Darwinbots Forum
Code center => Bugs and fixes => Topic started by: Old Henk on March 11, 2005, 01:31:15 PM
-
Hi,
Thanks to Num I got a buggy DB 3.36. Now, as it is buggy, I like to tell about bugs I noticed. I hope the programming team can do something with them.- When selecting a bot, the selection circle is really big (at least 10 ten times to big)
- 'Flying ties'. sometimes bots have ties to, well, nothing.
- I made a bot, and somehow it wasn't capable of rotating, while the DNA mentions .aimdx very clearly.
Henk
-
The selective circle was me playing with an idea. Everything within that circle is theoretically visible to the bot (obviously things behind it aren't, but it shows the range).
The tie bug has been driving me nuts! I didn't even touch ties. Why aren't they dying right? Arg!
Post the trouble bot and I'll see what the problem is.
-
I would really apreciate it if you'd make the big selection circle optional.
The bot's called WOLF, and I've attached it.
-
Found something weird...
There's an alga_minimalis floating around, it's brown, and the dialog boz show it has 12 overall mutations:
Deleted value 50 at pos 6
Deleted value 300 at pos 7
Deleted value 1 at pos 8
Deleted value 15 at pos 9
Deleted value 5 at pos 10
Deleted value 1 at pos 11
Yet the DNA is the same as a non-mutated Alga_Minimalis...
cond
*.nrg
6000
>
start
50
.repro
store
15
.aimdx
store
stop
end
-
Bots use anti-grav walls as elevators! :)
I found a pretty cool "bug" while playing around with walls in H-grav mode. I had just placed a couple down and the things started rising. The first two eventually stoped when they went over the barrier so I decided to make a straight line of them. These also rose but when they went over the line they had the others following behind forcing them to keep moving. A cool effect, seeing the bots being lifted up by wall/platform.
The limitless shell had a bizare effect due to the increased mass. Some of the bots would fall slower because their speed limit was more restrictive(more massive). Evolution picked up on this fairly quickly and set about giving bots as much shell as possible to both help on defense and allow bots to catch up with their food.
Endy B)
-
Oh man, that's a little crazy...
Maybe there should be a universal speed limit instead of dependant on gravity?
Walls have never been working well. I have an idea of using the Egrid to place down 'cement' blocks. That should help solve a number of problems.
-
Using the e-grid to make solid walls was always it's intention.
That way we can design coral reefs and stuff with little niches that robots can colonize.
:D PY :D
-
Oops, I worded that like it was my idea, didn't I? Here I've committed one of my own pet peeves!
I should have said 'using the EGrid to create walls has been an idea for a while. I had an idea...'
-
No worries
:D PY :D
-
--------------------
LISTEN - DJ HenkX site
Don't you just love angry smilies?
:tantrum:
:cuss:
:rant:
Yup sure do. That is why I have added them to the list of available emoticons.
:D PY :D
-
Cool! I got them from www.deephousepage.com I believe, but they aren't specific deephousepage-smileys
-
Looks like PY's short gene Devincio series aren't surviving anymore. Arg, so many changes, so many bugs.
-
Figured it out. They keep making slime as long as they can. That was fine before but now, with no limits on slime...
You might want to add some conditions to your Devincio series PY.
-
Looks like a whole bunch of robots are going to need a complete overhaul.
V3.0 is causing me a lot of geif here! :angry:
All the older robots that used slime relied on the fact that it was capped at 100 and they wouldn't be charged energy if they tried to go over that amount.
At least that should be a relatively easy fix for my poor litle devincios.
I think dropping slime altogether might be in order. It drains way too much energy in a short bot anyway.
:rolleyes: PY :rolleyes:
-
Anyone tried out HDV4 or early DOM in V3.0 yet. I think they will both have the same problem.
Not to mention Devincio Eversor
:( PY :(
-
I don't think making slime is a problem when made doing:
100 *.slime sub .mkslime store
is it? :huh:
-
Naw, that looks okay. Although if slime was somehow negative, it could screw it up.
So to be absolutely safe, try 100 *.slime sub 0 floor 100 ceil .mkslime store
The ceil and floor makes sure you know the range that's going into .mkslime.
