Darwinbots Forum
Code center => Bugs and fixes => Topic started by: Numsgil on April 03, 2005, 06:25:33 AM
-
Everyone download 2.36.1 and test it. Most of your bugs should be fixed.
Everyone repost any bugs still not fixed, and any new bugs found.
Schvarz, I couldn't get it to crash on me when I used graphs. Alot of bugs you found I can't seem to find, but then I haven't really been playing with it much.
-
The 'eye bug' has been fixed, yay! :)
I have two questions:- What is the recommended value for the 'energy per cycle per veggie' thing? What was the old value?
- For some reasons sometimes Alga_minimalis's "eye" is far in front of the bot itself. This isn't always the case. (see pic)
-
- What is the recommended value for the 'energy per cycle per veggie' thing? What was the old value?
The old value kept changing depending on how many vegs there were. This wasn't very realistic.
40 seems about right to me to get the feel the vegs had before.
-
I think the eye thing may be due to a new variable that Num added.
In the old system the position of the eye and the actual size of the circle making up the robot was calculated at the point of displaying it.
Now it is precalculated at one point of the cycle and saved as rob(n).radius. I am not sure but I think the new value may be saved as an integer. I noticed a few places where the old code was still used so I wonder if this may be causing a small difference in values.
I will have a look through the code to see if I am right.
-
Now that works a lot better! B)
Snapshot still doesn't work right though. The program just crashes completely and won't respond again. Wonder what happened there? :unsure:
Glad to see you included my new collisions routine for this release. On that subject I would like to get some feedback from everyone so I will start an opinion poll (I like polls) in a new thread.
-
Your collision routines should be in 2.36.0 too.
Snapshot worked for me. I think anyway. That's the button that looks like a camera, right? It did take for ever though.
-
Eye bug:
-
Your collision routines should be in 2.36.0 too.
Didn't really get the chance to test 2.36. The robots all died out in about 10 seconds so collisions were the last thing on my mind in that short period.
-
I suggest adding a new default settings file to the next version...
-
Good idea K0zm0
-
Eye bug:
Turn down the feed rate to something that makes sense and the problem goes away. 30 works pretty well!
-
The new 'eye bug' is a graphical bug only. And it's not a problem with the eyes, it's a problem with the radius of the veg. Those huge eyed veggies should be monstrously sized, but they're not.
I think PY was on to something with his idea about the radius variable that I added. I'll look into it.
All in all though, it doesn't effect gameplay.
-
it doesn't effect gameplay.
, I did not come here and download this thing to play games....
ok thats 3 for one day , :plzdie:
-
it doesn't effect gameplay.
, I did not come here and download this thing to play games....
ok thats 3 for one day , :plzdie:
What I meant was that it doesn't effect the robots in any way. It only effects how the robots are being displayed to the screen.
Hence, it's not a gameplay issue, but a graphics issue. Hence people can merrily still use the program without worrying about this new eye bug screwing with their mutations or bot building.
-
o , ic , better grahix in my opinion actualy , I like it better this way.
-
For fuck sake!
I am starting to get pissed off here! :angry:
-
Py, take it easy Bro...
****
A small bug in the settings,
In virsion 2.1 it sayed: "max energy in feild"
Now it says: "energy per veggie per cycle"
But when you load up any of the old settings it still loads HUGE numbers for the "energy per veggie per cycle" , because it still thinks its "max energy in feild".
The possible solutions are:
1.) Change all the settings files that are installed with the program.
2.) Make the program devide thouse huge numbers to give proper numbers.
3.) put the max energy in feild back.
I think the first one will be best because soon we will make the gap between veggie and hunter seamless.
-
PY posted a new settings file for default in the newbie forum I think.
edit: it's not newbie, it's in What is Darwinbots forum.
-
ok , thx
-
Think calm serene thoughts :bigginangel:
Think calm serene thoughts :bigginangel:
AHHHH bollox to that !
I just don't feel very serene right now
:firedevil:
-
Wow, he's so pissed he went out and uploaded a new smiley for it.
That's intense. :idea: :cyclops: :firedevil: :sleep2: :sad2:
-
HAH!
I put those up about a week ago
Caught you napping eh? :sleep2:
-
Haha, yeah. I haven't really looked lately.
-
I like the little sad bouncy one best I think
:sad2:
-
It reminds me of those commercials for the depression medication.
Sad? No longer take joy from the things you used to?
:sad2:
-
Yeah it does doesn't it?
I can't remember what the stuff was that they were advertizing though
-
Yeah it does doesn't it?
I can't remember what the stuff was that they were advertizing though
Coors Light.
-
Small one:
On Options - Physics and Costs - Energy Exchanged on Hit:
Increment arrows for both Proportional and Fixed increment the Proportional field only.
Did someone say the eyes were fixed in 2.36.1? I didn't notice anything about eyes in 2.36.0. Nope. Didn't notice. Wasn't suppose ta, ain't gonna. Not me. Nope.
