| --- Log opened Fri Oct 07 00:00:02 2011 |
| --- Day changed Fri Oct 07 2011 |
| 00:00 | | * Derakon gets a spam from someone trying to sell him CUDA stuff for Matlab. O_o |
| 00:00 | < Derakon> | That's...remarkably specific. |
| 00:00 | < Derakon> | I'm guessing they've been trawling the mailing lists I'm on? Maybe? |
| 00:02 | < Derakon> | Anyway, gotta run. |
| 00:02 | | Derakon [chriswei@Nightstar-f68d7eb4.ca.comcast.net] has quit [[NS] Quit: leaving] |
| 00:09 | < celticminstrel> | So, it works; I just need to make sure that the selection hilite isn't included and then I can move on to reading an SVG file. |
| 00:42 | | Attilla [Some.Dude@Nightstar-f29f718d.cable.virginmedia.com] has quit [Ping timeout: 121 seconds] |
| 00:44 | | Stalker [Z@Nightstar-3602cf5a.cust.comxnet.dk] has joined #code |
| 00:59 | < Rhamphoryncus> | ergh. #define int long long would break in so many ways x_x |
| 01:01 | < Rhamphoryncus> | Using structs for file formats is pretty nasty too. Can't actually do it in C, but people do it all the time :P |
| 01:37 | < celticminstrel> | Why can't you do it in C? |
| 01:38 | < celticminstrel> | And why would the #define break things? |
| 01:48 | <@ToxicFrog> | celticminstrel: as a rule redefining basic types is bad. |
| 01:49 | <@ToxicFrog> | That said, if your code is such that redefining int as long long actually breaks things, you should probably be using int32_t or similar instead of plain int. |
| 01:49 | <@ToxicFrog> | And you can totally use structs for on-disk formats in C, just don't store pointers in them or you are entering a world of pain. |
| 01:50 | | Vornicus [vorn@ServerAdministrator.Nightstar.Net] has quit [[NS] Quit: ] |
| 01:51 | < Rhamphoryncus> | ToxicFrog: You can use it as a cache for a single process, but there's no formally defined format outside what that one process sees. |
| 01:52 | < celticminstrel> | (Obviously you'd need the #define to occur after any system headers though.) |
| 01:52 | <@ToxicFrog> | Rhamphoryncus: what? |
| 01:52 | < Rhamphoryncus> | If you're ever relying on padding, byte order, or whatever, you're not using C anymore. You're using an extension to C. |
| 01:52 | <@ToxicFrog> | I think we're using 'struct' and 'file format' to mean different things. |
| 01:52 | <@ToxicFrog> | Aah. |
| 01:52 | <@ToxicFrog> | Yeah, using structs for portable file formats is madness. |
| 01:53 | < Rhamphoryncus> | The only time I could imagine doing it is when there's a substantial performance benefit to mmapping the file and using it directly |
| 01:55 | <@ToxicFrog> | This is basically Every DOS Program Ever, for starters~ |
| 01:56 | | Vornicus [vorn@ServerAdministrator.Nightstar.Net] has joined #code |
| 01:56 | | * Rhamphoryncus mumbles about nazi |
| 01:56 | < Rhamphoryncus> | *s |
| 01:58 | < gnolam> | That would be GodWindows, not DOS. |
| 01:59 | < Rhamphoryncus> | I don't even know what the performance of mmap is on DOS. It's not even the same mmap. |
| 02:01 | <@ToxicFrog> | Does DOS even have mmap? |
| 02:01 | <@ToxicFrog> | But what I mean is, this technique is ugly but so, so convenient, and will work fine if all of the following conditions are met: |
| 02:01 | <@ToxicFrog> | - you're using C and can't or don't want to upgrade to a HLL |
| 02:02 | <@ToxicFrog> | - you're releasing binaries, not source, and (if you're releasing binaries for multiple platforms at all) binaries for different platforms won't need to share data |
| 02:02 | <@ToxicFrog> | - you don't plan to make significant changes or upgrades to the program in the future, or if you are, you don't care about preserving the old files |
| 02:02 | <@ToxicFrog> | Which used to be really common. |
| 02:08 | < Rhamphoryncus> | I guess |
| 02:09 | < Rhamphoryncus> | Most of my experience is on linux and with a fairly good understanding of what the specs actually mean. I can't fathom not putting a little bit of extra work into correct, portable code |
| 02:09 | < McMartin> | TF: DOS doesn't have mmap because DOS doesn't, per se, have processes -_- |