-
ok thanks. Nice features, ceil and floor. :)
-
I'm not saying I didn't like the walls doing that. I was able to get a pretty cool elevator going for the bots using them. Would there be any way, using the E-Grid maybe, to have the walls be similar to the Edges but visible to the bots. This way walls would be part of the enviroment and less like another robot.
Endy B)
-
Wouldn't be hard at all. I'll put it on my list of things to do.
-
An improved Wall system would be really cool as it would allow mazes and such... B) :clap:
-
Simulated coral reefs has always been my intention. Bots also need to be able to tie to the e-grid building blocks in the "walls" level.
Sessile invertibrates! :clap:
Rubs hands in glee!
:sly: PY :sly:
-
Anybody know where I can buy some more Glee?
Mine is running out ans WalMart don't stock it.
:D PY :D
-
100 *.slime sub .mkslime store
Yes this works fine. I modded my two devincios with that exact line to test them and they survive just fine now.
:D PY :D
-
WalMart? PY, I thought you were working from Italy...
-
No, Carlo was from Italy. PY's a brit living in Missouri (I got that right didn't I?)
edited for correct state by PY
-
o ok
-
Oh, I was so close!
Well, sorta. Missouri, North Carolina...
-
Yeah Only missed by about 800 miles I figure.
Bout the same amount that my Ant bot would miss his Queen in a size 1,000,000 sim with a 120 X 90 e-grid.
:outahere:
-
I must be thinking of Orson Scott Card (the sci-fi writer). Shows you how my brain works, confusing PY and Orson Scott Card. Yikes.
-
Wibble Wibble!
:wibble: PY :wibble:
-
Found another weird one in H-grav mode. I set up an evosim using the largest field size with a Sharefeeder bot. Most of the time all the agla will fall down like normal. Ocasionally a few would stay suspended, it appears to be possibly caused by shots going over the screen(both times there was a mass of bots at the bottom) I'm not sure exactly why this occured; but the unmutated plants had normal mass values, no buoyancy, and were unfixed. Very odd but interesting population of plants hovering far above the "ground".
Another one that probably got overlooked is the sharefeeding glitch(bots staying at 4 nrg continously) I'm pretty sure Nums and PY got busy on the next vers and forgot to change this. :D
I'll try to post the Share Feeder when I have a chance.
Endy B)
-
3.0 should have the sharenrg bug fixed. I definately took the time to hunt it down and :bash:.
-
I saw this in several different sims observing multiple bots maintaining about 4 nrg. It is possible the bots were sharing at exactly the right amount to stay at 4, but it seems unlikely.
Endy B)
-
You're sure that you've noticed this in 3.0? Not 2.35?
-
If it's updated in 3.0 why not in 3.5? Which I found it in. I just figured it used the same basic code and then you improved upon it for 3.5, am I wrong?
Endy :unsure:
-
3.5? There's only 2.35 and 2.36(AKA 3.0)
-
Alright, let me be more clear, I found them in the buggy version you sent me "3.0". My reasoning is that any bug fixed in a older vers would be fixed in a newer "version". A reasonable assumption but not necessarily true, I'll admit, if an older base file was used.
The bugs were simultaneously weird and interesting. :D
The floating Agla thing was interesting but I couldn't stabilize it. Trying to cut population down by deleting over 400 bots in one MB is not wise :) trust me.
I'm going to try to find some sort of screen capture program so I can simply send a picture of these things, instead of trying (and failing) to explain them.
Endy B)
-
Just press "printscreen" button on your keyboard :)
-
3.0 should be the latest version you have. Check to see if you see the bug in it. I have a feeling you're running an older version thinking it's a newer version. 2.35 didn't have the bug, 2.36/3.0 should have the bug fixed. There is no 3.6 or whatever.
If you find the bug in 3.0, I'll go look it up.
-
Were is 3.0 , I cant find it on the server....
-
Again (for like the umpteenth time), 3.0 is not even at an alpha version yet, so it is not actually available for public release. I did this so that new people won't go trying to run 3.0 before they run 2.35, get disgusted with all the bugs, and give up.
However, if anyone wants to see what the Rock is cooking, send me an email address where I can send a ~705KB attachment of the program as it stands and some other files.
