| --- Log opened Fri Jun 15 00:00:15 2012 |
| 00:00 | < Rhamphoryncus> | The alternative interpretation is that storing the faces rather than the vertices is Fail, heh |
| 00:08 | < gnolam> | Tamber: I know :o |
| 00:44 | <~Vornicus> | Rham: I am /still/ not seeing where you get enough data to math this up |
| 00:44 | <~Vornicus> | What data precisely do you have. |
| 01:01 | | You're now known as TheWatcher[T-2] |
| 01:03 | | Number3 is now known as ShellNinja |
| 01:07 | | You're now known as TheWatcher[zZzZ] |
| 01:10 | | himi [fow035@D741F1.243F35.CADC30.81D435] has joined #code |
| 01:10 | | mode/#code [+o himi] by ChanServ |
| 01:12 | | cpux [cpux@Nightstar-c5874a39.dyn.optonline.net] has joined #code |
| 01:40 | | Vash [Vash@Nightstar-241cb5d4.wlfrct.sbcglobal.net] has joined #code |
| 01:40 | | mode/#code [+o Vash] by ChanServ |
| 01:44 | <~Vornicus> | Strange results of mathematics: the largest constructible polygon with an odd number of sides has 2^32-1 sides. |
| 01:48 | <~Vornicus> | (and all constructible polygons have a number of sides equal to a power of two times a factor of that same number) |
| 01:49 | < gnolam> | Are you sure that's not "the largest constructible polygon on a 32-bit machine"?~ |
| 01:50 | <@ToxicFrog> | "constructible polygon"? |
| 01:51 | <&McMartin> | You can build it with nothing but compass and unmarked straightedge |
| 01:51 | <~Vornicus> | That. |
| 01:51 | < celticminstrel> | Heh. |
| 01:51 | <~Vornicus> | You can construct a polygon if, when you take out all the 2s, it's a product of distinct fermat primes. |
| 01:51 | < celticminstrel> | @ gnolam |
| 01:53 | <~Vornicus> | (or rather, these are regular polygons) |
| 01:53 | < Rhamphoryncus> | Vornicus: the normal vector is centered at 0,0,0 for the triangle. The triangle is unit length, equilateral, and the main vertex is north (+y or whatever) |
| 01:53 | <~Vornicus> | I don't see how "centering" the normal vector helps you. |
| 01:54 | <~Vornicus> | A normal vector is generally given as a unit vector for easy calculation. |
| 01:54 | < Rhamphoryncus> | the normal is unit |
| 01:54 | < Rhamphoryncus> | The size of the triangle doesn't actually matter, that's why I called it unit length |
| 01:58 | <~Vornicus> | but th... |
| 01:58 | <~Vornicus> | Okay, give me all the data you /have/, we'll figure out what we're looking at |
| 01:59 | < RichyB> | Vornicus, there might be a bigger constructible polygon with an odd number of sides, but only if there's a sixth Fermat prime? |
| 01:59 | < Rhamphoryncus> | The height of the triangle and the slope of the triangle. I can apply the height after though (it shifts the whole thing up), so all that matters here is the slope |
| 02:02 | < celticminstrel> | I keep feeling like you're saying "triangle" when you mean "tetrahedron"... |
| 02:02 | < celticminstrel> | ...but also when you mean "triangle"... |
| 02:02 | < Rhamphoryncus> | no, it's a triangle |
| 02:02 | <~Vornicus> | RichyB: yes. |
| 02:03 | <~Vornicus> | But since it's widely believed that there are no more fermat primes, well |
| 02:03 | <~Vornicus> | Is this a triangle in R^2 or R^3? |
| 02:03 | < Rhamphoryncus> | Imagine you're looking at an orthographic projection from directly overhead. You cannot see the +z dimension. Now draw a triangle on your screen. What are the heights (+z) of the 3 corners? |
| 02:04 | < Rhamphoryncus> | Not sure what that means |
| 02:04 | < celticminstrel> | It's the number of dimensions. |
| 02:04 | < celticminstrel> | ie, 2D or 3D? |
| 02:05 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has quit [[NS] Quit: Leaving] |
| 02:05 | < celticminstrel> | More specifically, is the triangle positioned in three dimensions. |
| 02:05 | < Rhamphoryncus> | It's positioned in 3d but the triangle itself is just a triangle. |
| 02:06 | <~Vornicus> | Okay, so you've got an equilateral triangle in 3 space |
| 02:06 | <~Vornicus> | equilateral, unit triangle. |
| 02:06 | < Rhamphoryncus> | Before applying the slope it's equilateral |
| 02:06 | <~Vornicus> | And you have its normal vector, and one of its points. |
| 02:06 | <~Vornicus> | Oh oh oh, I see |
| 02:07 | <~Vornicus> | You have an isometric grid. |
| 02:07 | | Attilla [Obsolete@Nightstar-245c4d9c.as43234.net] has quit [Ping timeout: 121 seconds] |
| 02:07 | <~Vornicus> | You have a normal vector for each triangle in that grid. |
| 02:07 | <~Vornicus> | You need to generate heights from this. |
| 02:07 | <~Vornicus> | of the vertices. |
| 02:08 | < Rhamphoryncus> | something like that |
| 02:09 | <~Vornicus> | Okay, /this/ we can do. |
| 02:12 | < Rhamphoryncus> | hmm, I can rotate my normal. That simplifies what I need to do |
| 02:12 | <~Vornicus> | It's a pain though because technically your true normal is heavily limited by the normals of surrounding triangles. |
| 02:12 | < Rhamphoryncus> | or maybe solves it? |
| 02:12 | <~Vornicus> | So you may have trouble getting them to line up. |
| 02:13 | < Rhamphoryncus> | Vornicus: forget them. I'm going to average them out as a later step |
| 02:13 | < Rhamphoryncus> | This is more of an experiment to see what it looks like as I increase/decrease the LOD |
