whats' the right way to do that? Use our "fork" extension to clone the plt tree then somehow copy my commits to that clone?
16:11 samth: clklein: do the fork, add the as a remote, and then push to that remote
16:20 clklein: samth: Thanks, that seems to have done it.
16:21 (quit) em: Ping timeout: 260 seconds
16:22 (join) si14
16:25 (join) ic_
16:25 (quit) ic_: Client Quit
16:26 (quit) torche: Quit: torche
16:46 (quit) superjudge: Quit: superjudge
16:51 (join) ic1
16:52 (part) ic1
16:53 (join) ic1
16:53 (join) realitygrill
16:53 (part) ic1
16:54 (join) ic1
16:54 (part) ic1
16:57 (join) ic1
16:57 (part) ic1
16:58 (join) ic1
16:59 (join) torche
16:59 ic1: jemand aus deutschalnd?
17:00 bremner: I was there once, does that count?
17:00 (join) littlebobby
17:00 jonrafkind: i am a jelly donut
17:01 petey: i know a tiny bit of german
17:01 petey: but google translate probably knows more than i do :(
17:02 ic1: you can understand german^^
17:02 ic1: all from US?
17:03 samth: ic1: matthias felleisen, the originator of Racket, is from deutschland
17:04 ic1: ok
17:04 samth: und andern kann ein bisschen deutsch wenn es not ist
17:04 ic1: I asked, because of my wrong english
17:05 samth: i wouldn't worry about that, ic1, we deal w/ plenty of college students
17:05 samth: :)
17:05 (nick) samth -> samth_away
17:07 ic1: nice, can I ask something if I need something regarding Dr Racket?
17:08 (join) RacketCommitBot
17:08 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/dIDIaA
17:08 RacketCommitBot: [racket/master] `random' always produces `Float'. - Sam Tobin-Hochstadt
17:08 (part) RacketCommitBot
17:11 (part) ic1
17:13 (join) RacketCommitBot
17:13 RacketCommitBot: [racket] plt pushed 1 new commit to master: http://git.io/KZLRGg
17:13 RacketCommitBot: [racket/master] fixed and/or documentation, Closes PR 12197 - Matthias Felleisen
17:13 (part) RacketCommitBot
17:13 (join) ic1
17:13 (part) ic1
17:15 (quit) realitygrill: Quit: realitygrill
17:21 jonrafkind: are there any good use cases/examples of a name from the use of a macro bound to another macro in the template?
17:21 jonrafkind: like (foo x) => #'(let-syntax ([x ...]) x)
17:26 stamourv: t-r/rep/rep-utils.rkt probably does something like that.
17:26 stamourv: But you may not find that file especially enlightening/
17:26 stamourv: s@/@.@
17:27 jonrafkind: oh wtf theres typed-racket and typed/racket
17:27 jonrafkind: typed-racket is the old version, right?
17:28 stamourv: No. typed-scheme is.
17:28 stamourv: typed-racket is the implementation.
17:28 stamourv: typed/racket is the language.
17:28 stamourv: typed/scheme is the old language.
17:28 jonrafkind: why not put the implementation into typed/racket ?
17:29 stamourv: I argued for that, to save a collect. I forget the reasons.
17:29 stamourv: Reasons for going with typed-racket, that is.
17:29 jonrafkind: oh wel, anyway i dont see any uses of let-syntax in rep-utils.rkt, it does define-syntax a name but that name is just provided
17:30 jonrafkind: ok the question is if you have (foo x (x 5)) you could bind x to a let-syntax and have x transform the (x 5) part
17:30 jonrafkind: this is somewhat impossible in honu and im wondering if it matters
17:41 petey: icl: go ahead and ask
17:41 (join) ic1
17:41 (part) ic1
17:45 (join) realitygrill
17:46 (join) em
17:54 stamourv: samth: Your javascript graph thingy for DrDr should not truncate its y axis.