| 02:10 | < McMartin> | Unless you mean "mapping files", in which case, DOS tended to require hand-managing various sorts of 16-bit memory extension jackassery. |
| 02:10 | <@ToxicFrog> | Rhamphoryncus: yeah, so, first of all you're working on code that can reasonably be assumed to be cross-platform, and secondly, you have the luxury of writing code that will run on an actual OS. |
| 02:11 | <@ToxicFrog> | Neither of those hold if you're targeting DOS. |
| 02:11 | < Rhamphoryncus> | I'd do the same if I was on dos or even an embedded platform |
| 02:12 | < Rhamphoryncus> | No matter what size char is ^_^ |
| 02:12 | | Derakon[AFK] is now known as Derakon |
| 02:12 | < McMartin> | char is always one byte. |
| 02:12 | < McMartin> | char is not, however, always one character. -_- |
| 02:14 | < Rhamphoryncus> | How many bits is it? Hell, is it even bits? |
| 02:17 | < McMartin> | Every machine C has ever targeted has had an 8-bit byte. |
| 02:18 | < Rhamphoryncus> | bzzt, sorry, but thanks for playing :) |
| 02:18 | < McMartin> | Non-8-bit-byte machines haven't, to my knowledge, been mass-marketed since the Intellivision. |
| 02:18 | < McMartin> | Which C did not in fact target. |
| 02:18 | < McMartin> | (The PDPs don't count; they had 10-bit ROMs, but the RAM was 16-bit words.) |
| 02:19 | < Rhamphoryncus> | "mass-marketed" is the key word there. They're obscure, but they exist. |
| 02:19 | < McMartin> | The Intellivision was not obscure. |
| 02:20 | < Rhamphoryncus> | Afaik there's still such machines in production today for embedded market |
| 02:20 | < McMartin> | If you're writing C and targeting embedded, you should be targeting an ARM anyway. |
| 02:20 | < Rhamphoryncus> | http://en.wikipedia.org/wiki/36-bit is an example of a 9-bit byte |
| 02:20 | < Vornicus> | What drug addict builds computers where 8 bits is not a byte |
| 02:20 | < McMartin> | You can't address bits in C, anyway. |
| 02:20 | < McMartin> | (bitfields do not address fields, they translate to math.) |
| 02:21 | < McMartin> | In fact, on a lot of architectures, you can't sensibly address bytes. |
| 02:21 | < Rhamphoryncus> | http://stackoverflow.com/questions/2098149/what-platforms-have-something-other-t han-8-bit-char |
| 02:22 | < McMartin> | Yeah, it's pretty much safe to say that if they do that you're within your rights to blame them for running it. |
| 02:22 | < Rhamphoryncus> | McMartin: true, although there are of course bitwise ops |
| 02:22 | < McMartin> | Unless they're being retarded by saying "Win32 doesn't have an 8-bit char!" |
| 02:22 | < McMartin> | Which it doesn't, which is why you use TCHAR everywhere in Win32 code, not char. |
| 02:22 | < McMartin> | Because "char" is an 8-bit value. |
| 02:23 | < Rhamphoryncus> | My point is that it's a language variant to rely on 8-bit bytes. Which I don't mind in this case, I just keep in mind that it is a language variant. |
| 02:23 | < McMartin> | No, that's not something to keep in mind. |
| 02:23 | < McMartin> | If you're worrying about that, you're not worrying about things that actually matter. |
| 02:24 | < McMartin> | Alternately! |
| 02:24 | < McMartin> | "This is a program that takes UTF-8 input." |
| 02:24 | < McMartin> | Done. |
| 02:24 | < Rhamphoryncus> | I worry about all the fine print of C's defined/unspecified/implementation defined/undefined. |
| 02:24 | < McMartin> | UTF specifies that it gets its input as octets. |
| 02:25 | < McMartin> | Yes, and you shouldn't, because if someone needs to worry about it because it actually breaks something, porting it is their problem, not yours. |
| 02:25 | < McMartin> | Unless you have a 9-bit machine you're working on, in which case, use assembler. |
| 02:26 | < McMartin> | Worrying about whether chars are octets is an entirely different kettle of fish from worrying about whether chars are signed. |
| 02:26 | < McMartin> | The latter is something you can only rely on if you're fixing on two compilers. |
| 02:26 | < Rhamphoryncus> | this one case is not a big concern, but it helps understand the general insanity that is C types |