Steven Hawking tells a story about a public lecture he once gave. After the lecture an elderly matron came to him and scolded him for his view of the cosmos. She insisted that the Earth was flat and, indeed, rode on the back of a giant tortoise. He asked her what was underneath the tortoise and what was under that. She replied that he was a clever young man indeed but that it was turtles all the way down.
Bye, y'all.
-P
-
Steven Hawking tells a story about a public lecture he once gave. After the lecture an elderly matron came to him and scolded him for his view of the cosmos. She insisted that the Earth was flat and, indeed, rode on the back of a giant tortoise. He asked her what was underneath the tortoise and what was under that. She replied that he was a clever young man indeed but that it was turtles all the way down.
Sounds like some of the arguments I have with Zelos.
-
Steven Hawking tells a story about a public lecture he once gave. After the lecture an elderly matron came to him and scolded him for his view of the cosmos. She insisted that the Earth was flat and, indeed, rode on the back of a giant tortoise. He asked her what was underneath the tortoise and what was under that. She replied that he was a clever young man indeed but that it was turtles all the way down.
Well she got it half right.
I mean everybody knows that the world is supported on the backs of 4 giant elephants who in turn stand on the back of the great world turtle "A'Tuin"
Nothing is needed to support this turtle as it swims through the cosmos, stopping occasionally to mate or lay eggs (nobody is really sure of its sex)
-
Two other items:
Mutations Option (Sim level not Bot level) resets to 1 on every "Start New." Must start the sim, stop it, set mutation option then "Change" for the new value to take effect.
Racial memory is transferring. But only if the target .memval is 0. If a value already is present the parent's value does not get transferred. For my sims the initial values for generation 0 are randomly determined in DNA. All susequent generations need to have these values populated by mom (she gets and tests dad's value prior to mating). I see no other way to set these sex trait values (begin hint) in the absence of a working .sexrepro and its gene swap routine (end hint). If baby is a male that "initial" male gene kicks in and sluffs a value into the memloc. Dad's value (coming from mom) never gets transferred. Is there a way for the Bot to know it is a "subsequent" generation (not generation 0)?
Which brings up another item. I have not tested this but since I'm here and feeling lazy I'll just ask. (begin hint) When .sexrepro is working (end hint) I'll need to have 5 or 6 different males and 5-6 different females with various sex trait genes to start a simulation. Will the "Add Species" window in the Options menu allow a list of 15 or so "species?"
Some trivia:
Carolus Linnaeus (1700s) invented the binomial taxonomic system still in use today (Homo sapien, Canis lupus, Gallus gallus, etc.). Almost a hundred years prior to Darwin's On the Origins of Species, Linnaeus devised a classification system that so well described the bushy branching of life that Evolution grows. At the time, of course, Linnaeus thought he had uncovered the creational logic from the mind of God. So enthralled was he in his discovery that he applied the scheme to everything in nature. Including rocks. Quartzum aqueum was a "species" of transparent rock and Quartzum opacum was his name for flintstone.
Later.
-P
-
Will the "Add Species" window in the Options menu allow a list of 15 or so "species?"
Yes, it does.
-P
-
Mmmm, nice new version! Is it just me or "energy per veggie" allows much more balanced sims? No more tons of predators feeding off 3-4 veggies for me :)
Anyway, back to bugs: "Save/Load Simulation" thingie is seriously messed up. I can't even begin to describe what happens - just try it yourself... very strange stuff indeed.
"Snapshot" still does not work for me. Someone said that it takes a very long time... How long? I waited for 3 full minutes to snapshot 100 bots.
-
I haven't ever looked at or touched snapshot code. Who added it? Was it here since the dawn of time?
Load/Save routines probably do need an overhaul. I have some ideas, I'll look into them sometime.
-
PY added them a couple of months ago.
For Load/Save routines, remeber that each bot has to be saved, including all DNA, mutations and rates.
-
Is there a way for the Bot to know it is a "subsequent" generation (not generation 0)?
A bot cannot know that absolutely sure, but a gene like
cond
*.robage 0 =
*.eye5 77 =
start
bla
stop
or, since im unsure on the 77 value
cond
*.robage 0 =
*.eye5 70 >
start
bla
stop
Should do the trick
-
The program knows which generation a robot is. ie. it is stored in the robot array for each bot.
However the robot has no way at present to be able to read this value. The best way I can think of is to have the baby fire a tie at birth then in the following cycle have the mother program the baby with the new value.
Imagine that .generation (custom variable at some value) is used to define generation and .newmom (custom variable at some value) is used to tag each new birth. This sequence of genes should work out for you
'if .newmom is tagged then program the baby with a generation value
'then reset the tag
cond
*.newmom 1 =
start
1 .tienum store
*.generation 1 add .tieval store
.generation .tieloc store
0 .newmom store
stop
'make a baby and set the tag in .newmom
cond
*.nrg 6000 >
start
50 . repro store
1 .newmom store
stop
'make a tie at birth
cond
*.robage 0 =
start
.tie inc
stop
'best get rid of that tie I suppose before it becomes permanent
'do it at age 2 so that we have time for programming
cond
*.robage 2 =
start
.deltie inc
stop
Don't change the order of the first two genes or it won't work!