-
Nope, it's 3.0; it has the new stuff you mentioned high speed, shot aiming, and beta E-grid. Without the new settings I'd have serious trouble convincing my computer to even think of allowing a size 12 sim.
I worked more on the floating bug and found it can affect any bot causing them to rise to the top for size 10, slightly down from the top for size 11, and about 5% down from the top for size 12. The best bot/size to observe this was Din at size 12. One Agla would start the cycle by rising to the invisible "floor" whenever it would repro the offspring has a good chance to fall down. After a while several agla would find this "floor" but as they can still fall off(not sure of cause) Din would hang out right below and eat them. Eventually the Dins would be able to follow them up reaching the same level, eating the agla there and causing the cycle to begin again somewhere else.
I think this may be related to what is occuring with the walls but I have no easy way to prove it.
Endy B)
-
You got a size 12 sim working?
I thought it crashed on anything bigger than 4.
-
I'll get to work on isolating and removing bugs.
-
You got a size 12 sim working?
I thought it crashed on anything bigger than 4.
It does!
How the f(boing) can you get above size four without an overflow error?
As soon as any robot has an x coordinate bigger than 32000 it goes bye-bye.
:( PY :(
-
Maybe he really does have 3.6. From the Future. Tell us how we finally figured out the enzyme system, will ya?
-
To everyone that has the version of 3.0 with screwy eye codes (they grow slower as bots approach, making so that *.eye5 = 20 where it used to equal 50) I've just gone through and, with alot of math, fixed it so it's much closer to the way the old eyes worked.
Which means bots from previous versions should once again work unmodified in 3.0
-
Sure it works. Nums, you're the one who sped it up enough to make it work :D .
I got interested in the idea from the talk about using larger sims for more diverse evolution. I kept increasing the size wondering at what size it would stop working and to my surprise it didn't :lol: . I haven't spotted any overflow shut downs caused by the size, so it seems to work okay.
Weird thing that "floor" I'm trying to figure out how it works so I can use it in a bot. Already I've seen a primitive feeding cycle start-up so I'm wondering what could be done deliberatly with it.
Endy B)
-
I must have fixed the bug!
Well, I probably just added code that stopped xpos and ypos from returning anything larger than 32000. But still!
I'm so excited I'm going to be running a huge sim over night to see what happens.
-
Well if so then you must have done it quite recently. My source code crashes the moment I go to size 5.
Happens right in "writesenses" where the value is assigned to .xpos.
My planned fix for this is to switch to 32bit memory locations. Then there will be no limits on any of the values in memory. And robots can store decimals too :D (unless we use long integers. I haven't figured that out in VB yet)
:D PY :D
-
[Numsgils note: Bots just recently got the full source for 2.36/3.0, for those of you lost already]
ok , I seen the source code, "Version B" is missing a lot of cint() stuff, forget about the fact that I got at least 3 copy’s , including some "Version B" from Num.
Lets have the good News first.
1.) The Program Runs a Lot Faster!
2.) The Menus are a lot cooler!
Ok now lets have the not so good news, I hate them myself though
1.) The Huge Selection circle is ok, but there is a problem, If I change my mind and want to select another robot next to the one I just selected, I cant because the selection circle is not transparent
2.) When I pause the simulation the background turns Black for some reason.
3.)The Alga amount code is completely messed up, If we add the e-grid we are going to have to get rid of it completely… (Like max energy on field is about 125, if I change the number , its still says 125, alga does not reproduce at all also.)
4.) In the e-grid menu I cant find any “Methane gas” or anything I can use for the 50/50 system.
5.)This is an old one, !> is the same thing as <= , and , !< is the same thing as >= , I have no idea why we have separate operators here , we should have 2 operators instead of 4 .
6.)Currently the mutation is not visible in the Robot Data viewer, are my new mutation types being added? were they already added?
7.)I seen nice light energy control , but I did not see my robot plasment control yet.
-
1. Yep, I know.
2. That's the E-grid. It only shows up when the sim is paused. Reason? It's really slow otherwise. PY's looking into fixing that.
