| --- Log opened Wed Aug 01 00:00:04 2012 |
| 00:01 | | iofficespace is now known as iospace |
| 00:22 | | Nemu_ [NeophoxProd@Nightstar-55d158b3.asahi-net.or.jp] has quit [Ping timeout: 121 seconds] |
| 00:25 | | himi [fow035@D741F1.243F35.CADC30.81D435] has joined #code |
| 00:25 | | mode/#code [+o himi] by ChanServ |
| 00:25 | | Derakon[AFK] is now known as Derakon |
| 00:26 | | mode/#code [+ao Derakon Derakon] by ChanServ |
| 00:35 | | Rhamphoryncus [rhamph@Nightstar-5697f7e2.abhsia.telus.net] has joined #code |
| 00:43 | | You're now known as TheWatcher[T-2] |
| 00:55 | | You're now known as TheWatcher[zZzZ] |
| 01:13 | | gruber [lenin@9D46A2.F4E9D7.E4B4CF.2072AD] has joined #code |
| 01:14 | | gnolam is now known as NSGuest35214 |
| 01:14 | | gruber is now known as gnolam |
| 01:16 | | NSGuest35214 [lenin@Nightstar-202a5047.priv.bahnhof.se] has quit [Ping timeout: 121 seconds] |
| 01:18 | | Syloq_Home [Syloq@NetworkAdministrator.Nightstar.Net] has joined #code |
| 02:10 | | Attilla [Obsolete@Nightstar-429b0f61.as43234.net] has quit [Ping timeout: 121 seconds] |
| 02:13 | < Rhamphoryncus> | http://i.imgur.com/jUXDd.jpg |
| 02:14 | < Rhamphoryncus> | The pentagons are my favourite. They're still hexagons, just with a collapsed edge XD |
| 02:15 | <~Vornicus> | Yay |
| 02:15 | <~Vornicus> | The color's an eyesore though |
| 02:16 | <~Vornicus> | Also check out the ridiculous distortion on that hub hexagon. |
| 02:18 | < Rhamphoryncus> | yup |
| 02:19 | < Rhamphoryncus> | The colour was an accident I ran with |
| 02:19 | < Rhamphoryncus> | distortion is the next step |
| 02:20 | < Rhamphoryncus> | If I crank the "subdivision" to 45 (meaning each edge of a pentagon is split into 45 triangles) I get the same hexagonal tile count as civ 5's huge setting |
| 02:22 | <~Vornicus> | Note that Civ5 still doesn't do spherical worlds - higher connectivity of a sphere makes it feel smaller. |
| 02:22 | < Rhamphoryncus> | ahhh, well put. I had that feeling but couldn't put it in words |
| 02:24 | < Rhamphoryncus> | The system is flawed for civ 5 anyway. If you place a city directly on a pentagon you lose 1/6th of your surrounding area |
| 02:25 | | Syloq_Home [Syloq@NetworkAdministrator.Nightstar.Net] has quit [Client closed the connection] |
| 02:31 | < Rhamphoryncus> | whoever made this, I hate them: http://geomagmasters.com/img/spheres/Sphere810Stand.jpg |
| 02:31 | < Rhamphoryncus> | I got a bunch of those recently at a garage sale and couldn't get anything near that to stay up |
| 02:34 | | Syloq_Home [Syloq@NetworkAdministrator.Nightstar.Net] has joined #code |
| 02:39 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has quit [[NS] Quit: Leaving] |
| 02:50 | < Rhamphoryncus> | Vornicus: http://thomson.phy.syr.edu/thomsonapplet.php |
| 02:54 | < Rhamphoryncus> | I don't know if my goals *exactly* coincide with the thomson problem, but they're pretty damn close |
| 02:55 | <&McMartin> | OSX 10.8 has a new command line utility "caffeinate" to block system sleep |
| 03:01 | <~Vornicus> | pfff |
| 03:18 | < Rhamphoryncus> | Vornicus: it demonstrates that any sort of ideal solution will involve lots of regular grids |
| 03:24 | < Rhamphoryncus> | If avoiding regular grids was really that important to me then I'd have to change the problem such that I get a *worse* solution |
| 04:19 | < Rhamphoryncus> | Vornicus: did you use family 2? http://myweb.tiscali.co.uk/oaktree/butterfly/backpages/icosa_great_circles.htm |