| 02:13 | <~Vornicus> | Okay, so you're doing the triangles independently. Got you. |
| 02:14 | < Rhamphoryncus> | It should be sufficient to flatten the 3d normal vector into a 2d one |
| 02:16 | <~Vornicus> | 3d vector to gradient |
| 02:16 | <~Vornicus> | really though it's |
| 02:17 | <~Vornicus> | dot(normal, coordinate) = 0 |
| 02:20 | <~Vornicus> | solve for coordinate.z and you get heights. Problem there though is that neighboring triangles don't have /jack/ to do with each other, so averaging is going to be considerably harder. |
| 02:20 | < Rhamphoryncus> | I'm lazy. Averaging will consist of adding all 6 together and dividing by 6 ;) |
| 02:21 | < Rhamphoryncus> | I'm starting to question the whole approach though |
| 02:21 | <~Vornicus> | But the thing is |
| 02:21 | <~Vornicus> | THat DFW because the positions have nothing to do with each other. |
| 02:21 | < Rhamphoryncus> | doesn't fucking work? |
| 02:21 | <~Vornicus> | Yeah |
| 02:21 | < Rhamphoryncus> | Because they're randomly generated? |
| 02:23 | <~Vornicus> | No, because the positions are only valid as relative to other positions on that same triangle. |
| 02:24 | <~Vornicus> | THey are not, in any sense, global positions. |
| 02:24 | < Rhamphoryncus> | The triangle has an overall height and the x,y are predetermined |
| 02:24 | <~Vornicus> | "overall height"? |
| 02:24 | | Derakon[AFK] is now known as Derakon |
| 02:25 | < Rhamphoryncus> | the height parameter for the face |
| 02:25 | <~Vornicus> | Yeah I didn't have any idea you had that. |
| 02:25 | < Rhamphoryncus> | I've said it multiple times |
| 02:25 | <~Vornicus> | No you didn't. |
| 02:25 | <~Vornicus> | You said you had a normal vector. |
| 02:25 | < Rhamphoryncus> | [18:59:07] <Rhamphoryncus> The height of the triangle and the slope of the triangle. I can apply the height after though (it shifts the whole thing up), so all that matters here is the slope |
| 02:26 | <~Vornicus> | and at that point I was still under the impression you were doing an entirely different thing |
| 02:26 | <~Vornicus> | Okay, so here's your thing: |
| 02:27 | <~Vornicus> | dot(normal, coordinate) = dot(normal, centroid) |
| 02:27 | <~Vornicus> | Solve for coordinate.z and you have your answer. |
| 02:27 | < Rhamphoryncus> | I'll have to look that up |
| 02:27 | <~Vornicus> | that's dot product. |
| 02:28 | <~Vornicus> | By this point in your adventures you should have met it |
| 02:28 | < Rhamphoryncus> | Haven't needed it. Or I did and I found a work around because I wasn't familiar with it ;) |
| 02:28 | < Rhamphoryncus> | My brain was rejecting all math last night |
| 02:28 | < Rhamphoryncus> | Oh, I could also be doing it but not know the proper name |
| 02:29 | <~Vornicus> | dot(a, b) is, in 2 dimensions, a.x * b.x + a.y * b.y |
| 02:29 | <~Vornicus> | it's the summation of the cartesian product of two vectors. |
| 02:30 | | * Rhamphoryncus nods |
| 02:30 | <~Vornicus> | the centroid of your triangle is the average of the coordinates of the corners of the triangle for x, y, and your height for z |
| 02:30 | < Rhamphoryncus> | yeah |
| 02:31 | <~Vornicus> | That should get you pretty far. |
| 02:31 | < Rhamphoryncus> | I'll definitely read up on dot product soon since I'm sure I'll be needing it |
| 02:31 | < Rhamphoryncus> | But maybe you can help answer if this is the wrong approach |
| 02:31 | <~Vornicus> | Dot product is awesome; cross product is even more awesome. |
| 02:31 | <~Vornicus> | I don't know, how are you generating your heights and normals? |
| 02:31 | < Rhamphoryncus> | There's two reasons I can see to store the face rather than the vertices |
| 02:33 | < Rhamphoryncus> | One, vertices have more peculiar edge cases, particularly at the vertices of the global icosahedron (where 5 converge rather than 6) |
| 02:33 | <~Vornicus> | Sensible |
| 02:33 | < Rhamphoryncus> | Two, when downsampling faces to reduce LOD it's natural to average across multiple, but for vertices you're at the corner anyway so it's natural to just take that corner and ignore the rest |
| 02:34 | <~Vornicus> | (also your macro-edge and macro-vertex things aren't single-authority) |
| 02:34 | < Rhamphoryncus> | The second reason I'm really doubting |
| 02:34 | < Rhamphoryncus> | hmm? |
| 02:35 | <~Vornicus> | When considering vertices, vertices along the icosahedron edge and at the vertices are (generally) duplicated. |
| 02:35 | <~Vornicus> | Okay. |
| 02:35 | < Rhamphoryncus> | yeah |
| 02:35 | <~Vornicus> | About downsampling. |
| 02:36 | <~Vornicus> | Do you remember the Most Important Thing about making solid models? |
| 02:36 | | Kindamoody[zZz] is now known as Kindamoody |
| 02:36 | < Rhamphoryncus> | no :) |
| 02:36 | <~Vornicus> | OKay: The Most Important THing about making solid models is... that vertices of adjacent triangles must be coincident. |
| 02:36 | < Rhamphoryncus> | meaning the same spot exactly? |
| 02:37 | <~Vornicus> | Yes. |
| 02:37 | < Rhamphoryncus> | I know the opengl rule about vertices and edges being exactly identical |
| 02:37 | <~Vornicus> | If you cannot guarantee this, you are /doomed/ |
| 02:38 | <~Vornicus> | THere will be holes in your model and those make hulk mad. |