| 02:26 | < McMartin> | The thing is, you totally can fix on two compilers. |
| 02:27 | < McMartin> | Furthermore, if those two compilers disagree on something (C++ exception declarations, say), then you don't get to use that feature in portable code even if the standard says you can. |
| 02:27 | | * Rhamphoryncus nods |
| 02:27 | < McMartin> | (Those two compilers are, of course, cl.exe and gcc) |
| 02:29 | | * Rhamphoryncus ponders giving his language an octets type rather than a bytes type. (In name only. I have no intention of allowing non-8-bit bytes.) |
| 02:29 | < McMartin> | If you give a shit, define them in bit width. |
| 02:29 | < McMartin> | s8, i8, s16, u16, etc. |
| 02:29 | < McMartin> | Then have "int" be "whatever is fastest on the target platform" |
| 02:29 | < McMartin> | Lo, you have now reinvented stdint.h |
| 02:30 | | Kindamoody[zZz] is now known as Kindamoody |
| 02:30 | < Rhamphoryncus> | Oh I don't mean individual integers. I mean an array type for networking and file I/O. |
| 02:31 | < gnolam> | *cough*icc*cough* |
| 02:32 | < gnolam> | (ICC == GCC /for the most part/, except when it's extremely annoying if it's not.) |
| 02:33 | | gnolam [lenin@Nightstar-202a5047.priv.bahnhof.se] has quit [[NS] Quit: Z?] |
| 02:35 | < McMartin> | (This is why you don't have to fixate on icc~) |
| 02:38 | < McMartin> | Rham: Anyway, the main issue here is that you're more or less allowed to say "if you target something insanely exotic, it's up to you to make the runtime behave sensibly." |
| 02:39 | < McMartin> | A machine with 9-bit characters is going to have to have a bit-inserter inside the network driver, because the other machines out there sure as hell won't be serving up 9-bit chars, they'll be serving up UTF-8. |
| 02:39 | < McMartin> | At least if they know what's good for them. |
| 02:39 | < Rhamphoryncus> | Yes, but you have to know what insanely exotic is first. Most people stop at "it works for me". |
| 02:39 | < McMartin> | Yeah, so: x86, x64, PPC, ARM, THUMB~ |
| 02:39 | < Rhamphoryncus> | (people will argue that "x[i++] = i++;" is okay.) |
| 02:40 | < McMartin> | (That's language, not machine) |
| 02:40 | < McMartin> | (That *can* be OK if you define evaluation order strictly enough) |
| 02:40 | < Rhamphoryncus> | in C :P |
| 02:40 | < McMartin> | (Which doesn't!) |
| 02:41 | | * McMartin doesn't recall if Scheme guarantees left-to-right eager evaluation of arguments. |
| 03:02 | < celticminstrel> | What's icc? |
| 03:14 | < Vornicus> | http://en.wikipedia.org/wiki/Intel_C%2B%2B_Compiler |
| 05:19 | < celticminstrel> | I need to force a redraw of the window immediately so that I can eliminate the selection highlight from the output bmp. |
| 05:19 | < celticminstrel> | Is there a way to do this? Unsurprisingly, glutPostRedisplay+glFlush does notwork. |
| 05:19 | < celticminstrel> | (Technically I just need to update the buffer that glReadPixels reads from.) |
| 05:22 | < celticminstrel> | ^not work |
| 05:23 | < celticminstrel> | In other news, the GLenum typedef is the worst thing ever invented. |
| 05:23 | < celticminstrel> | Or one of them. |
| 05:25 | < celticminstrel> | Maybe just call my drawing functions... why didn't I think of that... |
| 05:26 | < celticminstrel> | Yup, that worked. Feel free to ignore this monologue! |
| 05:32 | < celticminstrel> | Oddly, the bmp output is much darker than the original. No idea why that is, so I'll probably just have to leave it. |
| 05:38 | | Janus [NSwebIRC@Nightstar-f1b7ad82.res.rr.com] has joined #code |
| 05:38 | | kwsn is now known as kw-slee-n |
| 05:45 | | Stalker [Z@Nightstar-3602cf5a.cust.comxnet.dk] has quit [Ping timeout: 121 seconds] |
| 06:04 | | celticminstrel [celticminst@Nightstar-5d22ab1d.cable.rogers.com] has quit [Connection closed] |
| 06:26 | | Derakon is now known as Derakon[AFK] |
| 06:57 | | Vash [Vash@Nightstar-f03c5637.sd.cox.net] has joined #code |
| 07:15 | < Janus> | okay. World generation. Putting that on the Hard list |
