search of the:   for: 
code logs -> 2010 -> Fri, 03 Sep 2010< code.20100902.log - code.20100904.log >
--- Log opened Fri Sep 03 00:00:45 2010
00:15 cpux [chatzilla@Nightstar-20a84089.dyn.optonline.net] has joined #code
00:16 Tarinaky [Tarinaky@Nightstar-f349ca6d.plus.com] has joined #code
00:32 Tarinaky [Tarinaky@Nightstar-f349ca6d.plus.com] has quit [Connection closed]
01:06 Derakon[AFK] is now known as Derakon
01:25 Derakon is now known as Derakon[CIV]
02:35 gnolam [lenin@Nightstar-38637aa0.priv.bahnhof.se] has quit [[NS] Quit: Z?]
02:52 Thaqui [Thaqui@27B34E.D54D49.F53FA1.6A113C] has quit [Connection closed]
02:55 Thaqui [Thaqui@27B34E.D54D49.F53FA1.6A113C] has joined #code
03:15 Derakon[CIV] is now known as Derakon
03:44 Stalker [Z@26ECB6.A4B64C.298B52.D80DA0] has quit [Ping timeout: 121 seconds]
04:07 Stalker [Z@3A600C.A966FF.5BF32D.8E7ABA] has joined #code
04:29
<@Derakon>
...hm, okay, that doesn't work.
04:29
< Vornicus>
?
04:29
<@Derakon>
I wanted to remove long edges from the graph, so I wrote up some code to remove any edge that's more than 2 standard deviations longer (or shorter) than the average.
04:29
<@Derakon>
Problem being that with 1k nodes and my node placement system, I have a very consistent edge length.
04:30
<@Derakon>
Just...not quite consistent enough to keep me from removing every edge from the graph. ¬.¬
04:31
<@Derakon>
...it would also help if I remembered how to calculate the standard deviation properly. It's the absolute value of the distance from the average, you nincompoop.
04:31
< Vornicus>
every edge? How'd you manage that? 3/4 of all values are /guaranteed/ to be within 2SD of the mean.
04:31
< Vornicus>
Um.
04:31
< Vornicus>
Square of the distance from mean, added up, divided by the number of elements, then square rooted.
04:31
<@Derakon>
Oh, really?
04:31
< Vornicus>
Yep.
04:32
<@Derakon>
Maybe I should have looked it up instead of assuming~
04:32
<@Derakon>
Thanks.
04:32
< Vornicus>
(in a sample instead of a population you divide by n - 1, not n.)
04:32
< Vornicus>
(I am a professional statistics tutor.)
04:32
<@Derakon>
So we keep discovering whenever someone wants to try a new dice system.
04:32
< Vornicus>
There's a one-pass algorithm though.
04:33
<@Derakon>
Okay, this works pretty nicely.
04:34
<@Derakon>
http://derakon.dyndns.org/~chriswei/temp2/delaunay.html
04:34
< Vornicus>
(the 3/4 thing is a theorem that people don't use very often; I know it because it's elegant and it makes for a reasonable smell test.)
04:34
< Vornicus>
Shiny.
04:34
<@Derakon>
What I find especially amusing is how this is amazingly faster than my previous graph generator...
04:37
<@Derakon>
Though I'm still bugged by the way it allows two edges with a very acute angle between them.
04:37
<@Derakon>
I don't know how that'd look in-game, but my suspicion is "inelegant".
04:37
< Vornicus>
That's really pretty, by the way.
04:38
< Vornicus>
Suggestion.
04:38
< Vornicus>
Remove convex hull edges.
04:38
<@Derakon>
I think that's functionally similar to the remove-long-edges trick, but yeah, makes more sense.
04:39
< Vornicus>
All the ridiculously acute angles will happen at the hull.
04:40
< Vornicus>
Well. Almost all.
04:41
<@Derakon>
Mmm...that still leaves several edges that I don't really like.
04:41
<@Derakon>
Maybe if I iterated it a few times.
04:42
<@Derakon>
...no, then the "construct hull" code breaks down.
04:42
<@Derakon>
(Since the hull is no longer convex)
04:43
<@Derakon>
There's also a secondary issue of edges that pass too close to nodes.
04:46
< Vornicus>
Also, why on earth is the "square" zone so regimented diagonal?
04:46
<@Derakon>
At a guess, BFS meant that the closest nodes each time were always down-right.