| 02:38 | < Rhamphoryncus> | That's why to calculate a vertex I take the adjacent faces, average them, then apply in the 1, 2, or 6 places it needs |
| 02:40 | <~Vornicus> | I'm not sure it's sensible to /do/ this for face-centric worlds. |
| 02:41 | < Rhamphoryncus> | Averaging that way when I'm just storing the faces? |
| 02:42 | <~Vornicus> | I mean, sure, you could /do/ it, but the real problem then is, well, you've got this... |
| 02:43 | <~Vornicus> | I... |
| 02:43 | < Rhamphoryncus> | I realized that it was smoothing over most of the detail. My random heightmap was rendering much smoother than the detail underneath. That's dropping a good chunk of resolution |
| 02:43 | <~Vornicus> | Right. You'll live, you can't win that fight. |
| 02:44 | < Rhamphoryncus> | That's why I added the slopes. That means I store more resolution but the actual rendering is used fully |
| 02:44 | < Rhamphoryncus> | but it's an ugly way to do it |
| 02:45 | <~Vornicus> | Unless you've got some awesome handtuned LOD mipper, you're not going to get your low-detail stuff to look at all like the high-detail stuff. |
| 02:46 | < Rhamphoryncus> | No, but simply averaging the faces is *crap*. Doing it because it has a "natural" way of downsampling is simply wrong. There's plenty of other ways I can downsample |
| 02:47 | <@Alek> | hrm. |
| 02:47 | < Rhamphoryncus> | If I store vertices directly then all I have to do is the downsampling. The rendering isn't taking any more away |
| 02:48 | <~Vornicus> | I mean |
| 02:48 | <@Alek> | if I have a deck of 40 cards. what are the odds I'll draw 3 specific ones, regardless of order? |
| 02:48 | < Rhamphoryncus> | 1/40 * 1/39 * 1/38 |
| 02:49 | < Rhamphoryncus> | hrm |
| 02:49 | < Rhamphoryncus> | No, that's not right :) |
| 02:49 | < Rhamphoryncus> | 3/40 * 2/39 * 1/38 |
| 02:49 | < Rhamphoryncus> | Nice to have a clue what I'm talking about lol |
| 02:49 | <~Vornicus> | 40C3 = 40! / (3! * 37!) = 40 * 39 * 38 / (3 * 2 * 1) = 20 * 13 * 38 = 980 distinct hands. |
| 02:50 | < Rhamphoryncus> | oh, more than 3 draws? |
| 02:51 | <~Vornicus> | so 1/980 chance you'll get the three cards you want. |
| 02:53 | <@rms> | 9880 |
| 02:54 | <~Vornicus> | er. 9800. |
| 02:54 | <~Vornicus> | I can Add. |
| 02:54 | <@rms> | How am I 80 off? |
| 02:54 | <~Vornicus> | wait, wtf, 9880? |
| 02:55 | <~Vornicus> | ...oh, because I am. |
| 02:55 | <~Vornicus> | 13*19 is 247, not 245 |
| 02:55 | <~Vornicus> | How the hell |
| 02:55 | | * Vornicus can't do math in his head at the moment apparently |
| 02:56 | < Rhamphoryncus> | afk |
| 02:56 | <~Vornicus> | http://www.wolframalpha.com/input/?i=40c3 I mean honestly this is why we have calculators and the internet, why was I trying to do it in my head |
| 02:57 | <@Alek> | o_o |
| 03:16 | < Rhamphoryncus> | back |
| 03:20 | | * Alek teraflops. |
| 03:21 | <~Vornicus> | I honestly do not know whether you'll be able to get -- okay, here's my thought - when you do the detail downsample you'll probably want a sort of gaussian blur to get the right value - this is Expensive though, so you'll want to hold on to those downsamples. |
| 03:21 | < Rhamphoryncus> | yeah |
| 03:22 | < Rhamphoryncus> | Which fortunately isn't hard. |
| 03:25 | < Rhamphoryncus> | Was it you that had the geodesic dome image? |
| 03:26 | < Rhamphoryncus> | oh geodesic grid gives the good results, unlike geodesic dome |
| 03:28 | <@Alek> | heh. nice. http://notalwaysright.com/well-to-do-if-a-bit-daffy/20373 |
| 03:29 | | * Alek snerks mightily. http://notalwaysright.com/sudo-religious/20367 |
| 03:35 | < celticminstrel> | That one's not really funny. |
| 03:36 | < celticminstrel> | The first one was though. |
| 03:37 | | Kindamoody is now known as Kindamoody|working |
| 04:07 | | * Rhamphoryncus goes on a tangent |
| 04:08 | < Rhamphoryncus> | If I took a totally different approach like delaunay triangulation for my map grid then the biggest issue is the lack of gridlines on which to store my vertex data |
| 04:09 | < Rhamphoryncus> | IOW, how do you break down a delaunay triangulation into triangle strips? |
| 04:10 | <&Derakon> | Start in upper-left corner, grab adjacent triangle that has leftmost vertex, repeat? |
| 04:11 | < Rhamphoryncus> | maybe |
| 04:13 | < Rhamphoryncus> | I have an urge to just use a texture.. which'd be square... funny that :P |
| 04:21 | | maoranma [nbarr@Nightstar-a092c26c.pools.spcsdns.net] has joined #code |
| 04:24 | | Noah [nbarr@Nightstar-a092c26c.pools.spcsdns.net] has quit [Ping timeout: 121 seconds] |
| 04:49 | < Rhamphoryncus> | http://www.flickr.com/photos/7861878@N06/3556592941/in/photostream |
| 04:49 | < Rhamphoryncus> | yup, was Vornicus |
| 05:05 | <~Vornicus> | I never got around to doing more with that. I should see what I can pull off. |
| 05:06 | | * Rhamphoryncus has gone in a circle again |
| 05:07 | < Rhamphoryncus> | Probably a reason why people like cubes. Simple, even if not terribly good |
| 05:07 | < Rhamphoryncus> | I'm trying to find a way to simplify this scheme and make it more vertex-oriented than face-oriented |
| 05:08 | <~Vornicus> | I don't really see that vertex orientation is doing you much good right now. |