| 07:17 | < Janus> | Also one of my pals convinced me to drop the minecraft clone. All it took was 5 back to back videos of clones already in progress, d'oh. |
| 07:17 | | * Vornicus patpats |
| 07:17 | < Vornicus> | It might not be a good, you know |
| 07:17 | < Janus> | At least it's only 2 weeks of code going down the drain |
| 07:17 | < Vornicus> | marketable idea |
| 07:17 | < Vornicus> | But I bet you learned a lot. |
| 07:18 | < Janus> | Yeah, I suppose! |
| 07:21 | < Janus> | I'm going back to the 2D ... ecosystem thingy I think I showed you eons ago. Not sure how I'm gonna make it into a 'game' though |
| 07:26 | < Janus> | And yes. Whenever I told anyone about it, it always had this air of "... but let me explain how it's different please" bleck |
| 07:38 | < Vornicus> | heh |
| 08:02 | | AnnoDomini [annodomini@60F158.737D66.CA1918.CC7562] has joined #code |
| 08:14 | < Janus> | I got. Um. Something so far. Something tells me if I understood basic math, this would be easier. http://goo.gl/yiz79 |
| 08:16 | < Vornicus> | That.... |
| 08:17 | < Vornicus> | Okay that appears to be a quairandom walk with curvature dependent on distance from the center. |
| 08:17 | < Janus> | ... that is exactly what it is. You're good! xD |
| 08:18 | < Vornicus> | Not knowing what you're aiming at I can't really give much advice. |
| 08:20 | < Janus> | Um! Well, I guess I'm after uh. That line is supposed to represent the beginning surface. I think after that, I can have tunnels be wormed through out it, and do erosion stuff. I want it to sorta be like a discworld thing. Where the mountains are in the middle, with plains around it maybe |
| 08:21 | < Janus> | There will be a space turtle and some elephants under it. |
| 08:22 | < Janus> | Most of the stuff I know now though, is for height maps. Which wouldn't make very interesting terrain |
| 08:24 | | * Janus has no idea how to mathimatically convey it though. What you seen was just me sticking a bunch of random/sin functions together until it stopped not working |
| 08:24 | | Vash [Vash@Nightstar-f03c5637.sd.cox.net] has quit [[NS] Quit: I <3Lovecraft<3 Vorn!] |
| 08:25 | < Vornicus> | Hm... |
| 08:29 | | * Janus looks at a terraria map. ... f-for reference! |
| 08:30 | < Vornicus> | Okay got a different one for you |
| 08:30 | < Vornicus> | Quasirandom walk |
| 08:31 | < Vornicus> | next slope is limited by two things: 1. a maximum slope based on distance from edge |
| 08:31 | < Vornicus> | 2. a curvature limit. |
| 08:32 | < Vornicus> | Has a weakness: your curvature limit means you can't get pointy bits. |
| 08:34 | < Janus> | Hmmm... *tries to do that* |
| 08:35 | < Vornicus> | Hm. Maximum angle might be better, max it out at about 100deg and you'll have yourself the possibility of some really awesome cliffies |
| 08:37 | < Vornicus> | Oh oh oh. Small chance - higher towards the center - that you get a sudden turn towards sanity. |
| 08:38 | < Janus> | How's this? http://goo.gl/Fp77V |
| 08:39 | < Vornicus> | THat has potential. What the crap is going on at that point about 40% along? |
| 08:39 | < Janus> | Beats me. So far all of them have had something pha--er... something weird jutting out like that |
| 08:40 | < AnnoDomini> | Looks like one of those rock formations in America I once saw in a western. |
| 08:40 | < Vornicus> | Yeah, it's a hoodoo all right |
| 08:41 | < Vornicus> | But was this my method or something? |
| 08:43 | < Janus> | Yeah, I put a limitation on the slope max, and also put a limit on the maximum angle to 1.1 to -1.1 radians |
| 08:44 | < Vornicus> | Maximum angle here is the curvature? |
| 08:45 | < Janus> | Yeah. I'm imposin' a bunch of them back to back to see if any of them turn out strange |
| 08:49 | < Janus> | http://dl.dropbox.com/u/3108480/abunchofnonsense.png |
| 08:49 | | * Vornicus wonders if it's possible to impose a limit on the maximum change in curvature without busting the slope demands. |
| 08:50 | < Vornicus> | Oh, very nice. |
| 08:51 | < Janus> | I've looked at the terraria maps though, and I've noticed it's almost always a straight heightmap |