04:47
< Vornicus>
...another suggestion. Instead of randomizing your queue inputs as they enter, shuffle them each time you reach a new length. Might want to put "None" in there as a sort of marker.
04:47
<@Derakon>
My queue inputs aren't randomized except insofar as I don't bother to order them.
04:48
< Vornicus>
mmm.
04:48
<@Derakon>
What would be the goal of shuffling them?
04:49
< Vornicus>
The goal of shuffling them would make it so it chooses randomly among all the possible paths to extend, instead of biasing towards the first path you chose on each pass.
04:49
<@Derakon>
Oh, to avoid the regimented diagonals?
04:49
< Vornicus>
Among other things; it will also tend to somewhat even out your pathwise node counts.
04:51 Thaqui [Thaqui@27B34E.D54D49.F53FA1.6A113C] has quit [Connection closed]
04:56
<@Derakon>
...that doesn't seem to have made any difference. I must have screwed it up somehow.
05:00
<@Derakon>
Okay, that's different.
05:00
<@Derakon>
http://derakon.dyndns.org/~chriswei/temp2/graph0004.png
05:00
<@Derakon>
(May need to clear cache)
05:02 Stalker [Z@3A600C.A966FF.5BF32D.8E7ABA] has quit [Connection reset by peer]
05:02 Stalker [Z@3A600C.A966FF.5BF32D.8E7ABA] has joined #code
05:05
< Vornicus>
That is different.
05:15 * Derakon ponders how to efficiently prune edges that pass too close to nodes.
05:16
<@Derakon>
Short of slapping every edge into a quadtree...
05:19 celticminstrel [celticminstre@Nightstar-f8b608eb.cable.rogers.com] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!]
05:26 * Derakon mutters at the problems with getting the shortest distance between two angles.
05:26
<@Derakon>
The problem is that discontinuity at 0...
05:29
< Vornicus>
"shortest distance between two angles" as in you've got an angle and you want to make sure it's not too acute?
05:30
<@Derakon>
I have three nodes; one central, one existing connection, one prospective connection.
05:30
<@Derakon>
I want to ensure that the angle with the prospective connection is not too acute with the existing edge.
05:31
< Vornicus>
Just kill it if it's less than 30 degrees apart. Dot product is your friend.
05:33 Stalker [Z@3A600C.A966FF.5BF32D.8E7ABA] has quit [Ping timeout: 121 seconds]
05:34
<@Derakon>
Man, I am way too rusty on my vector operations.
05:34
<@Derakon>
I think this is how I should be using the dot product, but it's pruning out way too many opportunities (with MIN_ANGLE_DISTANCE = pi / 8): http://paste.ubuntu.com/487620/
05:35
< Vornicus>
pretty much.
05:35 cpux_ [chatzilla@Nightstar-20a84089.dyn.optonline.net] has joined #code
05:37
< Vornicus>
Wait, no
05:37
< Vornicus>
acos the result of the dot.
05:37
<@Derakon>
...oh, dur. I even had the Wikipedia page open and there's the arccose right there.
05:38 cpux [chatzilla@Nightstar-20a84089.dyn.optonline.net] has quit [Ping timeout: 121 seconds]
05:38 cpux_ is now known as cpux
05:38
<@Derakon>
Yeah, that looks better.
05:39
<@Derakon>
Now I just have to deal with the nodes that are too close to edges.
05:39
< Vornicus>
I'd actually fold this into your vector library: given two vectors, get the angle between them.
05:39
<@Derakon>
Yeah.
05:41
<@Derakon>
return math.acos(alt.normalize().dot(self.normalize()))
05:41
<@Derakon>
That does mean that I'm calculating redundant normalizations...ehh.
05:42
< Vornicus>
You need the length anyway. Actually I'd -- if you can -- cache the length of each vector, y'knwo, lazily.
05:42
<@Derakon>
Hrm?
05:43
< Vornicus>
Whenever you call vec.length() the first time, vec stores and caches the length. Takes out the most onerous calculation for the later attempts.
05:44
<@Derakon>
Good idea.
05:45
< Vornicus>
def length(self): if not self._length: self._length = math.sqrt(self.x**2 + self.y**2);; return self._length
05:45
<@Derakon>
Yep.
05:55
<@Derakon>
Yeah, this looks pretty nice. http://derakon.dyndns.org/~chriswei/temp2/spanningtree.png
05:55
<@Derakon>
Still a bit too vertically-oriented for my tastes, and of course the proximity issue, but otherwise that looks entirely usable.
