Darwinbots Forum
Code center => Bugs and fixes => Topic started by: PurpleYouko on November 18, 2005, 03:09:42 PM
-
So, Am I to assume that the total and utter lack of overflows, runtime errors and other crap being reported for V2.37.6, is an indicator that it is finally stable? B)
Or is everybody just too lazy to comment? <_<
-
I am very busy at work and at home. Next week I'll have more time.
If you have time, maybe you could move some of the more advanced controls from Nums' 2.4 version (like fancier mutation frequency). Stuff that would still keep 2.37 as 2.37, but expand on convience.
-
Nums did a bit more than just put advanced controls in the options form.
I think he re-wrote a whole bunch of the mutations code.
-
I wouldn't recommend touching the mutations code for the weak of heart.
Had I known the sheer complexity of the code for it, I wouldn't have redone it so lightly. There are many, many special cases that mess with them...
Anyway, it's alot of work. Since 2.4 seems more or less stable now and you like some of its features, maybe you should upgrade? Mmmm?
-
I am waiting for C+ port. Besides, 2.37 has its own place. I would not call switching to 2.4 an upgrade, but rather a move sideways. Here are a couple of comparisons that come to mind: "Word" is not an upgrade to "Notepad", both are text editors, but one is not an upgrade from the other.
"Civ 4" is not an upgrade from the original "Civilization". It has better graphics and somewhat more complicated and just different rules, but the original still holds its own. In fact, with games it works quite well. I just finished "Dungeon Keeper", a game by Peter Molinoux from 1987, and was impressed by it much more than by its (much more recent) "B&W".
That's why I want to to just move to C+ and call it DB-III. This way it will be clear, that it is not just an upgrade, it is a different beast. And it will leave room for 2.37. My guess is that the new DB will be geared much more toward evolution, so for people who just want to have a battleground with simple rules, 2.37 might be a better choice.
-
That does seem to be more or less the case, seeing as how there are somewhat fewer battle bots being made as of late.
-
Lemme tell you something guys! The overflow bugs seems to have been SQUISHED, finally, but there are a few things I don't understand. From my point of view, as a user who know nothing about coding, I don't understand why some things, like some settings in the GUI doesn't get saved in the settings file! I've mensioned something about that before somewhere. (I've had a few drinks at the pub so I'm writing spontaniosly here). Somehow you programmers doesn't see the same thing as we users do. Maybe you just use the default settings with alga Minimalis and preservians, start a simulation with that on another computer and leave it at that with no thought of how the other settings will work. I like to ONLY use Alga whatever, Animal Minimalis and Triangulus. I've tried both pond mode and no pond mode. When I make settings and save them in the settings file, I find that when I start up the program I first have to load the settings file with the settings I used for the simulation, THEN load the simulating I saved, otherwise the simulation I load will use the default settings. If I were a programmer, I whould make sure EVERYTHING was saved in that simulation, ALL settings I used, so I JUST need to load the saved simulation and it whould continue EXACTLY as it was when I saved it! Also I don't understand why the day/night cycles doesn't work in 2.4.A as they do in 2.37.6. I get the feeling that Nums started a total remake of the program and forgot to put in the code that WORKED for day/night cycles. Are you following me? Also I don't understand why some settings in the GUI keeps reset to other values than the once I chose! Like on the physics and costs tab and the Advanced controls tab! It pisses me off that I can't make the settings STAY PUT! I want brownian motion set to ONE, damnit! It keeps going back to 10! Well...I don't see why I whould have to tell you this. See for yourself. Make different settings. Save it in a file. Quit. Start the program again. load that settings file and you will see that some settings is back at default values! This is very disturbing. Right now I think it's just the GUI settings not staying AS SET, day/night cycles in version 2.4.A doesn't work as it should (keeps start at night nomatter what), and the program runs TOO SLOW! 3 things left to fix, as I see it, then it will finally be perfect! Ok, that was long post. You think you can read all of it? :P
Oh, and it whould be nice if bots and corpses doesn't SINK in version 2.37.6 when gravity is turned OFF! They don't sink in 2,4,A. Why do they sink in 2.37.6 with the same settings??? GRRR!
-
When I mess around with the bot placement (ie negative placement etc.) the program freezes upon the start of the simulation. But I don't think that's a major bug since I normally don't desire negative placement for my bots :)
-
So, Am I to assume that the total and utter lack of overflows, runtime errors and other crap being reported for V2.37.6, is an indicator that it is finally stable? B)
Or is everybody just too lazy to comment? <_<
I've been getting runtime errors on occasion in 2.37.6 but ...
they occur in compiled mode and end the prog so no debugging info for you.
I forgot to see if I can insert that error writing routine ...
or maybe it's there and just needs to be uncommented.
then I think one could load the error.sim back into the VB version
and maybe get some info.
otherwise ... it seems to b pretty stable ... for 300000 cycles at least
playing with Boris' multiveg.
the veggie population controls are still kinda strange at times ...
but unconsistant enough so I can't point at anything.
ah ... here's something I noticed.
trying to play with/ look at boris's MultiVeg1.1 and using the console ...
I find that every veggie on my screen always has the same value
at ? 50
not just every 'part' of the multiveg ...
but every veg on the screen.
now I know that the 'leaf' and 'flower' portions .. #2 and #3 aren't
supposed to be able to 'grow' another segment ....
but I never fine a combination of 'types' in a multibot ...
I would expect at least one #1, ie stem, and then some other types ...
but find them always to be all the same type.
if I set position 50 to #1, ALL veggie bots then read back #1 ...
or whatever I set 50 to.
no idea what's happening with that ....
as they seem to 'grow' just fine, regardless of what value I find at ? 50.
just what I have been finding.
no idea if anyone else has been looking at this ...
but I would think Boris uses it to examine his creations.
an afterthought:
surely I don't have close and reopen the console for every bot I select?
that would be pretty strange but I'm grasping at any straw here ...
-
Ok,
I got VB; and while starting a simulation I stumbled on this runtime 9 error in the options form under list management
It highlights this section: (I assume that's where the problem is)
Shape1.width = Command3.left - Command2.left + Command2.width
My first time using VB, so forgive me when this is complete and utter nonsense :unsure:
BTW, is there an option in VB to display/export a bug found while debugging?
-
That looks like the robot placement box code.
I have never had a crash there before.
You are doing the right thing to find the error. When you click "debug" after a crash, the problem line will be highlighted in yellow.
This doesn't often mean that the actual error is in this line but more likely that this line simply catches an error generated eleswhere.
If you hold the mouse over the line of code (or anywhere in the subroutine) you will see little text boxes appear that show the present values in the variables. This information is extremely valuable for de-bugging.
-
Tried to generate bug again, and got it to chrash again. for clarity: it is in optionsform, under list management
Line:
Shape1.width = Command3.left - Command2.left + Command2.width
Values:
Shape1.width: 23
Command3.left: 2570
Command2.left: 7457
Command2.width: 135
-
The line of code appears under "Speclist_Click"
There is a commented heading that says "list Management" at the top of the section
Does that sound right?
Are you using V2.37.6 or 2.4 here?
How exactly do you make it crash. I can't seem to get it to do so.
-
Ah! I think I have reproduced the error.
You have to move the top right corner marker off the left side of the field to make a negative width.
Should be fixable.
-
Great! Yay I just found my first bug :) This is an historical moment... :)
Glad to hear it's fixable
-
an afterthought:
surely I don't have close and reopen the console for every bot I select?
that would be pretty strange but I'm grasping at any straw here ...
Yes you do. Why? Because you can open up a seperate console for each bot. The consoles belong to the bots, they're not a universal tool.
How can you tell this? The top of the console says "Robot X console". Obviously then the console belongs to robot X.
Why is it done this way? So you can open multiple consoles at once I suupose.
-
That looks like 2.37 code. 2.4 uses a different robot placement of course.
-
Yeah it's 2.37
Are you going to combine the 2.4 and 2.37 versions, or are they remaining separate? It's a bit confusing.
-
Yes you do. Why? Because you can open up a seperate console for each bot. The consoles belong to the bots, they're not a universal tool.
ok.
thanks
-
Yeah it's 2.37
Are you going to combine the 2.4 and 2.37 versions, or are they remaining separate? It's a bit confusing.
2.4 is a superset of 2.37. That is, 2.37 contains no features 2.4 does not have. Well, minus a few things with ties, which are rather confusing to me at the moment. But I'm working on it.
There just aren't alot of new bots specifically for 2.4, and older bots definately work better in 2.37. I think this is primarily a result of bots now getting really tiny.