Darwinbots Forum
Code center => Bugs and fixes => Topic started by: Griz on October 20, 2005, 11:19:22 PM
-
2.44 totally freezes up my computer.
I'm done.
~griz~
-
It's not supposed to do that...
-
Same here. Well, not totally, but I have to go through Ctrl-Alt-Del to close it. :wacko:
Why are you bothering with 2.4 if you are moving everything to C anyway? Just do it and stop giving us false hope, you sadist... :angry:
One thing I have to agree with MrMound and Griz is that changing so much stuff IS quite annoying to both those who evolve stuff and those who design bots. It is just too frustrating to see the bots stop working in newer versions.
One solution would be to separate versions of program that severely affect bot's behaviour, but keep maintaining the older versions. Like now with new physics and sizes old bots from 2.36 work poorly. So we should just give it version 3 and develop it, but at the same time keep ironing out the 2.36 version and increase the version count. This way we'll have a good stable version that will be targeted at newbies and will work consistently with all the bots. The version 3 would be called "alfa" on the server until most of features that are added work and most major bugs are removed. It is more work, but not by much. The 2.36.5 (?) version is already very stable and working very nicely, so it would only need to be modified when bugs are found or when we have an idea on how to improve interface.
-
The bots will have to be clearly labeled to indicate which version they are designed for, (2.3)-Dom invincibalis.txt or something..
-
Actually the GUI work better now, but I get run-time error 13 when trying to load a settings file wich I've saved with THIS version, and also when trying to access the GUI while the program is running. Using bots Alga_Minimalis and Animal_Minimalis 2.4 Remix as usual. Don't give up with this just yet! I think you're getting close to fix this. Because of the run-time error I haven't had a chance to check how well the settings STAY PUT, but it looked promising.
Also, maybe I should mension I did a clean install and just put this version into the DarwinBots folder.
And I agree with shvarz above, just that I think you should fix this version instead of the old 2.36 versions. I like the options on the settings tabs better with 2.4 versions.
-
6:33 am?? Hey, it's 2:40 pm over here! Rise and shine and start working! :P
-
Hmm, most odd. I just compiled are ran my source over night and, well, all the robots died, that's not supposed to happen, but other than that it ran just fine.
The GUI is working as well as I know how to test it. It definately doesn't freeze my computer or do other wacky stuff.
Back to the grind stone I guess.
-
It occurss to me that this might be related to lastexit.set. Specifically if you don't have a file named that.
So to humor me, everyone whose computers got frozen when they tried to run 2.4.4, copy default.set and rename the copy lastexit.set and see if that makes a difference.
I'm also aware of a bug in translocations that PY pointed out, which I've fixed in the source at home. I'll release 2.4.5 later today that (hopefully) will be mostly working.
-
I already had lastexit.set, but I deleted it and replaced with default.set and that solved the problem.
-
Num, why in the world are you messing with the defult settings file path searching... ?
-
The program now is supposed to remember the last settings you used when you exited the program.
See, before I'd do something, test it, exit the program, change the code, go back in, and have to reset all the old settings I was testing.
This way it's supposed to remember what you were running last time you ran the program. Obviously there are some issues however.
-
Ah, cool.
-
scout griz retuning from recon ...
my report:
ver 2.44
reloaded it ... started it and it immediately froze the program.
splash screen still in place. nothing happens at all ... I have zero control.
any click I make on anything results in the default windoze noise for an error.
takes ctrl-shift-del and aborting the program from the task panel to get my 'puter back.
ok.
copied the defalut.set and renamed it lastexit.set.
this time it fired up.
loaded in the Alga_Minimalis and Num's new Animal_Minimalis for 2.4.
noticed that gravity says zero ... however ...
if one clicks on the advanced button under the physics tab ...
one will find the Y-axis gravity set to 3. reset it to zero and saved the settings.
noticed the default fliud was set to thin and default solid set to teflon.
I didn't change anything else .. and saved the settings as 2.44start. set so I can get it back. ;)
onward.
starts off ok ... some of the animals eat some of the alga ...
although they are moving very quickly ... way too fast to make it interesting.
on the first run I only had 5 veggies at start and forgot to autofeed them and they
got eaten up quickly ... but no more appeared. there was no repopulation ...
even tho the settings should have kept limited them to 25 and repopulated under 10.
nope. didn't happen.
restart ... this time with 30 alga .... the animals fly about quickly ...
bump into an agla on occasion but none eats anything and no bot ever
changes energy ... no gain, no loss ... no interaction whatsoever other than
speeding around and bumbing into one another or veggies now and then.
I did a number of restarts ... same deal.
altered the fluid once to thick ... to see if the bots would slow up ...
well I guess they kinda, sorta did but ...
I got strange ghosted images of the bots ... as if I was seeing double ...
which happens to me now and then even in the 'real world' but then ...
you young bucks wouldn't know about that yet ...
[btw .. you should hope you ever get as old as me to have such problems.] ;)
so I changed it back to thin and the display returned to 'normal'. [ahem]
so ... no interactions between animals and agla.
the alga slowly reproduce and are not limited to 25 as set.
still have 5 animals that are not gaining or losing any energy ...
and at around 50 alga ... I get a visit from my old friend ...
runtime error 6.
we know each other well.
no indtroduction neccessary.
so that's it.
program shuts down ... no error.sim written.
runtime error6, I know him well enough to call him RT6 ...
and I sit down for a cup of tea and to shoot the .... uh ... breeze.
btw ... my computer barely fuctions now ... just draaaaaggggging along ...
as I haven't yet rebooted after this little encouter.
seems that is what I have to do to free up something or other after these
runtime errors.
end of report.
~griz~ over and out.
-
ok
2.44 again.
alga and Num's new animal ...
not much going on as the animals seem to far away to see the veggies.
but the veggies are gaining energy and once they hit about 6000 they reproduce.
this continues and at some point an animal 'see's one of them and starts to feed.
this seems to actually stimulate the alga into reproducing rapidly ...
AND they move ... as fast as the animal ...
as if they have bred with them and taken on their movement gene ...
even if I have 'blocked' checked.
soon the screen is filled with alga and the veggie limit just does not contain them ...
they just overpopulate ...
I even reduced the Veggie Energy from 30 to 1 ... but it didn't seem to help.
attaching that set ... (2.44bstart.set]
which does set the Y-axis gravity to zero and Veggie Energy/cycle to 3.
note: it uses Alga_Minimalis.txt and Animal_Minimalis2.4.txt ...
[that is what I have named Num's animal] ...
so you'll have to delete the species and add your own ...
or alter your robot's name or edit this 2.44bstart.set
well ... you guys may know this ... but newbies don't always ...
I know I didn't and nobody told me so maybe this will be of help to someone.
I've altered it so it should find your robot folder as long as it is in
the DarwinbotsII folder.
well ... I get an error ... saying I can't upload this type of file.
so renamed it 2.44bstart.txt ...
we will see.
~griz~
-
... I got strange ghosted images of the bots ... as if I was seeing double ...
Welcome to the pure evil Vb ran out of memory bug. Solutions = 0 (ok 1, simplify the porgram) But Num claims its a fixable memory leak so ... c++ it is. (although stayed tooned for a more or less working 2.37 I hope)
-
Ghosting bots is from an error in the way the simulation approximates forces.
See, in the real world all forces are executed instantaneously. In the bots' world, they are only calculated every cycle.
Which means for really high forces (like that seen in high drag environments) the bots end up oscillating back and forth in place.
This is a problem, but it's not just a simple fix. I'm looking into a few ways to fudge this to stop infinite oscillations, but it may be a while till I get around to that.
Turns out error.sim was disabled.
So I'll be uploading a tweaked 2.4 later today. It'll contain a few bug fixes, etc.
I'll take a look at your setttings.
-
Look Num it is not alright, most of real bactiria does not do this, Humans cats and dogs deffinatly do not do this eather:
When the birth tie appears in 2.44 it stratches so much that I can put 15 more robots in between the strached robots, the way it worked before was much better when I can only fit 2 robots in between.
I like the strachy tie but only in the case were the total force applyed is greater then 10 + 10 .up store. Its 10 .up store, its not some killer force , the tie should not do that.
-
The tie stretchy thing only happens with smaller mass bots. It too is a consequence of Euler Approximations of forces.
All these and more will be solved at some point in the future ;)
-
ok, fine (I will be waiting, over all: 2.44 runs great) but what about my modified DB source code?
-
Haven't had a chance to look at it yet. I've spent most of last night and this morning trying to get C++ to read the settings files. VB and C++ store things to disk a little differently.
-
Dam it man, Why the hell do you stay up trough the night (I can never imagine myself staying up just to program somthing) :blink:
(But I did abviosly show up early so I can see what you think of it, had to wake up to finish off some homework , ate brakefast and went right on-line waiting for a response from you...)
-
You do your best programming at 3:00 AM. Don't know why, it's just true.
-
...over all: 2.44 runs great)...
Hmm..Am I missing something here? :huh:
-
Yeah, I mensioned in another post I whould prefere to see 2.4 fixed because I liked the choices in the GUI better, but maybe 2.36 whould be easier to finish, bacause it doesn't seem to have so many bugs. Can't wait to try out a bug free version! Ok, keep up the good work folks, I'm getting ready for party. B)
-
Ok, keep up the good work folks, I'm getting ready for party. cool.gif
so you won't be worth a damn tomorrow, eh? :D
well ... have one for me!