05:55
< Vornicus>
Beautiful.
05:56
<@Derakon>
I should have gone down this route in the first place...
05:57
<@Derakon>
Man, the amount of pain and hacks I went through getting the first map generators working...
05:57
<@Derakon>
I mean, look at this... http://derakon.dyndns.org/~chriswei/games/jbrl/mapgen3.png
05:58
< Vornicus>
Damn
06:00 * Derakon eyes the above, has himself cited for ellipsis abuse.
06:00
< Vornicus>
Yeah, your dots should be vertical there.
06:47 Derakon is now known as Derakon[AFK]
06:49
< Vornicus>
Oh, one thing I noticed, Der: your start point always seems to be up high.
06:49
<@Derakon[AFK]>
It's randomly-selected from all nodes in my set.
06:49
<@Derakon[AFK]>
I just happen to be re-using the same RNG seed.
06:50
< Vornicus>
Ah.
07:02 Anno[Laptop] [annodomini@Nightstar-985bfa43.adsl.tpnet.pl] has joined #code
07:03 Kazriko [kaz@Nightstar-e09690fa.client.bresnan.net] has quit [Ping timeout: 121 seconds]
07:16 Thaqui [Thaqui@27B34E.D54D49.F53FA1.6A113C] has joined #code
07:22 cpux [chatzilla@Nightstar-20a84089.dyn.optonline.net] has quit [Ping timeout: 121 seconds]
07:24 cpux [chatzilla@Nightstar-20a84089.dyn.optonline.net] has joined #code
07:36 SmithKurosaki [Smith@Nightstar-7dbfa647.dsl.teksavvy.com] has joined #code
07:59 Kazriko [kaz@Nightstar-e09690fa.client.bresnan.net] has joined #code
07:59 mode/#code [+o Kazriko] by Reiver
08:56 You're now known as TheWatcher
09:15 Vornicus is now known as Vornicus-Latens
09:55 SmithKurosaki [Smith@Nightstar-7dbfa647.dsl.teksavvy.com] has quit [Ping timeout: 121 seconds]
09:58 SmithKurosaki [Smith@Nightstar-e73eb6b0.dsl.teksavvy.com] has joined #code
11:04 Reiv [orthianz@Nightstar-6d8a14fd.xnet.co.nz] has quit [Ping timeout: 121 seconds]
11:34 Reiv [orthianz@Nightstar-5cb6a951.xnet.co.nz] has joined #code
11:53 Tarinaky [Tarinaky@Nightstar-f349ca6d.plus.com] has joined #code
12:17 gnolam [lenin@Nightstar-38637aa0.priv.bahnhof.se] has joined #code
12:27 cpux is now known as shade_of_cpux
12:30 shade_of_cpux [chatzilla@Nightstar-20a84089.dyn.optonline.net] has quit [[NS] Quit: ChatZilla 0.9.86 [Firefox 3.6.8/20100722155716]]
12:53 Thaqui [Thaqui@27B34E.D54D49.F53FA1.6A113C] has quit [Client closed the connection]
13:02 Stalker [Z@3A600C.A966FF.5BF32D.8E7ABA] has joined #code
13:20 Rhamphoryncus [rhamph@Nightstar-bbc709c4.abhsia.telus.net] has quit [Client exited]
14:01 celticminstrel [celticminstre@Nightstar-f8b608eb.cable.rogers.com] has joined #code
16:31 Alek [omegaboot@Nightstar-c5f3565b.il.comcast.net] has quit [Ping timeout: 121 seconds]
16:56 Stalker [Z@3A600C.A966FF.5BF32D.8E7ABA] has quit [Ping timeout: 121 seconds]
17:01 You're now known as TheWatcher[afk]
17:19 Stalker [Z@26ECB6.A4B64C.298B52.D80DA0] has joined #code
17:37 Vornicus-Latens is now known as Vornicus
19:20 Abu[Laptop] [annodomini@Nightstar-fb7e5acb.adsl.tpnet.pl] has joined #code
19:22 Anno[Laptop] [annodomini@Nightstar-985bfa43.adsl.tpnet.pl] has quit [Ping timeout: 121 seconds]
20:08 You're now known as TheWatcher
20:33 Alek [omegaboot@Nightstar-8f41d887.il.comcast.net] has joined #code
20:37 * Alek could cry.