| 08:51 | < AnnoDomini> | Are you doing this on a plotter? |
| 08:52 | < Janus> | The only points where it's not, are pretty isolated. Maybe it's doing the 'whole digging' as a seperate step, even for the surface. |
| 08:52 | < Janus> | Oh, nono! I'm just putting red pixels at every vertice in the graph |
| 08:53 | < Vornicus> | Caves are generally done in this by building random graphs |
| 08:53 | < Vornicus> | Der had some interesting work in that field |
| 08:55 | < Janus> | Would they be the same as the random walking thing? Though, without the bias to go right of course. Maybe with a random width and length too |
| 08:55 | < AnnoDomini> | Janus: The labels look handwritten. |
| 08:56 | < Janus> | Yeah, it's easier to just draw it with my tablet than mess with photoshops text thingy |
| 08:56 | < Vornicus> | Janus -- it wouldn't have any of the characteristics of a random walk, no. |
| 08:58 | < Janus> | Oh, who's Der, by the way? |
| 08:58 | < Vornicus> | Derakon |
| 08:58 | < Janus> | Oh! ... duh. |
| 09:03 | < Janus> | It seems to get more and more unstable the longer it goes on though, it seems. |
| 09:04 | | * Janus tacked a zero to how long the graph needs to progress |
| 09:04 | < McMartin> | Heh |
| 09:04 | < McMartin> | When in doubt, multiply by ten~ |
| 09:06 | < Vornicus> | Make your angle condition scalefree |
| 09:06 | < Vornicus> | So the amount of distance you get off a single segment is in direct proportion to the amount of angle you can get off it. |
| 09:07 | < AnnoDomini> | Janus: Put it on a logarithmic scale! :V |
| 09:13 | < Janus> | Oh, that's much better! ... it'll take some, um, processing before I can show. I increased it 100 times. And. The graph I made sucks, so it had to be 102400 pixels wide too |
| 09:27 | < Janus> | ... I'm gonna make some tea |
| 09:28 | | * Janus accidently photoshop. It's now throwing an error box every second that it's out of memory. It will do this 200 more times, because that's how many frames there are, and it doesn't know when to give in |
| 09:29 | < Janus> | ... oh nevermind, it stopped at 180/200 and crashed. Well. Just take my word for it, it works. :[ |
| 09:30 | < Janus> | Also, instead of proximity to the edge, I replaced it with biome stuff, so it gets jaggy and plainy based on that instead |
| 09:30 | | Kindamoody is now known as Kindamoody|out |
| 09:31 | | Attilla [Some.Dude@Nightstar-f29f718d.cable.virginmedia.com] has joined #code |
| 09:31 | < Janus> | Whoohoo! That's down then, for now! Now I can do caves! |
| 09:33 | < Janus> | And... after that. ... um. Figure out how to fill it in with rigid bodies. That will be, uh, fun. |
| 09:34 | | gnolam [lenin@Nightstar-025a7b1e.dialup.ice.net] has joined #code |
| 09:39 | | * Vornicus wants to seeeeeeeeee |
| 09:41 | | AnnoDomini [annodomini@60F158.737D66.CA1918.CC7562] has quit [[NS] Quit: leaving] |
| 09:50 | < gnolam> | Vornicus: been binging on wood alcohol? |
| 09:54 | < Vornicus> | ...whut |
| 09:57 | | Vornicus is now known as Vornicus-Latens |
| 10:00 | < Janus> | There we go, finally! http://dl.dropbox.com/u/3108480/SurfaceScanTestHuge.gif |
| 10:02 | < Vornicus-Latens> | glory |
| 10:05 | < Janus> | Thanks! Got more done tonight than I would've ever thouzzz |
| 10:06 | | Janus [NSwebIRC@Nightstar-f1b7ad82.res.rr.com] has quit [[NS] Quit: zzz] |
| 10:06 | | AnnoDomini [annodomini@FFB3F3.4C5BE8.2014E2.DC0864] has joined #code |
| 10:18 | < gnolam> | Yay, the revised version actually works on the radiotherapy computers as well. \o/ |
| 10:48 | | Phox [Phox@Nightstar-3e5deec3.gv.shawcable.net] has joined #code |
| 10:49 | | Fantastic_Phox [Phox@Nightstar-3e5deec3.gv.shawcable.net] has quit [Ping timeout: 121 seconds] |
| 10:50 | < gnolam> | Phox: shouldn't that be "Phantastic"? ;) |
| 11:18 | | kw-slee-n is now known as kwsn |
| 11:58 | | AnnoDomini [annodomini@FFB3F3.4C5BE8.2014E2.DC0864] has quit [[NS] Quit: Away!] |
| 12:28 | | Rhamphoryncus [rhamph@Nightstar-14eb6405.abhsia.telus.net] has quit [Client exited] |