| 04:25 | < iospace> | Rhamphoryncus: how goes pointer snorting? :P |
| 04:36 | <~Vornicus> | Rham: uh. No, I didn't. |
| 04:36 | <~Vornicus> | I need to show you what I really did so you can see how it works. |
| 04:36 | < Rhamphoryncus> | iospace: all done. See screenshot. No more gaps. http://i.imgur.com/jUXDd.jpg |
| 04:37 | < iospace> | nice |
| 04:37 | < Rhamphoryncus> | Vornicus: I appreciate it |
| 04:45 | < Rhamphoryncus> | Right now I'm subdividing before flattening, which I think is my worst source of distortion |
| 04:50 | < Rhamphoryncus> | maybe useful? http://en.wikipedia.org/wiki/File:Icosahedral_reflection_domains.png |
| 04:55 | | Kindamoody[zZz] is now known as Kindamoody |
| 05:00 | <~Vornicus> | Probably. |
| 05:02 | | * Derakon sads at Python |
| 05:03 | <&Derakon> | {}.get('foo', []).append('bar') doesn't result in {'foo': ['bar']} |
| 05:04 | < Rhamphoryncus> | setdefault |
| 05:04 | < Rhamphoryncus> | aka worst named method in the language |
| 05:05 | <&Derakon> | Heh. |
| 05:05 | <&Derakon> | The specific use case here is that I have a mapping of things to lists of things. |
| 05:05 | <&Derakon> | And I need to either create a list with one element, or append to the list, depending on if it exists yet. |
| 05:07 | | * Rhamphoryncus checks help(collections.defaultdict), finds only autogenerated garbage x_x |
| 05:08 | < Rhamphoryncus> | x = defaultdict; x['foo'].append('bar') # Works |
| 05:09 | <&Derakon> | ...yikes. |
| 05:09 | <&Derakon> | That, uh, makes me think of Perl. >.< |
| 05:11 | <~Vornicus> | defaultdict works too, but you should use defaultdict(list) to be explicit |
| 05:11 | < Rhamphoryncus> | ergh sorry |
| 05:11 | < Rhamphoryncus> | I meant defaultdict(list) |
| 05:11 | <~Vornicus> | I use defaultdicts a lot because of this very problem. |
| 05:11 | <&Derakon> | I think I'd rather be explicit. |
| 05:12 | <&Derakon> | I was just hoping there was an elegant way to be explicit and compact. |
| 05:12 | < Rhamphoryncus> | x = {}; x.setdefault('foo', []).append('bar') |
| 05:12 | < Rhamphoryncus> | It's exactly the semantics you're asking for. A get that will assign the default if it wasn't there already |
| 05:13 | <&Derakon> | Yeah, but it subverts the expectations for how dicts work. |
| 05:13 | <&Derakon> | You have to see where the dict is created to understand what's going on. |
| 05:13 | <&Derakon> | It's a bit too magic for me. |
| 05:14 | <&Derakon> | (I'd rather assume that anyone reading my code needs to have their face shoved in exactly what is going on, because that is all too often the case for me) |
| 05:14 | | * Vornicus pokes at his ancient POV knowledge. |
| 05:16 | < Rhamphoryncus> | Derakon: a short comment is justified ;) |
| 05:16 | <&Derakon> | Heh. |
| 05:17 | <&Derakon> | Out of 3413 lines in this project, 816 are comments and 571 are whitespace. |
| 05:17 | <~Vornicus> | Yeah. Now I remember some things I wished I'd figured out |
| 05:18 | | * Vornicus tries to get toruses to appear where he wishes. |
| 05:20 | < Rhamphoryncus> | Vornicus: that thomson problem applet has a torus mode, fwiw |
| 05:20 | <~Vornicus> | Gnah. |
| 05:20 | <~Vornicus> | No, not what I mean |
| 05:21 | < Rhamphoryncus> | Didn't think so |
| 05:21 | <~Vornicus> | I'm using toruses to draw circles on a sphere; but in order to do so I must figure out how to transform the toruses to get them where i want. |
| 05:21 | <~Vornicus> | This means, in particular, figuring out the rotation. |
| 05:22 | <~Vornicus> | but I've never gotten around to doing that, so now I'm sitting here with ridiculous math to do. |
| 05:22 | <&Derakon> | Move the torus straight up to the radius of the sphere, then rotate it about the X axis to the desired latitude, then rotate it about Z to the desired longitude. |
| 05:22 | <~Vornicus> | "rotate it about the X axis to the desired latitude" - desired latitude of what part of the torus? |
| 05:23 | <&Derakon> | You're placing the center of the torus on a specific point on the sphere, yes? |
| 05:23 | <~Vornicus> | Ah, okay. |
| 05:23 | <&Derakon> | With the plane of the torus tangent to the sphere surface? |
| 05:23 | <~Vornicus> | Wait, no |
| 05:23 | <~Vornicus> | No, no. |
| 05:23 | <~Vornicus> | I'm placing the center of the torus at the center of the sphere. |
| 05:24 | <&Derakon> | But ultimately you want the torus on the surface of the sphere, right? You're trying to draw circles on a sphere? |
| 05:24 | <~Vornicus> | Through two points on the sphere. This is a great circle. |
| 05:24 | <&Derakon> | ...ah, that kind of circle. |
| 05:24 | <&Derakon> | You're trying to draw the lines of latitude/longitude? |
| 05:24 | <~Vornicus> | No. |
| 05:24 | <~Vornicus> | I have a pair of points, I want a torus through them. |
| 05:25 | <&Derakon> | Two points on the sphere, and a torus, also on the surface of the sphere, that connects the two points. Hm. |
| 05:25 | <&Derakon> | Clearly the torus must have the same primary radius as the sphere does, then. |
| 05:25 | <~Vornicus> | Right, that part's obvious |
| 05:25 | < Rhamphoryncus> | Could be used for longitude, if you were so inclined, but not for latitude? |
| 05:26 | <~Vornicus> | I can take the cross product of the two points to get the axis of symmetry of the great circle. |
| 05:26 | <~Vornicus> | Rham: that's great circles, yeah. |
| 05:27 | <&Derakon> | Okay, this is too complex for me to do in my head, sorry. |
| 05:27 | <~Vornicus> | hng |
| 05:27 | <~Vornicus> | Okay, cross product gives me a point. |
| 05:27 | <~Vornicus> | Well, a point on the sphere. If I can get the point, uh... hang on, which plane are toruses rendered in. |
| 05:28 | <~Vornicus> | okay, need to get y to the appropriate point, which means rotate about x and then about y. |
| 05:30 | <~Vornicus> | Okay. How much do I rotate about x? acos(axis.y); How much do I rotate about y? atan2(...not sure which is first) |
| 05:30 | <~Vornicus> | but it's axis.x and axis.z in there |
| 05:30 | <~Vornicus> | OKay, I think I can do this now. |
| 05:33 | | iospace is now known as iospacedout |
| 05:45 | | * Derakon spends 20 minutes tracking down a stupid bug. |
| 05:46 | <&Derakon> | I determined that a function a) was doing nothing specifically with the "self" variable, and b) might be useful elsewhere, so I moved it to an external module, and swapped its argument order so that it would be in "source, dest" order. |
| 05:46 | <&Derakon> | Except I forgot to swap the point in the function itself where it recursed. |
| 05:48 | <~Vornicus> | ...I think it worked. |
| 05:48 | <&Derakon> | \o/ |
| 05:49 | <~Vornicus> | torus {1, 0.01 rotate <degrees(acos(axis.y)), degrees(atan2(axis.x, axis.z)), 0>} |
| 05:50 | <~Vornicus> | Yes, victory |
| 05:50 | <&McMartin> | Derakon: Maybe a stupid bug but not a simple one >_< |
| 05:52 | <~Vornicus> | okay. Now let's see if I can' |
| 05:53 | <~Vornicus> | t get this to do the things I need. |
| 05:53 | <&Derakon> | McM: and it wouldn't even have been caught by strict typing, since both of the inputs were dicts of similar structure. |
| 05:54 | <&McMartin> | Boo. |
| 05:55 | | * Derakon finds himself about to type "if not canContinue: continue", swaps "canContinue" for "canAllolcate", sighs. |
| 05:55 | | * Vornicus tries to remember how he did evenly-spaced locations along a great circle. |
| 05:56 | <~Vornicus> | well, okay, I know how I did it. I have no idea how I'll do it when I don't have "rotate about axis" |
| 06:07 | < Rhamphoryncus> | Vornicus: you used the model editor to do the rotations before? |
| 06:12 | | Kindamoody is now known as Kindamoody|breakfast |
| 06:14 | < Rhamphoryncus> | hmm, radius of ?3 |
| 06:34 | <&McMartin> | Allolcate? |
| 06:35 | <&Derakon> | Yes. |
| 06:35 | <&Derakon> | Hence the sigh. |
| 06:35 | <&McMartin> | x = [Lolcat() for x in range (nyanyanyan)] |
| 06:36 | | Derakon is now known as Derakon[AFK] |
| 06:45 | | Kindamoody|breakfast is now known as Kindamoody |
| 06:47 | < Rhamphoryncus> | woo fail. Length of vector in 3d is still square root, not cube root |
| 06:54 | < Rhamphoryncus> | http://i.imgur.com/VfWZE.jpg |
| 06:55 | < Rhamphoryncus> | So much better that I'm not sure there is any more readjustment to be done |
| 06:57 | < Rhamphoryncus> | All I changed was normalizing the radius of each pentagon's centroid *before* subdividing, rather than later when I normalize every radius |
| 06:58 | <~Vornicus> | Rham: I used a handmade matrix lib to do the rotations before. |
| 06:58 | <~Vornicus> | Like I made it myself. |
| 06:58 | < Rhamphoryncus> | ahh |
| 06:59 | <~Vornicus> | This was... foolish, but I was ten years younger then. |
| 06:59 | < Rhamphoryncus> | heh |
| 06:59 | <~Vornicus> | And yeah, that looks good. |
| 07:00 | < Rhamphoryncus> | thanks |
| 07:01 | < Rhamphoryncus> | Enough that the inherent distortion around each pentagon is the worst |
| 07:03 | <~Vornicus> | But yeah, I couldn't get openGL to go right, so I wrote myself a 3d vector-matrix lib and poured the results into POV-Ray for final rendering. |
| 07:04 | | * Rhamphoryncus nods |
| 07:04 | < Rhamphoryncus> | I've learned a lot.. what's most surprising is how solid the result is |
| 07:05 | < Rhamphoryncus> | Yes, it's hideous and slow, but there are no gaps between the triangles, nor any overlap. Not "unlikely, implementation dependant", but rather guaranteed none; it's in the specs |
| 07:07 | < Rhamphoryncus> | Even my cheat of rendering the pentagons as hexagons with two identical vertices, that's perfectly safe |
| 07:14 | <~Vornicus> | Indeed it is. |
| 07:15 | | * Rhamphoryncus deletes 60% of his file as old code XD |
| 07:18 | < Rhamphoryncus> | So, next step.. terrain and/or object placement? |
| 07:20 | <~Vornicus> | Probably. |
| 07:20 | <~Vornicus> | I remember a few things that snuck up on me when I was trying to do terrain. |
| 07:21 | < Rhamphoryncus> | Well, it does mean I'm back to needing addressing again as well as both dense (terrain) and sparse (object) storage |
| 07:22 | < Rhamphoryncus> | An octree or half-assed equivalent should be sufficient for object storage. Not optimal, but the ease of implementation should easily win out |
| 07:23 | <~Vornicus> | Specifically: any 3d models that are tiles will need to do some crazy stuff to line up right. |
| 07:24 | < Rhamphoryncus> | Oh, yeah, that's a grounding assumption |
| 07:24 | < Rhamphoryncus> | Buildings will need to be somewhat flexible, rather than absolutely rigid models |
| 07:25 | < Rhamphoryncus> | But on a real mapsize that'll be a fraction of a percent so it won't be user visible |
| 07:26 | < Rhamphoryncus> | For most it should be the ground around that soaks up the difference, so the massive skyscraper doesn't end up larger at the top *g* |
| 07:37 | < Rhamphoryncus> | I suppose the absolute laziest coordinate system is to use floats :/ |
| 07:45 | < Rhamphoryncus> | I half expect to end up with something like quaternions; "redundant" axis to avoid extreme situations |
| 07:52 | < Rhamphoryncus> | gack. A webpage that describes itself as "heavily under construction - hard hat area!!!" |
| 07:52 | < Rhamphoryncus> | At least there's no animated gif :P |
| 07:53 | <~Vornicus> | Man. I thought you'd found a door to the 90s. |
| 07:54 | < Rhamphoryncus> | ... 2005 |
| 07:55 | < Rhamphoryncus> | And it was actually from MIT research that modelled muscle tissue |
| 07:57 | < Rhamphoryncus> | Okay, I need to quantify what I want from a coordinate system. Starting with portability; I don't want to rely on C/C++'s vague float/double semantics. However, that's particularly for stored points, so intermediate calculations (do a track from A to B) could relax that |
| 07:58 | < Rhamphoryncus> | I also want to avoid significant variation in numerical precision. E.g. longitude at the pole vs equator |
| 07:59 | < Rhamphoryncus> | I shouldn't need to worry about alignment of the numerical precision, so long as I throw on enough spare bits |
| 08:00 | < Rhamphoryncus> | (if the GUI only lets you place tracks to within ?1 cm then storing at 1 mm is sufficient, regardless of rotation) |
| 08:01 | < Rhamphoryncus> | It sounds like a fairly simple x/y/z should be sufficient |
| 08:02 | < Rhamphoryncus> | in integer form |
| 08:03 | < Rhamphoryncus> | which is.. a little over 32 bits |
| 08:07 | < Rhamphoryncus> | If I'm lazy and bump that to 64 bits.. I can get 10% of the way across the milky way :P |
| 08:11 | < Rhamphoryncus> | I do want to include aircraft and possibly even satellites.. but they may be better off under a separate system anyway, since they're obviously highly mobile and not fixed to the ground |
| 08:14 | < Rhamphoryncus> | Hrm, but I can't occlude the core of the earth. That's key to using a simple x/y/z vector :P |
| 08:36 | | himi [fow035@D741F1.243F35.CADC30.81D435] has quit [Ping timeout: 121 seconds] |
| 08:56 | < Rhamphoryncus> | hmm FLT_EVAL_METHOD == 0 |
| 08:59 | < Rhamphoryncus> | caught me by surprise. Apparently gcc and clang on 64-bit x86 no longer play with floating point types on you |
| 09:00 | < Rhamphoryncus> | the defined values are 0: use the type you gave, 1: promote to double, and 2: promote to long double |
| 09:00 | < Rhamphoryncus> | Or anything else being implementation defined |
| 09:01 | < Rhamphoryncus> | The way I learned C was in the "anything else" camp: yes, it'd promote to long double, but when was totally arbitrary |
| 09:32 | | Rev [Reiver@5B433A.F67240.CBBC50.C3A90E] has joined #code |
| 09:55 | | Vash [Vash@Nightstar-e8057de2.wlfrct.sbcglobal.net] has quit [[NS] Quit: I lovecraft Vorn!] |
| 10:06 | | You're now known as TheWatcher |
| 10:09 | | Rev [Reiver@5B433A.F67240.CBBC50.C3A90E] has quit [Client closed the connection] |
| 10:10 | | Rev [Reiver@5B433A.F67240.CBBC50.C3A90E] has joined #code |
| 10:24 | | Kindamoody is now known as Kindamoody|lunch |
| 10:56 | | himi [fow035@Nightstar-5d05bada.internode.on.net] has joined #code |
| 10:56 | | mode/#code [+o himi] by ChanServ |
| 11:10 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has joined #code |
| 12:10 | | iospacedout is now known as iospace |
| 12:20 | | Attilla [Obsolete@Nightstar-429b0f61.as43234.net] has joined #code |
| 12:24 | | * TheWatcher eyes this form |
| 12:24 | <@TheWatcher> | ... how is this getting past the validator, what |
| 12:24 | | * TheWatcher applies the "print STDERR"s |
| 12:55 | | Kindamoody|lunch is now known as Kindamoody |
| 13:02 | | iospace is now known as iofficespace |
| 13:52 | | Rev [Reiver@5B433A.F67240.CBBC50.C3A90E] has quit [Connection reset by peer] |
| 14:08 | | EvilDarkLord [jjlehto3@Nightstar-a5db08cc.org.aalto.fi] has quit [Operation timed out] |
| 14:19 | | EvilDarkLord [jjlehto3@Nightstar-a5db08cc.org.aalto.fi] has joined #code |
| 14:40 | | Kindamoody is now known as Kindamoody|afk |
| 14:44 | | Rhamphoryncus [rhamph@Nightstar-5697f7e2.abhsia.telus.net] has quit [Client exited] |
| 15:15 | | Kindamoody|afk is now known as Kindamoody |
| 16:11 | < iofficespace> | yay grounding +3v lines by mistake |
| 16:11 | < iofficespace> | whoops! |
| 16:13 | <@TheWatcher> | >.< |
| 16:13 | < froztbyte> | haha |
| 16:13 | < froztbyte> | seems in line with this wednesday |
| 17:11 | | Vash [Vash@Nightstar-e8057de2.wlfrct.sbcglobal.net] has joined #code |
| 17:11 | | mode/#code [+o Vash] by ChanServ |
| 17:42 | < iofficespace> | froztbyte: that was yesterday actually |
| 17:43 | < iofficespace> | the sad thing is these are DFNs and the pin i was trying to ground was right next to the +3v pin |
| 17:43 | < iofficespace> | add in the fact that I have to time the grounding well |
| 17:45 | < iofficespace> | it wasn't until I got a board without an obstructive heat frame on that I was able to do it reliably >_> |
| 17:47 | < iofficespace> | oh and as an added bonus! It was the pin attached to the flash where the firmware was stored |
| 17:48 | < iofficespace> | so yeah, i had to manually reflash the thing Dx |
| 17:51 | < froztbyte> | hmm |
| 17:51 | < froztbyte> | well, at least you could still access the programming path |
| 17:51 | < froztbyte> | and didn't burn anything out :) |
| 17:53 | < iofficespace> | yeah :P |
| 17:53 | < iofficespace> | mind you i never shorted any of the pins |
| 17:53 | < iofficespace> | just grounded +3v |
| 17:59 | < iofficespace> | (plus, the pin that was next to the +3v? the clock pin) |
| 18:04 | < froztbyte> | haha |
| 18:04 | < froztbyte> | so this is $dayjob for you, I'm guessing? |
| 18:05 | < froztbyte> | not a university student in the lab thing? |
| 18:07 | < iofficespace> | yup |
| 18:10 | < iofficespace> | i'm testing our firmware when hardware faults are injected |
| 18:59 | < froztbyte> | I see |
| 19:01 | < iofficespace> | ^^;; |
| 19:30 | | Kindamoody is now known as Kindamoody[zZz] |
| 19:59 | | RichyB [MyCatVerbs@Nightstar-3b2c2db2.bethere.co.uk] has quit [[NS] Quit: Leaving] |
| 21:07 | | cpux|2 [cpux@Nightstar-c5874a39.dyn.optonline.net] has joined #code |
| 21:09 | | cpux [cpux@Nightstar-c5874a39.dyn.optonline.net] has quit [Ping timeout: 121 seconds] |
| 21:25 | < gnolam> | https://www.youtube.com/watch?v=0FLlxq5LSlo |
| 21:56 | | himi [fow035@Nightstar-5d05bada.internode.on.net] has quit [Ping timeout: 121 seconds] |
| 23:25 | | himi [fow035@D741F1.243F35.CADC30.81D435] has joined #code |
| 23:25 | | mode/#code [+o himi] by ChanServ |
| 23:31 | | You're now known as TheWatcher[T-2] |
| 23:44 | | You're now known as TheWatcher[zZzZ] |
| 23:59 | | cpux|2 is now known as cpux |
| --- Log closed Thu Aug 02 00:00:18 2012 |