20:37
< Alek>
so we have a 12Mbps pipe, theoretically.
20:38
< Alek>
we actually get more like 24, actually. >_>
20:38
< Alek>
and guess what Comcast went and did?
20:38
< Alek>
put a monthly transfer cap of 250GB on all residential lines. ;_;
20:38
< Alek>
and since we transferred over a terabyte over the past month, we were blocked. ;_;
20:40
< simon_>
boo!
20:42 * jerith wishes for problems like that.
20:42
<@jerith>
We pay per-gb, and my line is 384kbps. (Theoretically.)
20:43
< Alek>
I remember when we had DSL.
20:43
< Alek>
we started with 1.5MBps.
20:43
< Alek>
then that wasn't enough, we upped to 3.
20:44
< Alek>
no noticeable performance increase. so we upped to 6.
20:44
< Alek>
we were paying all that money for months before we found that our line was poor quality and couldn't support any more than 1.5. -_-
20:44
< Alek>
so we switched to cable.
20:44
<@jerith>
:-/
20:44
< Alek>
and now this. -_-
20:45
< Alek>
they didn't even warn us! >:(
20:45
<@jerith>
4mbps is the best we can get here. Although they're testing 10mbps in certain places, but it's "burst up to" and doesn't much.
20:46
<@jerith>
Not warning you is grounds for a lawsuit, surely?
20:46
<@jerith>
Isn't that waht you people do over there? ;-)
20:48
< Alek>
-_-
20:48
< Alek>
not us.
20:48
< Alek>
at most, we complain to customer service and get a discount. :P
20:49
< Alek>
that's what my family does anyway.
20:49 * simon_ can download at 400KB/s on good days. normally it's 200KB/s
20:49
< Alek>
-_-
20:50
< Alek>
funny, I thought Europe in general had much better broadband availability than the US.
20:50
< simon_>
except today, I downloaded a six-part .rar and everytime I started a new connection, the download speed of the others didn't drop. so eventually it was something like 1MB/s.
20:50
< Alek>
so maybe it's a per-file limit?
20:50
< simon_>
well
20:51
< simon_>
I live in Copenhagen, which is the capital of Denmark. in general we don't have as much fiber as everyone would like (that's Sweden you want to look at)
20:51 * Alek does some researching, finds something interesting
20:51
< Alek>
Comcast's new 250GB transfer cap is, get this...
20:51
< simon_>
the highest DSL companies advertise with is 10/1mbps.
20:51
< Alek>
nationwide.
20:51
< simon_>
that's crazy.
20:51
< Alek>
yep.
20:52
< simon_>
only wireless broadband providers have caps here.
20:52
< simon_>
(and they're at 10-30GB)
20:52
< Alek>
especially since their plans range from 1.5Mbps (DSL equivalent) to 50Mbps.
20:52
< Alek>
wireless broadband from Comcast is capped at 2 to I think 6 or 10 per month.
20:53
< Alek>
nationwide 3G. metro is uncapped as far as I can tell, or has a higher cap.
20:53
< simon_>
I used to live in a dorm where there was a 3G/day cap.
20:53
< simon_>
(so that's one good reason to volunteer for the network group :))
20:58
< Vornicus>
So I have discovered that 1. Multiplication Drill's results table does not work on IE6. 2. the computer labs at school use IE6.
21:00
<@TheWatcher>
I note you have also discovered a third point:
21:00
<@TheWatcher>
whoever manages the school's computer lab software deserves to be strangled with their still-living entrails.
21:03 * TheWatcher suggests changing all the home pages to http://www.bringdownie6.com/ ¬¬
21:13 celticminstrel [celticminstre@Nightstar-f8b608eb.cable.rogers.com] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!]
22:02 Rhamphoryncus [rhamph@Nightstar-bbc709c4.abhsia.telus.net] has joined #code
23:10 ToxicFrog [ToxicFrog@ServerAdministrator.Nightstar.Net] has joined #code
23:10 mode/#code [+o ToxicFrog] by Reiver
23:37 Stalker [Z@26ECB6.A4B64C.298B52.D80DA0] has quit [Ping timeout: 121 seconds]
--- Log closed Sat Sep 04 00:00:46 2010
code logs -> 2010 -> Fri, 03 Sep 2010< code.20100902.log - code.20100904.log >
 search of the:   for: