| --- Log opened Thu Aug 25 00:00:10 2011 |
| 00:06 | < Attilla> | it chafes uncomfortably, that's why I prefer leather |
| 00:10 | <@Tarinaky> | Not funny. |
| 00:12 | <@Tamber> | I thought it was. |
| 00:20 | | * Alek sends someone to fit Attilla for a polyester gimp suit. |
| 00:32 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has joined #code |
| 00:41 | | celticminstrel [celticminst@Nightstar-5d22ab1d.cable.rogers.com] has quit [Ping timeout: 121 seconds] |
| 00:53 | | Derakon[AFK] is now known as Derakon |
| 00:55 | | Thaqui [Thaqui@27B34E.D54D49.F53FA1.6A113C] has quit [Connection closed] |
| 00:59 | | Vornicus [Vash@Nightstar-f03c5637.sd.cox.net] has quit [Connection closed] |
| 01:27 | | Derakon [Derakon@Nightstar-c826c088.ca.comcast.net] has quit [Client exited] |
| 01:27 | | Derakon [Derakon@Nightstar-c826c088.ca.comcast.net] has joined #code |
| 01:55 | | Attilla [Some.Dude@Nightstar-f29f718d.cable.virginmedia.com] has quit [Ping timeout: 121 seconds] |
| 02:27 | | celticminstrel [celticminst@Nightstar-5d22ab1d.cable.rogers.com] has joined #code |
| 03:46 | < Phox> | Hey, I have a problem. My instructor kept telling me to put a star in front of a variable to make it a pointer, so I loaded up a japanese IME and put a star in front of my variable, and it wouldn't compile. int ?b = &a; |
| 03:47 | < Phox> | Should I tell him he's a moron, and should use proper terminology? |
| 03:48 | <@McMartin> | Er |
| 03:48 | <@McMartin> | That looks like it should be right. |
| 03:48 | <@McMartin> | Oh. You're using a five-pointed star |
| 03:48 | <@McMartin> | No, you're being a twatwaffle. |
| 03:48 | <@McMartin> | "star" is a perfectly good name for the unicode code point *. |
| 03:49 | < Phox> | XD, twatwaffle. |
| 03:51 | < Phox> | Nah, I actually kid. I have a lot of respect for the guy. I think he's just finishing up his Masters. I do things like this because I get bored in class. |
| 03:53 | <@McMartin> | If you want to have Fun With Official Unicode Names, "=" has QUADRATHORPE as an alias. |
| 03:55 | < Phox> | ... I even have a use for that. |
| 03:57 | | Vash [Vash@Nightstar-f03c5637.sd.cox.net] has joined #code |
| 03:58 | | Vash is now known as Vornicus |
| 04:04 | | cpux is now known as shade_of_cpux |
| 04:08 | | Kindamoody[zZz] is now known as Kindamoody |
| 04:35 | | Omega [omegaboot@Nightstar-10752b3e.il.comcast.net] has joined #code |
| 04:35 | | Alek [omegaboot@Nightstar-10752b3e.il.comcast.net] has quit [NickServ (GHOST command used by Omega)] |
| 04:35 | | Omega is now known as Alek |
| 05:13 | | PinkFreud [WhyNot@NetworkAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 06:11 | | AnnoDomini [annodomini@Nightstar-f7eedefa.80-203-17.nextgentel.com] has joined #code |
| 06:27 | | Stalker [Z@Nightstar-3602cf5a.cust.comxnet.dk] has quit [Ping timeout: 121 seconds] |
| 06:34 | | Derakon is now known as Derakon[AFK] |
| 06:36 | | Kindamoody is now known as KiMo|prog |
| 06:46 | | celticminstrel [celticminst@Nightstar-5d22ab1d.cable.rogers.com] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] |
| 07:09 | | Thaqui [Thaqui@27B34E.D54D49.F53FA1.6A113C] has joined #code |
| 08:13 | | jeroid [jerith@687AAB.5E3E50.73E66E.A55161] has joined #code |
| 08:21 | | Vornicus [Vash@Nightstar-f03c5637.sd.cox.net] has quit [[NS] Quit: Spaaaaaaaaaaaaaaaaace sleeeeeeeeeeeeeeeep!] |
| 08:39 | | You're now known as TheWatcher |
| 08:55 | | KiMo|prog is now known as Kindamoody |
| 09:06 | | jeroid [jerith@687AAB.5E3E50.73E66E.A55161] has quit [[NS] Quit: Bye] |
| 09:10 | | Kindamoody is now known as Kindamoody|out |
| 10:05 | | Attilla [Some.Dude@Nightstar-f29f718d.cable.virginmedia.com] has joined #code |
| 13:50 | | PinkFreud [WhyNot@NetworkAdministrator.Nightstar.Net] has joined #code |
| 14:14 | | celticminstrel [celticminst@Nightstar-5d22ab1d.cable.rogers.com] has joined #code |