| 12:41 | < Tarinaky> | Okay. Stupid question. What's a 'plain association' (in the context of aggregation and composition)? |
| 12:42 | < Tarinaky> | This book doesn't seem to say. |
| 13:06 | | gnolam [lenin@Nightstar-025a7b1e.dialup.ice.net] has quit [[NS] Quit: Je suis arrivé] |
| 13:15 | < TheWatcher> | Tarinaky: association is "Foo uses Bar", or "Foo is related to Bar", it's the least specific way of talking about the interaction of two objects |
| 13:18 | | Stalker [Z@Nightstar-3602cf5a.cust.comxnet.dk] has joined #code |
| 15:12 | | You're now known as TheWatcher[afk] |
| 16:38 | | Stalker [Z@Nightstar-3602cf5a.cust.comxnet.dk] has quit [[NS] Quit: If the world didn't suck, we'd all fall off.] |
| 16:57 | | Derakon [chriswei@Nightstar-f68d7eb4.ca.comcast.net] has joined #code |
| 16:58 | < Derakon> | So I added these two files to my project and now atcore.h (a third party library) is throwing syntax errors at me. I'm guessing I screwed something up. Any ideas? http://pastebin.com/TDTNG02x |
| 17:07 | < Derakon> | I guess it'd help to show atcore.h too. Here; first error is on line 12 with the AT_WC identifier. http://pastebin.com/xxHMt0nm |
| 17:13 | < Derakon> | ...okay, that's weird. Replacing memhandler.h with nothingness and memhandler.c with just "#include "atcore.h"" still gets me the compile errors. What the heck. |
| 17:20 | < Tarinaky> | Why does the recommended book on UML for my course recommend using 'plain association' instead of Aggregation and Composition? |
| 17:21 | < Tarinaky> | Surely being more specific when you have the power to do so is a good thing... otherwise you'd never be specific. |
| 17:34 | <@ToxicFrog> | Derakon: what's the actual error? |
| 18:02 | | Stalker [Z@Nightstar-5aa18eaf.balk.dk] has joined #code |
| 18:11 | < Derakon> | TF: "syntax error : identifier 'AT_WC'" |
| 18:22 | | Stalker [Z@Nightstar-5aa18eaf.balk.dk] has quit [Client closed the connection] |
| 18:22 | | Stalker [Z@Nightstar-5aa18eaf.balk.dk] has joined #code |
| 18:25 | <@ToxicFrog> | o.O |
| 18:25 | <@ToxicFrog> | What compiler? |
| 18:26 | < Derakon> | VS2008's nmake. |
| 18:42 | < Derakon> | ...okay, that's a novel error. "cp: /Volumes/Users/Public/Documents/neo-dev/test.bat: Cannot allocate memory" |
| 18:42 | < Derakon> | (Trying to copy files off of a fileshare) |
| 18:45 | < Derakon> | Also really irritating, since I'm trying to get the updated files into my local Mercurial repo. |
| 18:51 | < Derakon> | And now I've somehow broken the fileshare entirely so I can't even mount the external drive. |
| 18:51 | < Derakon> | Fuck. |
| 19:22 | | Kindamoody|out is now known as Kindamoody |
| 19:22 | | * Derakon fixes the AT_WC identifier error via "#include "wchar.h"" before including atcore.h. WTF. |
| 19:23 | < Derakon> | I'm still getting "syntax error : 'string'" elsewhere in atcore.h though, which is weird because there aren't any strings! |
| 19:23 | < Derakon> | Just character arrays! |
| 19:23 | < Tamber> | But... that is a string... |
| 19:30 | < Derakon> | No, it's a wchar_t*. |
| 19:30 | < Derakon> | A string is a std::string. |
| 19:30 | < Derakon> | Or a Boost string or something. |
| 19:30 | < Derakon> | I.e. something actually sensible for handling text data. |
| 19:33 | < Tamber> | Welcome to C; a string is a char* |
| 19:33 | < Tamber> | :) |
| 19:33 | < Tamber> | Please deposit your sanity in the basket by the door. =p |
| 19:34 | < Derakon> | Allow me to quote from a GameDev.net sig: "If you are asking a question and your code contains "char*" then you have a bug. std::string roxxors the big one one one." |
| 19:35 | < Tamber> | Aheh. |
| 19:35 | < Tamber> | If you're using C++, maybe. |
| 19:37 | < Tamber> | If your header files have .h on them, most likely you're dealing with C (I believe the .h ext. was deprecated for C++); in which case, a string may also be an array of characters. |
| 19:38 | < Tamber> | In which case, the sig is completely and utterly irrelevant. :) |
