Darwinbots Forum
Code center => Bugs and fixes => Topic started by: Griz on October 23, 2005, 11:29:02 PM
-
lifted from announcements ... re: ver 2.45
excellent. (haven't tried it yet tho]
I'm just hoping for a stable version ...
then I'll stop bad-mouthing you Nums. ;)
did you try out Bot's bot placement rountine?
if I download the source for 2.45 ... is it possible to add it?
Oh, and veggies will now repopulate inside their little square like they're supposed to. So you can run your high gravity sims where vegs repopulate blocked in the upper half...
It's actually a pretty cool idea.
yeah. I'm not a complete fool, you know ...
and am capable of having revelations every now and then.
not often but .... ;)
-
I've actually psoted Bots' home code to the FTP, which uses the robot placement codes. They aren't perfect, but they're 1000 times better than what we have now.
It shouldn't be too hard to add them to 2.4.5's source. I leave that as an excercise for the student ;)
I'm running a gravity sim (largest gravity possible) with blocked vegs in the upper fifth of the screen.
It's going to be really cool :P
-
Animal Minimalis 2.4 Remix.
There are some weird bugs I'm seeing, I must admit. Bots "falling" up out of the simulation area, then getting stuck on some invisible cieling about half again the simulation height above the simulation...
So don't freak out when you start seeing that, it's not the old LSD comming back into your blood stream ;)
-
I have no idea actually. If you click and drag them off the ceiling, they float back up, like a helium balloon.
Might be something to do with my vector math. I need to go hunting and figure out exactly what's causing it before I can fix it.
Maybe I need to add a graphic that draws arrows representing the forces acting on the bot.
-
Yeah, I don't imagine too many people would be interested in the forces on every bot at once ;)
-
There are some weird bugs I'm seeing, I must admit. Bots "falling" up out of the simulation area, then getting stuck on some invisible cieling about half again the simulation height above the simulation...
I remember this one :) I used to see it in larger sims. Some Vegs used to travel up to the ceiling and just kind of chill there. Some of the bots followed them up and was able to have a nice feast on the invisible floor. :lol:
And you all were thinking I was nuts talking about seeing such a thing. :P
-
This is too crazy. I think I'll give you guys another week before testing the prog. Maybe 2.4.6 or 2.4.7 might be worth trying out. <_<
-
I've found a good article on the physics they used for Hitman Code: 47 (If any of you remember that game ;))
Anyway, I think I can reimplement some of the physics to be strictly rigid body, so no more bouncing out of bounds, (which still baffles me even after I stepped through the code.)
I'll look at your settings Griz later tonight after work, but the most obvious thing might be that your mutation rates are too high, and you're getting a billion mutations in like 10 seconds...
-
Student? I probebly deleted the virsion I was working on to add the plasment control... So Num, a.) you are the student here as much as me b.) you add it , I chewed it up for you you only have to eat it...
Not Perfict
What did you have in mind to make it more perfict??
Another thing: How did this end up in bot traven?
-
Ah Bots, I knew you would misunderstand, and I wrote it anyway.
"An excercise for the student" is the cope out message most textbooks give. They do something like:
"A and B, are true. Thir proof's are left as an excercise for the student."
More like damn lazy writers if you ask me. "Excercise for the student" indeed!
-
I tried 2.45 in the directory that I normally keep darwinbots in and it kept messing up like the previous incarnations, however I tried a new install in a different directory, with just 2.21 and 2.45 and the problems I had before seem to have gone, it hasn't crashed once yet. So im thinking there must be a conflict with some older part of darwinbots possibly
-
Zip up your problematic directory and sendit to me at Numsgil2002@yahoo.com.
I'll get to the bottom of this if it kills me.
-
and you lot should be more gratefull to Num, if it wasn't for him there wouldn't be anywhere near the kind of progress that has been made. He's only 1 man doing this in his spare time and I for one are very gratefull to him for iit.
-
Yay, you get a cookie.
-
Zip up your problematic directory and sendit to me at Numsgil2002@yahoo.com.
I'll get to the bottom of this if it kills me.
even if its like 40MB big?
-
Num. Let him upload it to the ftp site
-
and you lot should be more gratefull to Num, if it wasn't for him there wouldn't be anywhere near the kind of progress that has been made. He's only 1 man doing this in his spare time and I for one are very gratefull to him for iit.