| 15:24 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has quit [Client closed the connection] |
| 15:24 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has joined #code |
| 15:34 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has quit [Ping timeout: 121 seconds] |
| 15:39 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has joined #code |
| 16:01 | | Stalker [Z@Nightstar-3602cf5a.cust.comxnet.dk] has joined #code |
| 16:11 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has quit [Ping timeout: 121 seconds] |
| 16:17 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has joined #code |
| 16:38 | | Thaqui [Thaqui@27B34E.D54D49.F53FA1.6A113C] has quit [Connection closed] |
| 17:47 | | Derakon [chriswei@Nightstar-c826c088.ca.comcast.net] has joined #code |
| 17:52 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has quit [Ping timeout: 121 seconds] |
| 17:55 | | * Derakon mutters at SWIG and Python and nonresponsive mailing lists. |
| 17:57 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has joined #code |
| 18:07 | | * Derakon sticks "int main(int argc, char** argv) {}" into a file, does "gcc test.c", gets an "undefined symbol: _main" error. |
| 18:07 | < Derakon> | Oh wait, I'm a dumbass. |
| 18:09 | < Derakon> | (Forgot to make the forward function declaration match the function -- I had an errant "static") |
| 18:33 | | * Derakon mutters some more at the use of wide characters in general. |
| 18:34 | < Derakon> | Especially since the library I have to interoperate with a) just uselessly renames the standard wchar_t to "AT_WC", and b) provides conversion functions between wchar_t* and char*, which makes me think they aren't even using the extra byte. |
| 18:36 | | * TheWatcher hates all over wide characters >.< |
| 18:36 | < Derakon> | I understand the need for them in certain areas, but this is really not one of them. |
| 18:37 | < Derakon> | (I'm trying to talk to a camera, and strings are used only for choosing options by name or for error codes) |
| 18:40 | < Derakon> | Ideally there'd be some way in the SWIG interface file to say "Here, these wchar_t things? Convert them to chars and then handle them normally, please", but I'm having a devil of a time figuring out how to do that and the mailing list appears to be dead. |
| 18:40 | < Derakon> | (And there's no associated IRC channel) |
| 19:09 | | Kindamoody|out is now known as Kindamoody |
| 19:10 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has quit [Ping timeout: 121 seconds] |
| 19:11 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has joined #code |
| 19:19 | | Vash [Vash@Nightstar-f03c5637.sd.cox.net] has joined #code |
| 19:19 | | Vash is now known as Vornicus |
| 19:20 | | * Derakon spends a half-hour trying to figure out why his typemap isn't getting used, only to realize it's because he wasn't generating the necessary intermediary file after having added it. |
| 19:20 | < Derakon> | I hate it when that happens. |
| 19:21 | < Vornicus> | typemap? what're you playing with today? |
| 19:21 | < Derakon> | SWIG. |
| 19:21 | < Derakon> | I need to teach it how to convert Python strings into wchar_t*s |
| 19:21 | < Vornicus> | aha |
| 19:22 | < Derakon> | So I was modifying my SWIG interface file, building the wrapper file, linking it, and trying to use it...but between steps 1 and 2 I was supposed to be generating the wrapper file from the interface file. >.< |
| 19:24 | < Derakon> | ! |
| 19:24 | < Derakon> | It worked! |
| 19:24 | | * Derakon does a small version of the happy coder dance in his chair. |
| 19:24 | < Vornicus> | Happy Coder Dances, yay! |
| 19:28 | < Derakon> | This is one of those tasks that would have been trivial to anyone who knew what they were doing~ |
| 19:29 | < Vornicus> | Yeah, I noticed that happens a lot. |
| 20:02 | | Kindamoody is now known as Kindamoody[zZz] |