3. Max energy on the field control will be gone soon. I'll keep the robot population control control though.
In my opinion, run a sim with no limits, and I think you'll see more realistic evolution.
4. The egrid is so underdeveloped right now it's not even funny. PY just went through and defined the basic way the Egrid interacts with the program. There's still alot to do with it.
5. Yup, same thing. The program even treats them as such. Why are they there? So DNA programmers can use whichever they like better.
6. What do you mean? I've seen mutations being listed before, I think.
7. I got sick and tired of the options module. Did you see all of what I changed? That was like a solid week of work.
If you want to add it yourself, feel free. The beauty of it is that the program is now so modulare we can all work on it at once.
-
Ok some of these things I haven't seen in my source code (which admitedly is a little bit different than Num's)
1) Huge selections circle. Personally I don't like this. If you want to select another robot, click somewhere else on the background and it will disappear. Now you can choose a new bot.
2) Background goes black because it is defaulting to the output display of the e-grid on level 1 (currently not even used for anything much, no matter what the label says.) Fixed in my source.
3) Haven't really checked this at all. I will have to look into it.
4) Methane gas hasn't been added to the grid yet. About the only level that works is the O2 level. Temperature also reads back some syuff I think. (at least mine does but I don't know if Num's version does yet. Can't remember if I had that working when I sent it to him.
5) That was deliberate just for variety. :P
6) Huh? Works perfectly here ???? :blink:
7) Don't know about this one. That is Num's area.
:D PY :D
-
1.) Yep I know means: you will fix it? right?
2.) Currently the e-grid is a black screen, there are no lines to Indicate where one grid square ends and the other grid square begins.
3.) I hope I will.
4.) Ok , good luck.
5.) Not cool , but ok if you added it already , no point of removing it.Just wanted to know, what you mean in the help, that “<= is the logical >” and “>= is the logical <”
6.) I see the mutations listed but I dont see the dna changing., thats the problem, Qustion was: Did Py add my new mutations yet; The mutation rates windows still looks the same -no new controls.
7.) I don't think I have the source code for that on my computer any more, I can check.... Its better that you add it yourself , I can mess somthing up. I'll look into it anyway.
-
2.) I cant say its a grid , its a black screen not a grid, Is each square supposed to turn a different shade of color?
-
1. Yup
2. Go to the menu, and try the different egrid controls. You'll see what we mean.
5. In what help file do you mean?
6. Weird. I'll look into it.
7. It's probably still attached to the post you made on this board somewhere.
-
PY and I haven't synchronized our code in a while, so neither of us have what the other is working on.
-
5.) DarwinBotsSource3.0\Darwinsource2.36\Darwinsource2.36B\DarwinBotsII.vbp , Run it....
-
"PY and I haven't synchronized our code in a while, so neither of us have what the other is working on. "
Man, Thats why I am all for making a computer company (selling it) , So we can have a networked office with SourceSafe and stuff.
-
7.) Its Probebly still on the old board, I am not going back there to search.
The fact that I have oval Insurtion as well means that I have to add a new property to the settings file incoding,. I dont want to mess with no settings file incoding, thats for sure.
:pray: Just find time to add it please...
Here for oval insurtion just to remind you what to do, use this formula for each robot
step1 Genarate Random Angle
step2 Generate Random Distance from 0 to 1
step3 Do sin and cos and the whole nine yards on it
[span style=\'font-size:8pt;line-height:100%\']x = Sin(angle)* width.of.oval * GenerateRandomDistance / Scaleconvirsionformula
y = Cos(angle)* Height.of.oval * GenerateRandomDistance / Scaleconvirsionformula[/span]
[span style=\'font-size:8pt;line-height:100%\'](Backclick and drag for oval)
(Frontclick and drag for Reguler Rectangule, the way it works now)[/span]
-
Oh, the zip files and stuff inside the source are all old backups of things.
Don't look at them.
Everything in the first folder is current.
-
I'll add the robot placement code eventually, but I'm still a little sheepish about looking at the options form code. Ugh.
-
It's kind of messy in there isn't it?
What is even worse is that at some point all this new stuff has to be run through simopts and tempopts so that the cancel button works. Then we have to update all the file saving routines.
I think UGH! describes it pretty well.
:D PY :D