18:03 (join) jeapostrophe
18:09 (join) feor
18:12 (quit) littlebobby: Quit: Ex-Chat
18:14 (quit) feor: Remote host closed the connection
18:14 (join) feor
18:15 (quit) jeapostrophe: Quit: jeapostrophe
18:42 (join) jeapostrophe
18:44 (quit) jeapostrophe: Client Quit
18:52 (join) otzi
19:09 (join) RacketCommitBot
19:09 RacketCommitBot: [racket] plt pushed 2 new commits to master: http://git.io/jHF9sw
19:09 RacketCommitBot: [racket/master] allow different phases for `free-identifier=?' arguments - Matthew Flatt
19:09 RacketCommitBot: [racket/master] add `syntax-transforming-module-expression?', `variable-reference->module-base-phase' - Matthew Flatt
19:09 (part) RacketCommitBot
19:15 (quit) janne: Ping timeout: 260 seconds
19:30 (join) janne
19:35 (quit) masm: Quit: Leaving.
19:56 (quit) machuga: Ping timeout: 260 seconds
20:08 (quit) flying_rhino: Ping timeout: 260 seconds
20:17 (join) jeapostrophe
20:22 (quit) jeapostrophe: Client Quit
20:48 (quit) dmac: Ping timeout: 276 seconds
21:18 (quit) jonrafkind: Read error: Operation timed out
21:43 (join) dmac
21:51 (join) machuga
22:10 (quit) machuga: Ping timeout: 260 seconds
22:19 (join) RacketCommitBot
22:19 RacketCommitBot: [racket] plt pushed 4 new commits to master: http://git.io/ZMRrMA
22:19 RacketCommitBot: [racket/master] doc clarifications on `eq?' - Matthew Flatt
22:19 RacketCommitBot: [racket/master] scribble: more nowraps in HTML output - Matthew Flatt
22:19 RacketCommitBot: [racket/master] simpify an example - Matthew Flatt
22:19 (part) RacketCommitBot
22:47 (quit) em: Read error: Operation timed out
22:50 (join) em
22:51 (quit) dmac: Ping timeout: 260 seconds
23:21 (quit) mceier: Quit: leaving
23:33 (join) Apoc_
23:33 Apoc_: Hello?
23:35 mithos28: Apoc_: Hello there.
23:35 Apoc_: I've got a quick question been a little stuck was hoping to get some clarification that I couldnt get from the docs
23:36 Apoc_: not sure if this is the place to do that or not :P
23:36 mithos28: This is one place, and if you cannot get the help you need hear the mailing lists are available
23:37 mithos28: here*
23:37 Apoc_: alright, its fairly simple, more of a syntax question really
23:37 mithos28: ok, ask away
23:37 Apoc_: if i have a function with two arguements, how do i write the function body to then use both?
23:38 Apoc_: define: expected only one expression for the function body, but found 1 extra part
23:38 Apoc_: im getting that error
23:38 mithos28: Can you write out what you have
23:38 Apoc_: sure
23:38 Apoc_: (define (star-at-angle a b)      (overlay (rotate a STAR) MTS)     (overlay (rotate b STAR) MTS)     )
23:38 Apoc_: thats my function
23:38 Apoc_: MTS and STAR are defined beforehand
23:39 mithos28: ok so the problem is that you have two expressions in the body
23:39 mithos28: each of which correspond to a picture (I think)
23:39 Apoc_: alright, so how would i call that expression with the two arguements?
23:39 Apoc_: and i think you are right
23:40 mithos28: what do you want to do with the two pictures
23:40 Apoc_: im going to be calling animate on them
23:41 mithos28: put them next to each other, above each other, overlaid?
23:42 mithos28: Can you tell me in words what (star-at-angle a b) is supposed to do?
23:44 Apoc_: im thinking that it should flip between two images
23:44 Apoc_: that are rotated 2 different amounts
23:44 (join) jonrafkind
23:45 (join) flying_rhino
23:45 mithos28: you cannot do that with having two expressions in the body of a function
23:45 Apoc_: hmm
23:45 Apoc_: alright, i think im missing something else here
23:46 Apoc_: i'm going to do a little more digging
23:47 mithos28: are you using 2htdp/image?
23:47 (join) neilv
23:50 Apoc_: yep
23:51 mithos28: I don't think you can do any animation with just that, but if you look at 2htdp/universe you should find what you are looking for
23:52 Apoc_: alright, will do, thanks!
23:52 (quit) Apoc_: Quit: Page closed
23:53 (join) dnolen
23:58 (join) dmac