General > Off Topic
What I've done so far
Numsgil:
This is so you know what's going down.
1. Bots' size is now realistically linked to body. (see screenshot)
2. All physics reworked to use Forces. Result: a few instabilites here and there, slower from not being optimized, but generally more realistic reactions between bots and ties, walls, gravity, etc. Haven't worked out drag yet, and friction is acting a little screwy still.
As an unintended result, bots conserve angular momentum from ties very well now.
Still not quite stable for gravity sims yet, which is a disapointment.
3. Shots now collide mouch more often with bots they're supposed to. It's still not 100%, but it's definately an improvement.
4. Options form has been slightly tweaked, with all physics constants being moved to a seperate form. There are some pull down menues for physics defaults.
5. Bots now allowed up to 10 ties.
6. Solved a few logic errors and bugs that freeze the program.
7. Some bits of a "bucket" system started to speed up the simulation.
8. Started the new mutations menu, and some basic DNA tools for the new mutations.
9. Started a new robot debugging form, but no code just yet.
10. Environment grid worked out a bit more. Still not quite there.
That's pretty much it.
shvarz:
That is awesome! So many changes! I feel ashamed I have not updated the web-site yet, which I promised to do a long time ago. Well, I am almost there, probably finish over the weekend.
Woo-hoo!
PurpleYouko:
--- Quote ---5. Bots now allowed up to 10 ties.
--- End quote ---
Hmm??
That is going to screw up the .tielen1, .tielen2 etc. format.
Oh well, it was a bit makeshift anyway and was just put in to test an idea.
I wonder if we can actually simplify the whole thing by using a slightly modified kind of sysvar for the input.
I am thinking of something like .tielen#1 where the "1" can be any integer value and the DNA code will automatically search out the ".tielen#" substring then apply the number that follows it, to the internal tie array address.
A similar system could also replace the clunky tie addressing system that we now use. We wouyld then have an internal tie address (1-10) to be used for setting lengths and angles, and an external tie address (phase) with values from 1 to 32000 to be used as we do now.
The idea needs fleshing out a little but is the basis for the core of a new and simplified tie addressing system that I am starting to work out.
Stryke:
--- Quote ---5. Bots now allowed up to 10 ties.
--- End quote ---
I forsee many program crashes in the future. Just think of a bot mutating the ability to reproduce quickly and eventually being tied to ten other bots that are tied to nine other bots, etc. :wacko: . And also, I have seen that when bots go from one side of the screen to the other when the screen runs out (toroidial) and they have a whole bunch of ties, the tie will sometimes stretch all the way in the opposite direction instead of going through and out the other side. When it happened, it frequently crashed the program for me when in large numbers of ties (I'm not sure if it's been fixed in the new versions). But if it works, I don't see anything wrong with it. All the other improvements are great.
Numsgil:
Yep, it did cause alot of bugs and freezes, but I think I managed to sort most of them out. I tested it with a cancerous veg that ties to everything it sees. It manages to create some very fun webs and strings and things to play with.
I found and fixed a bug that caused large conglomerations of vegs (50+) to freeze the program if they touch the edges.
Everything looks pretty good. The only real problem at this point is that the simulation has slowed down dramatically (alot of square roots taken).
Navigation
[0] Message Index
[#] Next page
Go to full version