| 19:39 | < Derakon> | That still does nothing to explain why the compiler is giving "syntax error : 'string'" when "string" doesn't appear in the file. |
| 19:39 | < Tamber> | I didn't say it did. =p |
| 19:39 | < Derakon> | So basically you'd rather snark than be useful. :P |
| 19:39 | < Tamber> | ...wait, hang on. |
| 19:39 | < Tamber> | You're running C code through a C++ compiler? |
| 19:39 | < Tamber> | Is 'string' a reserved word? |
| 19:40 | < Derakon> | It'd better not be. |
| 19:40 | < Derakon> | As I said earlier, it's "std::string". |
| 19:40 | < Tamber> | Hum. |
| 19:40 | | * Tamber shrugs. Doesn't use C++. |
| 19:41 | < Tamber> | But, anyway~ |
| 19:44 | < Tamber> | Hmmm. So 'string' doesn't appear anywhere in atcore.h? Does it tell you where this syntax error occurs? |
| 19:44 | < Derakon> | Yes, on line 75. http://pastebin.com/xxHMt0nm |
| 19:45 | < Derakon> | (And "String" does occur, but even if you replace all instances of it with other words the same errors occur) |
| 19:45 | < Tamber> | *headscratch* |
| 19:58 | | You're now known as TheWatcher |
| 20:11 | < TheWatcher> | Dera: ... maybe it's complaining about the "C", in the 'extern "C" __devlspec(dllexport)' it's replacing AT_EXP_MOD with... although I have no idea why it should, as VS2008 should allow that |
| 20:23 | | Kindamoody is now known as Kindamoody[zZz] |
| 20:34 | < Derakon> | TW: removing the "extern "C"" allows compilation to complete, but will it work... |
| 20:36 | < kwsn> | guilty pleasures: finding ways to reduce how much code is used yet still doing the exact same functionality |
| 21:03 | < TheWatcher> | Derakon: ... |
| 21:03 | < TheWatcher> | I think my only response to that can be "Bwuh" |
| 21:15 | | SchoolPhox [ccae3c5f@Nightstar-ef290c21.mibbit.com] has joined #code |
| 21:15 | < SchoolPhox> | So, riddle me this. I've got a matlab program, and part of it involves filling an array inside a for loop. |
| 21:16 | < SchoolPhox> | I'm told that if I initialize the array outside the for loop, then populate it inside, this cuts down on the computational overhead |
| 21:17 | < SchoolPhox> | That's compared against each iteration, adding another column to my matrix |
| 21:17 | < SchoolPhox> | Does matlab burn clock cycles when it has to redefine matrices? |
| 21:17 | < Tamber> | Could it involve reallocating memory every time? |
| 21:18 | < SchoolPhox> | Dunno. That's kind of what I'm asking. |
| 21:18 | < SchoolPhox> | Are Matlab matrices conticuous blocks of data? |
| 21:18 | < SchoolPhox> | *contiguous |
| 21:19 | < Derakon> | If it wants to be remotely efficient, then it'll be allocating the array as a single contiguous block, yes. |
| 21:20 | < Derakon> | This should be the kind of thing you can easily test, though. Just do 10k iterations of "allocate an entire array" vs. 10k iterations of "allocate array one column at a time". |
| 21:20 | < Derakon> | And time it and see which one's faster. |
| 21:20 | < SchoolPhox> | Ah, yeah, I suppose I could |
| 21:20 | | * SchoolPhox goes to write a quick script for that kind of thing |
| 21:21 | < SchoolPhox> | Does Matlab feature some kind of timer or clock cycle counter? |
| 21:22 | < Derakon> | Ah ha, fixed my compilations problems by changing the filename from memhandler.c to memhandler.cxx. |
| 21:22 | < Derakon> | Go go magic filename-based behaviors! |
| 21:24 | | SchoolPhox [ccae3c5f@Nightstar-ef290c21.mibbit.com] has quit [[NS] Quit: http://www.mibbit.com ajax IRC Client] |
| 21:27 | | SchoolPhox_ [ccae3c5f@Nightstar-4fab16c5.mibbit.com] has joined #code |
| 21:27 | < SchoolPhox_> | 10K was too small, so I went bigger |
| 21:27 | < SchoolPhox_> | Too big, I guess, because I locked up the machine |
| 21:27 | | Stalker [Z@Nightstar-5aa18eaf.balk.dk] has quit [[NS] Quit: If the world didn't suck, we'd all fall off.] |
| 21:27 | | SchoolPhox_ is now known as SchoolPhox |
| 21:27 | < Derakon> | Heh. Well done. |
| 21:27 | < Derakon> | You should probably free the array between each iteration. :) |
| 21:27 | < SchoolPhox> | Isn't there an emergency bailout button in Matlab? |