-
Henk, PY,
Thanks for the ideas. I like both.
I haven't seen this on the "Bugs" board before:
"Save Sim" puts a .sim file in the DBII/saves directory just fine. "Load Sim" is appearenly not working. I cannot get to the saved sim. When I "Load Sim" the field stays blank. I tried hitting the "start" arrow. Field remains blank and after 3 seconds I get the notice that all bots are dead. I tried going into the "Options" menu and hitting "Continue." Get a "Run-time Error 55 file already open."
If this is a known/repeat-of-what-you-already-knew then, well, never mind. If not, well, then now you know.
Just to add my 2-cents of limited knowledge on "Snapshot": I tried a snapshot of a 45 bot sim (including veggies) and went to get more coffee (the true elixir of all life). After 6 min I got tired of waiting. Task Manager showed DBII as "Not Responding." I've attached the resultant .txt file from the Saves directory.
Speaking of Zodiac signs (say what?), Otto Preminger had the best ever response to being asked his sign. "I am a Do Not Distuuuurb sign."
Much obliged,
-P
-
Another bug: "Disable all mutations" does not work - my veggies are mutating like crazy.
-
Another bug: "Disable all mutations" does not work - my veggies are mutating like crazy.
:blink:
Works perfectly in my copy and also in the source code.
You sure your veggies aren't using .mrepro or something?
-
"Save Sim" puts a .sim file in the DBII/saves directory just fine. "Load Sim" is appearenly not working. I cannot get to the saved sim. When I "Load Sim" the field stays blank. I tried hitting the "start" arrow. Field remains blank and after 3 seconds I get the notice that all bots are dead. I tried going into the "Options" menu and hitting "Continue." Get a "Run-time Error 55 file already open."
It would appear that for some unknow reason, everything to do with file handling has suddenly become a pile of testicular material in V2.36.
I can't think why because neither Numsgil or I have touched any of that code in this release.
There is a possibility that the work that Num did to speed up the program (changing robot address nodes) could have screwed up the way the program accesses the robot list to save them.
One of us will check it out in the next few days.
-
Another bug: "Disable all mutations" does not work - my veggies are mutating like crazy.
I just ran a quick sim with Hunter 2.2 and you are right, the buggers are mutating even with everything set to zero.
What message are you getting in the mutation detsils?
Just turned off and restarted the same sim. Now they aren't mutating at all. I wonder if this is yet another instance of the now infamous "Numsgil robot list code"? :unsure:
I have a nasty feeling that some robots might be inheriting stuff from different robots sometimes when the robot list is sorted.
-
About mutating veggies: Well, I am sure they don't use mrepro. What I am not sure about is whether they actually mutate. The "average mutation" graph certainly shows them accumulating mutations. I'll try and investigate that some more by looking at the actual DNA code.
-
My hunters actually were mutating.
Here is what I did.
1 started a sim with Hunter 2.2 only and all mutations set to 5000
2 ran it a few minutes
3 disabled mutations for Hunter 2.2 and restarted the sim
4 Hunter kept right on mutating
5 turned off DB and restarted the same sim with mutations off from the start. No more mutations?
-
I think the save/load problems and snapshot problems relate to the fact that robot DNA is no longer 1000 elements long. It's set dynamically as need arises. The snapshot doesn't know that it should stop reading from DNA when it hits the last end statement I think. I'll look into it to be sure, but that would be my guess.
PY! My robot sorting thing isn't the main culprut of all the bugs. I've changed so much, let's not blame all the bugs in one part of the code I changed!
Mutation rates I'll look into. I'm sure it's a problem of mutation rates not resetting.
-
PY! My robot sorting thing isn't the main culprut of all the bugs. I've changed so much, let's not blame all the bugs in one part of the code I changed!
I know but it is so much more fun just blaming it on one thing :P
-
Its allways usefull to have a scape goat on hand :bigginangel:
-
I even have a lab tech who is the official scapegoat.
Whenever something goes wrong, it is always Joe's fault. Even when he is on vacation at the time
It's all in fun though :D
-
Not sure what and where is broken, but 2.36.2 is very unstable for me - it always crashes in under 1 hour. The only simulation I ran on 2.36.1 lasted for over 10 hours and crashed due to "single cycle" bug. Maybe I just got lucky....
-
Alot of code was copy and pasted to make 2.36.2. If you want stability, I'd run 2.36.1. 2.36.2 only really adds a few mutation changes and other things like that.
-
Yep, that's what I did - went back to 2.36.1 and it's been running for 1.5 hours fine and dandy. Something is definately broken in 2.36.2