I am respectful and I appreciate the work, but I'm starting to get impatient. If you could just squish those damn bugs we're reporting about you whould get a program that is far superior from the rest out there! It's so frustrating that I can't even run a simulation through the night. I really want to see my bots evolve! :burnup:
-
99% of bugs are gone. That you are still having problems frustrates me to no end.
I'll PM the FTP details to you Light...
-
It's so frustrating that I can't even run a simulation through the night.
I bet it isn't as frustrating as trying to find those darn bugs and squish them.
-
I jsut checked back on my simulation, and it produced an error.sim file.
So I loaded it into VB, and it ran as smooth as silk.
Confused, I loaded the error.sim into the compiled version, and it ran just fine!
Motto: if it crashes, try loading the error.sim and see if it fixes it.
-
Once again, I think all versions higher than 2.37 should be removed from FTP and Nums valuable time should not be wasted on fixing bugs in a version that is not going anywhere.
Nums, port it to C+, add whatever major features you want to have in there (metabolism, env grid, even if they are as buggy as hell) and then release a version.
2.37 should be the last version of DB II. All versions we have above that are breaking all current bots/sims/setting files and should be called DB 3-somthing to make sure these two beasts don't cross-bread and don't create any ugly offsprings.
-
I'll second that Shvarz. Let's treat the new stuff that Num is doing as a seperate entity for the time being.
-
Aright, I'll create a seperate directory on the FTP later tonight when I get a chance, and load all the 2.4 stuff in there.
-
Just ocurred to me: if we call the new version 3.0, then we can drop the whole "backward-compatible" stuff. The older bots don't work too well in the 2.4 anyway, so why torture them - leave them in 2.37 and make some new ones for 3.0.
This way we'll have much more freedom in designing metabolism, which we always tried to keep backwards-compatible. Same goes for new DNA structure (whatever it ends up to be).
-
That's true...
I've been thinking about older bots... I think I can get them to run right without modifying them and not modifying the physics engine.
Basically I'd just set up the voluntary motions (.up, .dn, etc.) to be scaled by mass. You could turn this off with a preprocessor type declaration, like:
def NewMove
Same thing with turning, if I ever implement that.
-
I've implemented the NewMove thing.
use NewMove
instructs the simulation to not multiply all robot .up, .dn, etc. by mass, allowing for greater control.
Works very well, all my old bots once more work as well as they ever did before.
I view it as sort of cheating, so newer evo bots, etc. should probably include use NewMove at the top of their code.
I'll release the version later tomorrow, I still have to add Bots' code and figure out why the bots like to float up and over the top of the sim.
-
I've found a fixed a major bug with the stack.
It wasn't clearing! Every bot iniherited the stack left behind by its predecessor in the array.
This also was causing the "Bots float up and hit a ceiling" problem.
Endy mentioned the stack acting funny, I should have remembered to check it.
I just need to debug Light's installation and add Bots' bot placement code and I think I'll have a rather useful version ready ;)
-
Light, I found the problem with your version.
A sort-of error message was popping up when you started the program. Unfortunately, it was popped under that initial splash scrren.
I've fixed this in my source code. In the mean time, just hit enter when you start the sim and it'll all work.
-
Hopefully fewer and fewer bugs each version. That's the goal anyway.
-
This also was causing the "Bots float up and hit a ceiling" problem.
Alright I just want to know how? I was puzzeling over this for awhile, it seemed like there would be spots like an invisible gas vent that allowed the bots to float up.
(kind of cool actually if only they were even vaguely stable ecosys)
It seemed to be fixed for awhile then it's suddenly back. :wacko:
a few steps back, a few forward ...
as long as you end up with more forward than back ... progress.
as submariners say ...
as long as you surface as many times as you dive ... no problem.
Eventually we'll make progress it just will take awhile. B) This is one of the most radical (weird using this to describe Alife program :) ) changes made so far. I think every time the physics engine is moded is when bugs rear up the worst.
-
I don't know if it's the same invisible floor/ciling you were seeing or not Endy.
This one eventually caused my bots to fire 1000 and 1000 right every cycle. That was strong enough to propel them far outside normal bounds, until the force extered on them to go back into the sim was the same as the force they exerted to leave it.
The result was a sort of equilibrium point they drifted back to if you moved them.