| 05:08 | < Rhamphoryncus> | You mean the face orientation? |
| 05:08 | < Rhamphoryncus> | Rendering uses vertices, period |
| 05:08 | <~Vornicus> | right. |
| 05:09 | <~Vornicus> | I mean if you're doing what I was planning - a Civ - face orientation is how you store shit |
| 05:10 | < Rhamphoryncus> | Most user stuff will be objects placed at real coordinates |
| 05:10 | <~Vornicus> | but vertex orientation is Where It's At when you want to do terrain generation. |
| 05:11 | < Rhamphoryncus> | But the map itself, to the degree that it's persistent and mutable, needs an ultimate level of detail. It also needs reduced LOD derived from that |
| 05:11 | < Rhamphoryncus> | yeah |
| 05:12 | < Rhamphoryncus> | So.. if I scrap this idea of recursive subdivision of the triangle.. define the grid as the whole planet from step 1, then set *arbitrary* limits on which areas are active.. |
| 05:12 | <~Vornicus> | You want to do fun stuff with LOD? You can do it the hard way with crazy face shit, or you can just blur down your details using vertex data |
| 05:13 | < Rhamphoryncus> | Use fixed point math for object placement, probably with 3 axis to ensure I always have decent precision (or find something better later on) |
| 05:13 | < Rhamphoryncus> | Blur sounds like the way to go |
| 05:15 | < Rhamphoryncus> | I don't need it to be painful. I have certain goals, "OMGWTF" levels of precision in my LOD are not amongst them. Keeping it simple enough that I can throw hardware at it and render enough triangles to make it irrelevant is. |
| 05:16 | < Rhamphoryncus> | Damnit, I have an urge to play spore XD |
| 05:18 | < Rhamphoryncus> | Maybe I'm doing this all wrong |
| 05:18 | < Rhamphoryncus> | Think about this from an opengl point of view |
| 05:19 | < Rhamphoryncus> | What I want is a single global megatexture (upcoming hardware form is PRT - Partially Resident Texture) |
| 05:19 | < Rhamphoryncus> | map areas in and out as needed |
| 05:20 | < Rhamphoryncus> | mipmaps of the entire thing |
| 05:21 | <~Vornicus> | partially resident texture, like, you have a giant sparsely populated texture? |
| 05:22 | < Rhamphoryncus> | It's virtual memory for the GPU |
| 05:22 | < Rhamphoryncus> | You allocate address space for the whole thing, including the mipmaps, and it uses whatever detail is available as well as providing hooks for when you should start loading new spots |
| 05:24 | < Rhamphoryncus> | So, if I take this approach, the implication is there's only *one* problem to deal with: seamless texture edges |
| 05:29 | | * Rhamphoryncus looks at cube maps |
| 05:30 | < Rhamphoryncus> | Gonna go mull it over |
| 05:36 | | iospace is now known as iospacedout |
| 06:06 | | Vash [Vash@Nightstar-241cb5d4.wlfrct.sbcglobal.net] has quit [[NS] Quit: I lovecraft Vorn!] |
| 06:36 | < celticminstrel> | Is it worth it to keep around a surface many times larger than the window? |
| 06:37 | < celticminstrel> | It would be something that's drawn once and then copied from as needed. |
| 06:42 | | Derakon is now known as Derakon[AFK] |
| 06:49 | < Rhamphoryncus> | That's what a megatexture does. On disk you have the full texture but you only load pieces at a time |
| 06:59 | < celticminstrel> | Well, this would be keeping it in memory, not on disk. |
| 06:59 | < celticminstrel> | It's formatted text, basically. |
| 07:01 | < Rhamphoryncus> | err, is this a different subject from what I'm doing? |
| 07:01 | < celticminstrel> | Not sure? |
| 07:02 | < Rhamphoryncus> | I'm talking about my global map in opengl |
| 07:02 | < celticminstrel> | My question had nothing to do with what you were talking about before that. |
| 07:03 | < Rhamphoryncus> | okay. Ignore my response involving megatextures :) |
| 07:04 | < Rhamphoryncus> | Is yours a 2d scrolling background? |
| 07:04 | < celticminstrel> | More or less. |
| 07:05 | < Rhamphoryncus> | I suppose it depends on how much ram it takes relative to the rest of the program |
| 07:05 | < Rhamphoryncus> | As well as how rapidly you'll be switching between areas |
| 07:05 | < celticminstrel> | I don't really know the answer to the ram question. |
| 07:05 | < Rhamphoryncus> | What's the resolution? |
| 07:06 | < celticminstrel> | As for the other question, this can be accessed from anywhere just by pressing F1, but in practice I wouldn't expect it to be accessed frequently. |
| 07:06 | < celticminstrel> | 800x600 currently |
| 07:06 | < Rhamphoryncus> | That's the window. How about this image? |
| 07:06 | < celticminstrel> | Oh. |
| 07:06 | < celticminstrel> | I don't know. |
| 07:06 | < celticminstrel> | Width is the same as the window. |
| 07:07 | < Rhamphoryncus> | Rule of thumb for an uncompressed image is 4 bytes/pixel. 800*600*4=1.9 megs |
| 07:07 | < celticminstrel> | Height is at least ten times the window at the moment, and it might get larger later. |
| 07:07 | < Rhamphoryncus> | So if your image is a hundred windows high then you've got ~200 megs |
| 07:07 | < celticminstrel> | Okay. |
| 07:08 | < Rhamphoryncus> | You're in the grey area in between. Obviously on a typical computer you'd be fine at 200 megs |
| 07:08 | < celticminstrel> | Okay, so quite likely worth it? |
| 07:09 | < Rhamphoryncus> | Well, what'll happen if you tile it is you'll have it on disk in compressed form, which the OS will cache into ram anyway |
| 07:09 | < celticminstrel> | Tile it? |
| 07:09 | < celticminstrel> | What? |
| 07:09 | < Rhamphoryncus> | chop it up into window-sized portions |
| 07:10 | < celticminstrel> | At the moment I'm rendering it on demand. |
| 07:10 | < Rhamphoryncus> | Oh, by surface do you mean an actual hardware surface? opengl or d3d surface? |
| 07:11 | < celticminstrel> | I don't have a separate surface at the moment. I'm rendering it straight to the screen (except for the bits where I'm scaling things down by half). |
| 07:11 | | * Rhamphoryncus nods |
| 07:11 | < Rhamphoryncus> | I'd say it's cleaner to render on the fly |
| 07:11 | < celticminstrel> | Okay then. |
| 07:11 | < Rhamphoryncus> | So if you've already got it implemented.. keep it |
| 07:12 | < celticminstrel> | So now I need to find a way to input a pixel location and get a line index back, |
| 07:12 | < celticminstrel> | So I know where to start rendering. |
| 07:13 | < celticminstrel> | Currently I'm just keeping track of the line number, but that makes for somewhat irregular scrolling. |
| 07:24 | < celticminstrel> | I suppose I could just run through the entire rendering process minus the actual rendering until I get to the onscreen part... though that seems inefficient... |
| 07:28 | < celticminstrel> | Yeah, there's a noticeable lag towards the bottom if I do that. |
| 07:33 | < celticminstrel> | Maybe I can build up a list of line heights as I render... |
| 07:33 | < Rhamphoryncus> | Keep the two nearest lines and interpolate between them? |
| 07:34 | < celticminstrel> | How? |
| 07:36 | < Rhamphoryncus> | no clue :) |
| 07:36 | < Rhamphoryncus> | I really don't understand what you're doing |
| 07:49 | < celticminstrel> | I'm not sure if I understand it myself. :P |
| 07:49 | < celticminstrel> | Anyway bedtime. |
| 07:50 | | 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!] |
| 08:08 | | Kindamoody|working is now known as Kindamoody |
| 08:48 | < froztbyte> | http://developer.ubuntu.com/showdown/ |
| 09:07 | | Attilla [Obsolete@Nightstar-245c4d9c.as43234.net] has joined #code |
| 09:14 | | Rhamphoryncus [rhamph@Nightstar-5697f7e2.abhsia.telus.net] has quit [Operation timed out] |
| 09:14 | | You're now known as TheWatcher |
| 09:32 | | Rhamphoryncus [rhamph@Nightstar-5697f7e2.abhsia.telus.net] has joined #code |
| 09:33 | | himi [fow035@D741F1.243F35.CADC30.81D435] has quit [Ping timeout: 121 seconds] |
| 10:14 | | Atreus is now known as Tarinaky |
| 10:21 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has joined #code |
| --- Log closed Fri Jun 15 10:39:44 2012 |
| --- Log opened Fri Jun 15 10:39:52 2012 |
| 10:39 | | TheWatcher [chris@Nightstar-3762b576.co.uk] has joined #code |
| 10:39 | | Irssi: #code: Total of 26 nicks [6 ops, 0 halfops, 0 voices, 20 normal] |
| 10:39 | | mode/#code [+o TheWatcher] by ChanServ |
| 10:40 | | Irssi: Join to #code was synced in 43 secs |
| 12:10 | | Rhamphoryncus [rhamph@Nightstar-5697f7e2.abhsia.telus.net] has quit [Client exited] |
| 12:26 | | * Tarinaky WTFs at Python. |
| 12:26 | < Tarinaky> | class Menu(ui_abstract.menu.Menu): |
| 12:26 | < Tarinaky> | AttributeError: 'module' object has no attribute 'menu' |
| 12:26 | < Tarinaky> | It's right there! |
| 12:26 | < Tarinaky> | I'm looking at it! |
| 12:28 | < Tarinaky> | Changed name to be less ambiguous |
| 12:28 | < Tarinaky> | class DefaultMenu(ui_abstract.menu.Menu): |
| 12:28 | < Tarinaky> | AttributeError: 'module' object has no attribute 'menu' |
| 12:28 | < Tarinaky> | So that wasn't it :/ |
| 12:28 | < Tarinaky> | ui_abstract has a module called menu... Which has the class Menu. |
| 12:28 | < Tarinaky> | I can't see why it can't 'find' this menu sub-module :/ |
| 12:29 | <&jerith> | Tarinaky: Have you imported it? |
| 12:29 | <&jerith> | Pastebin the code, perhaps? |
| 12:30 | < Tarinaky> | import ui_abstract doesn't work. |
| 12:30 | < Tarinaky> | If I change it to from ui_abstract.menu import Menu it works. |
| 12:30 | <&jerith> | "import ui_abstract.menu"? |
| 12:30 | < Tarinaky> | (changing the reference from ui_abstract.menu.Menu to just Menu ofc) |
| 12:31 | < Tarinaky> | Elsewhere I'm only importing the top level module and everything else is getting imported. |
| 12:32 | <&jerith> | That only works if the top-level module imports the thing you're after. |
| 12:33 | < Tarinaky> | Ah. Interesting. |
| 12:37 | | celticminstrel [celticminst@Nightstar-5d22ab1d.cable.rogers.com] has joined #code |
| 13:33 | | Attilla [Obsolete@Nightstar-245c4d9c.as43234.net] has quit [[NS] Quit: ] |
| 13:45 | | Attilla [Obsolete@Nightstar-245c4d9c.as43234.net] has joined #code |
| 13:58 | < Tarinaky> | What does TypeError: 'builtin_function_or_method' object has no attribute '__getitem__' mean? |
| 13:58 | < Tarinaky> | Because I thought dictionaries had __getitem__... >.> |
| 13:59 | <&jerith> | Tarinaky: What code is throwing that? |
| 14:00 | <&McMartin> | Dictionaries do, but functions that return dictionaries don't until you call them. |
| 14:00 | <&jerith> | Yeah. It's complaining that you're accessing '__getitem__' on a builtin. |
| 14:02 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has quit [Ping timeout: 121 seconds] |
| 14:04 | < Tarinaky> | Well, I deleted the code and just refactored it away anyway >.> |
| 14:04 | < Tarinaky> | But I wasn't using a method to obtain the dictionary. |
| 14:04 | < Tarinaky> | I was indexing a dictionary :/ |
| 14:04 | < Tarinaky> | So I don't know what was wrong. |
| 14:05 | <&jerith> | Where did you get the dict from? |
| 14:06 | <&jerith> | Because "foo = dict; foo['blah'] = 'frobboz'" would do that. |
| 14:06 | <&jerith> | Also, in general, pastebinning the relevant code leads to more specific debugging assistance. :-) |
| 14:22 | < Tarinaky> | I assigned it {} |
| 14:22 | < Tarinaky> | Yeah, but I didn't have a test case. |
| 14:22 | < Tarinaky> | I had a sphegetti of broken python. |
| 14:25 | < iospacedout> | ... |
| 14:30 | < Tarinaky> | I should probably have a better method of refactoring than just breaking it and running it... then pathing the errors as they come up. |
| 14:30 | < Tarinaky> | *patching |
| 14:31 | <&jerith> | Tarinaky: Do you have decent automated test coverage? |
| 14:31 | < Tarinaky> | No. |
| 14:32 | <&jerith> | Because that's *very* useful for refactoring. |
| 14:32 | < Tarinaky> | "Does it run?" |
| 14:32 | <&jerith> | (Useful enough that I consider it mandatory, personally.) |
| 14:32 | < Tarinaky> | If it can get as far as the first sleep without crashing or raising an exception. |
| 14:32 | < Tarinaky> | That more or less calls all of the code in the program anyway. |
| 14:34 | <&jerith> | that "more or less" scares me. |
| 14:34 | < Tarinaky> | I may implement a monkey test that generates random key and mouse events. |
| 14:35 | < Tarinaky> | Yeah, I know. |
| 14:35 | < Tarinaky> | I don't really have a clear enough idea what I'm doing for unittests though. |
| 14:35 | <&McMartin> | I tend to sketch architectures on paper before I write a single line of code. |
| 14:35 | < Tarinaky> | It's just turn into a millstone around my neck. |
| 14:35 | < Tarinaky> | *It;d |
| 14:36 | <&jerith> | Tarinaky: I find test cases often help clear that up, actually. |
| 14:36 | <&McMartin> | I don't personally accept the idea that the test is the spec |
| 14:37 | <&McMartin> | The spec comes first, but the tests *embody* the spec, ideally as directly as possible |
| 14:37 | <&McMartin> | If this doesn't work, then your spec is in some sense wrong. |
| 14:37 | < Tarinaky> | Yeah. I don't have a spec. |
| 14:38 | < Tarinaky> | I have a to do list of features. |
| 14:38 | < Tarinaky> | But no spec. |
| 14:38 | < Tarinaky> | >.> |
| 14:38 | < Tarinaky> | In case anyone cares >.> http://a6.sphotos.ak.fbcdn.net/hphotos-ak-snc7/600308_3854761981059_923582626_n. jpg |
| 14:38 | <&jerith> | I don't really do the test-first thing, because I tend to code bottom-up and rearrange interfaces as I figure out what the shape of the solution is. |
| 14:41 | | Tsubaki [Phox@Nightstar-d1963151.asahi-net.or.jp] has joined #code |
| 14:43 | <&jerith> | Tarinaky: GUI code is hard to unit-test, but I find that a clean separation between GUI and game logic makes testing the latter much easier. |
| 14:44 | <&jerith> | It also makes both sides nicer to work with, because you don't have weird coupling to bite you. |
| 14:47 | | Tsubaki [Phox@Nightstar-d1963151.asahi-net.or.jp] has quit [Client closed the connection] |
| 14:47 | < Tarinaky> | There isn't a whole lot of game logic tbh. |
| 14:48 | | iospacedout is now known as io|driving |
| 14:48 | < Tarinaky> | That's... well... game logic is the easy part :/ |
| 14:49 | <&jerith> | Not if it's buried in the GUI code. ;-) |
| 14:51 | < Tarinaky> | Anyway. Does anyone see a problem with randomly generating pygame keydown and mouse events to test the GUI? |
| 14:54 | < gnolam> | What is it supposed to accomplish? |
| 14:56 | < Tarinaky> | Making people less concerned that my only test is "Does it compile?" |
| 14:57 | | Tsubaki [Phox@Nightstar-ec25ea75.asahi-net.or.jp] has joined #code |
| 15:07 | <&jerith> | Tarinaky: More directed tests are probably better. |
| 15:07 | <&jerith> | Since you can't usefully test state flows with random events. |
| 15:09 | <&jerith> | If you're going to craft events to inject, you may as well make them useful events. |
| 15:15 | <&McMartin> | That said, whitenoise is one useful test as a pathological input |
| 15:15 | <&McMartin> | BUt it can't be the only one |
| 15:28 | <&jerith> | Randomness in tests is dangerous. |
| 15:28 | <&jerith> | Because you get stuff that sometimes fails and sometimes doesn't. |
| 15:30 | <@ToxicFrog> | That's why the test harness records the seed so that it's reproduceable. |
| 15:31 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has joined #code |
| 15:34 | < Tarinaky> | Is it possible to seed the default PRNG exposed as random in python? |
| 15:35 | < Tarinaky> | No. Wait. |
| 15:35 | < Tarinaky> | Stupid question. |
| 15:57 | | * Vornicus is glad he knows how to use Excel. Makes the graph show its tick labels as for instance 3?. |
| 16:17 | | gnolam [lenin@Nightstar-202a5047.priv.bahnhof.se] has quit [[NS] Quit: Reboot] |
| 16:23 | | gnolam [lenin@Nightstar-202a5047.priv.bahnhof.se] has joined #code |
| 16:29 | | Vash [Vash@Nightstar-241cb5d4.wlfrct.sbcglobal.net] has joined #code |
| 16:29 | | mode/#code [+o Vash] by ChanServ |
| 16:44 | | io|driving is now known as iofficespace |
| 16:56 | | maoranma is now known as Noah |
| 17:08 | | McMartin [mcmartin@Nightstar-adac7b01.pltn13.sbcglobal.net] has quit [Ping timeout: 121 seconds] |
| 17:30 | | Tsubaki [Phox@Nightstar-ec25ea75.asahi-net.or.jp] has quit [Ping timeout: 121 seconds] |
| 17:56 | | Tsubaki [Phox@Nightstar-e11eaaee.asahi-net.or.jp] has joined #code |
| 18:07 | | 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!] |
| 19:04 | | iofficespace [Alexandria@Nightstar-e67f9d08.com] has quit [Operation timed out] |
| 19:04 | | franny [fran@Nightstar-e67f9d08.com] has quit [Operation timed out] |
| 19:06 | | franny [fran@Nightstar-e67f9d08.com] has joined #code |
| 19:08 | | iofficespace [Alexandria@Nightstar-e67f9d08.com] has joined #code |
| 19:12 | | Derakon[AFK] is now known as Derakon |
| 19:16 | | Kindamoody is now known as Kindamoody[zZz] |
| 19:43 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has quit [Ping timeout: 121 seconds] |
| 19:52 | | Vornicus [vorn@ServerAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 19:53 | | Vash [Vash@Nightstar-241cb5d4.wlfrct.sbcglobal.net] has quit [Ping timeout: 121 seconds] |
| 20:16 | | Tsubaki [Phox@Nightstar-e11eaaee.asahi-net.or.jp] has quit [Ping timeout: 121 seconds] |
| 21:18 | < iofficespace> | ... T________________________T |
| 21:18 | < iofficespace> | i have to test practically every option in our setup utility... there are over 650 |
| 21:19 | | * iofficespace commits ritual suicide |
| 21:20 | < iofficespace> | mind you some of them are tested already by virtue of being default values |
| 21:25 | | Tsubaki [Phox@Nightstar-a7956243.asahi-net.or.jp] has joined #code |
| 21:27 | <&jerith> | Combinatorics? |
| 21:31 | <&Derakon> | Even if she doesn't have to test every combination, that's a lot of options to test. |
| 21:35 | < iofficespace> | jerith: each setting has to be put at every possible setting it can be, individually (thankfully i don't have to do X! tests) |
| 21:36 | < iofficespace> | there's over 650 settings i have to test, some of which are default |
| 21:36 | | celticminstrel [celticminst@Nightstar-5d22ab1d.cable.rogers.com] has joined #code |
| 21:55 | < Tarinaky> | I'm having trouble with Inkscape. |
| 21:55 | < Tarinaky> | I've set the page at 128x128px |
| 21:55 | < Tarinaky> | And I've turned on a grid. |
| 21:55 | < Tarinaky> | But 1: the grid isn't aligned with the page |
| 21:55 | < Tarinaky> | 2: I can't get my drawing to 'snap to grid' anyway :/ |
| 22:00 | | Tsubaki [Phox@Nightstar-a7956243.asahi-net.or.jp] has quit [Ping timeout: 121 seconds] |
| 22:04 | | Vornicus [vorn@ServerAdministrator.Nightstar.Net] has joined #code |
| 22:04 | | mode/#code [+qo Vornicus Vornicus] by ChanServ |
| 22:06 | < Tarinaky> | Ahah! |
| 22:06 | < Tarinaky> | Figured it out! |
| 22:06 | <~Vornicus> | what have we figured out? |
| 22:08 | < Tarinaky> | How to make snaping work |
| 22:09 | < Tarinaky> | Black is 0x0 right? Which... if I set an Alpha channel to 0x0 would that be invisible or visible? |
| 22:10 | <&jerith> | Being a double-agent wizard who is mean to children? |
| 22:10 | <@Tamber> | Tarinaky: Invisible, I think. |
| 22:10 | < Tarinaky> | Okay. |
| 22:10 | <&jerith> | Alpha is "opacity". |
| 22:10 | < Tarinaky> | So I've drawn a black circle as a vector graphic. |
| 22:10 | <&jerith> | Zero alpha means zero opacity. |
| 22:11 | < Tarinaky> | How do I make a white circle on a black background? |
| 22:11 | < celticminstrel> | I thought alpha was backwards and 255 was fully transparent? |
| 22:11 | < Tarinaky> | Because I want the middle to the the area that's 'visible' and the back to be invisible. |
| 22:11 | < Tarinaky> | Like a dough cutter! |
| 22:11 | < celticminstrel> | Does Inkscape do the path directions thing? |
| 22:12 | < Tarinaky> | (Obviously there's an implicit svg->png step I'm ignoring for now >.>) |
| 22:12 | < celticminstrel> | ie if you want one path to cut out from another, you make their directions opposite. |
| 22:13 | < Tarinaky> | I have no idea how to do that :/ |
| 22:15 | < Tarinaky> | Ah! Found what I needed! |
| 22:20 | < Tarinaky> | Things I dislike about Inkscape: It won't 'failback' in a sensible way if you tell it to open a blank file. |
| 22:20 | < Tarinaky> | Which is silly given it's a plain-text format. |
| 22:33 | | Tsubaki [Phox@Nightstar-daa59e27.asahi-net.or.jp] has joined #code |
| 22:59 | | iofficespace is now known as io|driving |
| 23:16 | | Vash [Vash@Nightstar-e8057de2.wlfrct.sbcglobal.net] has joined #code |
| 23:16 | | mode/#code [+o Vash] by ChanServ |
| 23:17 | | McMartin [mcmartin@Nightstar-3f2d52e0.pltn13.sbcglobal.net] has joined #code |
| 23:17 | | mode/#code [+ao McMartin McMartin] by ChanServ |
| 23:20 | | Netsplit *.net <-> *.split quits: Tsubaki, Orthia, @Vornicus, Tarinaky, Attilla, Namegduf, @Vash, celticminstrel, @McMartin, @rms, (+5 more, use /NETSPLIT to show all of them) |
| 23:23 | | Netsplit over, joins: @ToxicFrog, Tarinaky, sshine, &McMartin, froztbyte, @rms, &Derakon, Attilla, Orthia, EvilDarkLord (+4 more) |
| 23:23 | < Tarinaky> | I meant logical leap >.> |
| 23:23 | <~Vornicus> | [2**a for a in range(8)] |
| 23:23 | < Tarinaky> | I want to start on 8. |
| 23:23 | <~Vornicus> | ah |
| 23:23 | < Tarinaky> | So 8->128 in powers of 2. |
| 23:24 | | PinkFreud [WhyNot@NetworkAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 23:24 | | PinkFreud [WhyNot@NetworkAdministrator.Nightstar.Net] has joined #code |
| 23:25 | | mode/#code [+o PinkFreud] by ChanServ |
| 23:25 | <&Derakon> | Well, 128 is 2^7. |
| 23:25 | <&Derakon> | And 8 is 2^3. |
| 23:25 | <&Derakon> | So [2 ** n for n in xrange(3, 8)] |
| 23:27 | | franny [fran@Nightstar-e67f9d08.com] has quit [Ping timeout: 121 seconds] |
| 23:28 | | Netsplit *.net <-> *.split quits: Orthia, @Vornicus, Tarinaky, Attilla, Namegduf, @Vash, celticminstrel, @McMartin, @rms, EvilDarkLord, (+4 more, use /NETSPLIT to show all of them) |
| 23:30 | | Netsplit over, joins: ToxicFrog, Tarinaky, sshine, McMartin, froztbyte, rms, Derakon, Attilla, Orthia |
| 23:30 | | franny_ [fran@Nightstar-e67f9d08.com] has joined #code |
| 23:30 | | Netsplit over, joins: EvilDarkLord, @Vash, ~Vornicus, Namegduf, celticminstrel |
| 23:30 | | ServerMode/#code [+oaooaoooq ToxicFrog McMartin McMartin rms Derakon Derakon Vash Vornicus Vornicus] by *.Nightstar.Net |
| 23:31 | <&Derakon> | Okay, I can see a bug. |
| 23:31 | <&Derakon> | But what do you think is going wrong? |
| 23:31 | <&Derakon> | And what debugging steps have you taken? |
| 23:31 | < Tarinaky> | Well, I spotted a missing space on line 11. But adding that didn't help. |
| 23:32 | <&Derakon> | What goes wrong, then? |
| 23:32 | < Tarinaky> | I added the print statement on line 10 as 'debugging' and... yeah. |
| 23:32 | < Tarinaky> | I can't really see what else to try. |
| 23:32 | < Tarinaky> | Sample output: |
| 23:32 | < Tarinaky> | Exporting sample_composite.svg at size 128 |
| 23:32 | < Tarinaky> | The system cannot find the path specified. |
| 23:32 | <&Derakon> | And what do you think that means? |
| 23:32 | < Tarinaky> | That either the command string is wrong, or the filename is wrong. |
| 23:33 | < Tarinaky> | But the filename appears right, and so does the inkscape invocation. |
| 23:33 | < Tarinaky> | I'm not 'used' to using Python for shell scripts. |
| 23:33 | <&Derakon> | Are you certain about both? How would you go about testing them? |
| 23:33 | < Namegduf> | 21:51 -!- ZLOK [Z@Off.With.Her.HEAD] has quit [Singularized: If I had a world of my own, everything |
| 23:33 | <&Derakon> | When in doubt, simplify your test case. |
| 23:33 | < Namegduf> | Er... |
| 23:33 | < Namegduf> | Okay, I don't know how that happened. |
| 23:33 | < Tarinaky> | I'm not sure how I'm supposed to simplify 15 lines :/ |
| 23:33 | < Tarinaky> | It's 15 lines. |
| 23:33 | <&Derakon> | It could be 2. |
| 23:34 | <&Derakon> | How would you test that your program invocation is correct? |
| 23:35 | < Tarinaky> | Well, aparently Windows doesn't like c:\Program%20Files any more :/ |
| 23:35 | < Tarinaky> | What's the right thing you injsert on Windows for a space in a filename? |
| 23:35 | <&Derakon> | I believe you can simply quote the entire path. |
| 23:36 | <&Derakon> | E.g. "C:\Program Files\whatever" |
| 23:36 | < Tarinaky> | Yeah, but there used to be a control sequence for it... |
| 23:36 | < Tarinaky> | Like '\ ' on UNIX. |
| 23:36 | <&McMartin> | No |
| 23:36 | <&McMartin> | Windows shell is distressingly lacking in this. |
| 23:37 | < Tarinaky> | Oh. |
| 23:37 | < Tarinaky> | Where the hell did I get %20 from? |
| 23:37 | <&Derakon> | Spaces in URLs, probably. |
| 23:39 | < Tarinaky> | Which Windows 98 extended into the OS sphere :x |
| 23:39 | < Tarinaky> | I swear it was valid in 98 |
| 23:39 | <&Derakon> | You might have noticed, it's been awhile since Win98 was a valid target platform. |
| 23:40 | <&Derakon> | These things change. |
| 23:41 | < Tarinaky> | How can I take a string "foo.a" and convert it to "foo.b"? |
| 23:42 | <&Derakon> | 'foo.a'.replace('.a', '.b') |
| 23:42 | <&jerith> | In PHP, you can do it with "my_str++" |
| 23:42 | | * jerith vomits. |
| 23:43 | <@rms> | JS: 'foo.a'.replace(/\.a$/, '.b') |
| 23:43 | < Tarinaky> | Okay! Internal error tastic! |
| 23:45 | < Tarinaky> | "c:\Program Files\Inkscape\Inkscape.exe" -zf svg/colony_transport_icon.svg -p colony_transport_icon.png -C --export-width=16 --export-height=16 |
| 23:45 | < Tarinaky> | This invocation causes Inkscape to crash. |
| 23:45 | < Tarinaky> | I -assume- this is because I have messed something up. |
| 23:45 | <&Derakon> | Usually the preferable initial assumption. :) |
| 23:46 | < Tarinaky> | It appears to crash for all permutations run in the script, that's just the first one it tries. |
| 23:47 | < Tarinaky> | Any thoughts? >.< |
| 23:48 | < Tarinaky> | Sorry, that's the second one it tries. |
| 23:48 | <&Derakon> | This is an Inkscape thing, and I don't know it. Hit up the docs. *shrug* |
| 23:50 | < Tarinaky> | Yeah... I already started there >.< |
| --- Log closed Sat Jun 16 00:00:35 2012 |