| 21:06 | | Vornicus [Vash@Nightstar-f03c5637.sd.cox.net] has quit [Connection closed] |
| 21:19 | < Derakon> | There, now I have functional string returning too. |
| 21:19 | < Derakon> | ...though there might be the possibility for a memory leak here. |
| 21:20 | < Derakon> | Yeah, memory leak. Hm. |
| 21:21 | < Derakon> | Wow, my C is rusty. How do you generate and return a C string (array of chars) without memory leaks or returning the address of a local variable? |
| 21:22 | < Derakon> | mallocing the array of course causes the memory leak, and doing "char buffer[1000]" means returning the address of a local. |
| 21:24 | <@Tamber> | Maybe I'm thinking of this problem in a different way to you, but why is it leaking memory? You're malloc-ing the memory inside foo(), returning the pointer to it, then not freeing it in the caller when you're finished with it? |
| 21:24 | <@McMartin> | (a) malloc, and rely on the caller to dispose of it (ownership model!) |
| 21:24 | <@Tamber> | ^ |
| 21:24 | < Derakon> | That gets difficult when the caller is Python. |
| 21:24 | <@McMartin> | (b) pass in a pre-allocated array as an argument, a la sprintf, and trust they got the size right. |
| 21:24 | < Derakon> | Likewise. |
| 21:24 | <@Tamber> | ...ah, yes, that doesn make it a little difficult. |
| 21:24 | | mode/#code [+ooooooooooooooooo Derakon[AFK] Phox TheWatcher simon_ Reiver Attilla PinkFreud celticminstrel AnnoDomini Derakon Kazriko Kindamoody[zZz] ToxicFrog Syloqs-AFH Alek Stalker shade_of_cpux] by Tamber |
| 21:24 | <@McMartin> | That's not "your C is rusty" |
| 21:24 | <@McMartin> | that's "your Python/C FFI is rusty" |
| 21:24 | <@Derakon> | Ah. |
| 21:25 | <@Derakon> | For some reason I remembered there being some way to pass character arrays around by value. |
| 21:25 | <@McMartin> | A struct with a char array in it, passed by value |
| 21:25 | <@McMartin> | If it's fixed size |
| 21:25 | <@McMartin> | Viz. Not A String. |
| 21:26 | <@Derakon> | The ultimate goal is to call an API function which accepts a wchar_t* buffer and stuffs a result value into it. |
| 21:26 | <@Derakon> | I'm trying to make that look, to the Python, like "call this function with no arguments, and get a string back". |
| 21:26 | <@Derakon> | So somewhere in there I have to allocate a buffer of some arbitrary size and pass it to the API function. |
| 21:27 | <@McMartin> | Python must have some form of "this chunk of memory belongs to you now, and it's a string, or "build a string from this buffer, and the buffer can be destroyed later". |
| 21:30 | | * McMartin throws together a small Haskell program over lunch to decode the clues he's collected in M&M2. |
| 21:30 | <@Derakon> | All of the actual conversions are being done inside SWIG, which is AFAICT entirely little C stubs. |
| 21:31 | <@Derakon> | So I have access to the Python C libraries, which I have zero familiarity with, but no actual Python. |
| 21:31 | <@Derakon> | In short: tricky. |
| 21:41 | <@Derakon> | ...hm, this gets worse, since I'm actually allocating two distinct buffers. |
| 21:41 | <@Derakon> | Though I can eliminate one of those... |
| 21:42 | <@Derakon> | Yeah, that's gone but I still have a memory leak because of the other buffer. :\ |
| 21:45 | <@Kazriko> | There's StringIO for building strings as if they were files, but i'm not sure as to its internal implementation. |
| 21:45 | <@Derakon> | Hokay, there's a solution that works and doesn't leak memory, but it required me to write a one-off wrapper for a specific function. |
| 21:45 | <@Derakon> | That's not so good for generalizing to all functions that return strings in this fashion. |
| 21:46 | <@Kazriko> | http://www.python.org/dev/peps/pep-3137/ << Buffers in python 3.x? |
| 21:49 | <@Derakon> | (This solution involves telling SWIG "Okay, replace this function with a function that allocates a wchar_t buffer, calls the original function, allocations a char buffer, copies the wchar_t buffer to it, creates a PyObject from the char buffer, then frees both buffers and returns the PyObject" |
| 21:49 | <@Derakon> | The only part of that that's function-specific is where the original function is called, mind you... |
| 21:49 | <@Derakon> | Seems like it %Bshould%B be possible to generalize. |
| 21:50 | <@Derakon> | Er. should |
| 22:40 | | AnnoDomini [annodomini@Nightstar-f7eedefa.80-203-17.nextgentel.com] has quit [[NS] Quit: Zzz.] |
| 23:17 | | * Derakon eyes wcstombs. |
| 23:17 | <@Derakon> | Why aren't you inserting a null character at the end of the array? |
| 23:18 | <@Derakon> | I'm having to set it manually. What the heck. |
| 23:23 | <@McMartin> | Does the length not include the null? |
| 23:24 | <@Derakon> | Here, this is what I'm talking about: http://paste.ubuntu.com/674868/ |
| 23:24 | <@Derakon> | If I don't have line 6, then when I print cBuffer it wanders off the end of the array. |
| 23:26 | <@McMartin> | That's because wcslen doesn't include the null character |
| 23:26 | <@McMartin> | So you're running out of space because it doesn't have "room" to convert it. |
| 23:27 | <@McMartin> | n is wcstombs isn't the "length of the string", it's the size of the destination buffer |
| 23:27 | <@Derakon> | Derp. Thank you. |
| 23:27 | <@McMartin> | So it's running out of space - or so it thinks - before it can convert the terminator |
| 23:27 | <@Derakon> | Yes, once you said the line about not having room I got it. |
| 23:27 | <@McMartin> | Also, this code will break if you actually have any multibyte chars; you should call it with n=0 and buffer=NULL first to get the amount of space you need to malloc. |
| 23:27 | <@Derakon> | Clearly I am le tired. |
| 23:28 | <@McMartin> | Er |
| 23:28 | <@McMartin> | plus 1 |
| 23:28 | <@Derakon> | I didn't follow that bit... |
| 23:28 | <@McMartin> | instead of having len be wcslen(wideStr) |
| 23:29 | <@McMartin> | the man page suggests using wcstombs(NULL,src,0)+1 |
| 23:29 | <@McMartin> | Then that is your amount of bytes to malloc and the size of the buffer you pass in |
| 23:29 | <@Derakon> | Ah. |
| 23:29 | <@McMartin> | That will - in the event that some characters *do* require multiple bytes - ensure space for exactly enough to hold them. |
| 23:30 | <@Derakon> | Right, makes sense. |
| 23:30 | <@McMartin> | However, it can also return (size_t)(-1), which means "this has characters I can't convert". |
| 23:30 | <@McMartin> | You probably don't want to try to malloc that. -_- |
| 23:31 | <@Derakon> | Is that "negative size_t" or "one less than maxint"? |
| 23:31 | <@Derakon> | Or something else? |
| 23:31 | <@McMartin> | That is, in fact, "precisely MAXINT", since size_t is unsigned and twos-complement |
| 23:31 | <@Derakon> | Oh, right, a cast. |
| 23:31 | <@McMartin> | It might even be MAXINT64. |
| 23:31 | <@McMartin> | Whatever that's really called. |
| 23:33 | <@Derakon> | Man, I am missing semicolons every time. |
| 23:34 | <@Derakon> | In the absence of the desire to properly handle this, I'll just return "Whoops, that didn't work at all!" if invalid characters show up. >.> |
| 23:34 | <@Derakon> | Maybe I'll deal with it properly tomorrow when my brain returns. |
| 23:36 | <@Derakon> | Oh look, I managed to trigger that behavior. I guess trying to print the 1024 characters that come after 'a' doesn't work so well~ |
| 23:37 | <@Derakon> | Guess i'll need to figure out a saner default buffer size. |
| 23:46 | <@McMartin> | Odd, that should work if your default locale is UTF-8 |
| 23:53 | | * McMartin shakes the spiders out of his brain |
| 23:59 | | Derakon [chriswei@Nightstar-c826c088.ca.comcast.net] has quit [[NS] Quit: leaving] |
| --- Log closed Fri Aug 26 00:00:26 2011 |