| 21:28 | < SchoolPhox> | I thought I remember hearing something about it |
| 21:28 | < SchoolPhox> | Nah, this was just testing the time with a pre-initialized array |
| 21:28 | | * SchoolPhox deletes some of those trailing zeros |
| 21:28 | < SchoolPhox> | .... 1 billion |
| 21:28 | < SchoolPhox> | That's what I used. oops |
| 21:28 | < Derakon> | Careful with those orders of magnitude. |
| 21:31 | < Derakon> | Now I just have to figure out why the SWIG wrapper for my custom mini-library isn't finding the functions in it... |
| 21:32 | | SchoolPhox [ccae3c5f@Nightstar-4fab16c5.mibbit.com] has quit [[NS] Quit: http://www.mibbit.com ajax IRC Client] |
| 21:38 | | SchoolPhox_ [ccae3c5f@Nightstar-36f67fd0.mibbit.com] has joined #code |
| 21:39 | < SchoolPhox_> | Yus, verified. I don't really want to time it, but we're talking less than a second at 100K, versus 10 seconds~~ without a pre-initialized matrix |
| 21:40 | < Derakon> | :) |
| 21:40 | < SchoolPhox_> | It only required 3 crashes, too. |
| 21:40 | < Derakon> | Heh. |
| 21:41 | < SchoolPhox_> | I think I asked about a bailout button, but I crashed before I saw a response |
| 21:41 | < SchoolPhox_> | Is there one? |
| 21:41 | < Derakon> | You did, and I don't know. |
| 21:41 | < Derakon> | I try to avoid using Matlab if I can. |
| 21:41 | < SchoolPhox_> | Oh yeah? What do you use? |
| 21:41 | < Derakon> | Python. |
| 21:42 | < SchoolPhox_> | Ah. I've been looking at Sage a bit. It's supposed to be good, but I haven't had the time to start playing with hit |
| 21:42 | < SchoolPhox_> | *it |
| 21:42 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has quit [Ping timeout: 121 seconds] |
| 21:42 | | Reiver [orthianz@3CF3A5.E1CD01.C6689C.33956A] has joined #code |
| 21:43 | | * Derakon looks it up. |
| 21:43 | < Derakon> | Heh, built on Python. |
| 21:44 | < SchoolPhox_> | I wouldn't doubt it. Python does it well, and I think Sage incorporates existing stuff when it can |
| 21:47 | < Derakon> | There's an extensive community dedicated to reimplementing Matlab in Python, more or less. |
| 21:47 | < Derakon> | It'd be silly to ignore them. |
| 22:09 | | kwsn is now known as kw|GO_BREWERS |
| 22:14 | < Derakon> | Woo, now I'm calling the API successfully! |
| 22:14 | < Derakon> | And getting error codes back! |
| 22:15 | < Tamber> | Progress! |
| 22:16 | < SchoolPhox_> | You know, only in programming is an error a good thing |
| 22:18 | < Derakon> | It's the right kind of error! |
| 22:18 | < Tamber> | It is when it means you're getting to the point that it's /talking/ to you. =p |
| 22:18 | < Derakon> | As opposed to the one where the system is all "what the heck you aren't even speaking comprehensibly let alone making sense" |
| 22:21 | | Rhamphoryncus [rhamph@Nightstar-14eb6405.abhsia.telus.net] has joined #code |
| 22:24 | | SchoolPhox_ [ccae3c5f@Nightstar-36f67fd0.mibbit.com] has quit [[NS] Quit: http://www.mibbit.com ajax IRC Client] |
| 22:33 | | * Derakon :wtc:s at his code. |
| 22:33 | < Derakon> | My "walk the API through its paces" program works fine. The program that actually tries to use the API is getting errors out, despite AFAICT doing exactly the same things. |
| 22:33 | < Derakon> | So clearly I'm missing something somewher. |
| 22:33 | < Derakon> | Er, somewhere. |
| 23:21 | < Derakon> | Ah. Strange. |
| 23:21 | < Derakon> | The first time I call this function, the string I pass along isn't null-terminated. The second time, it is. |
| 23:21 | < Derakon> | I'm calling it exactly the same way -- I just made the same call twice in a row. |
| 23:23 | < Derakon> | ...because I'm operating with an old version of this file from before I fixed the null-termination issue. Good job Derakon. |
| 23:23 | < McMartin> | A mystery is solved |
| 23:38 | < Derakon> | Hey hey hey! Got a "timed out" error instead of the usual type! \o/ |
| 23:43 | < Tamber> | \o/ |
| 23:53 | < Derakon> | Ugh, initialization of this library can take anywhere from 3 to 30 seconds, even when done successfully. |
| --- Log closed Sat Oct 08 00:00:13 2011 |