Author Topic: Another interesting mutation bug  (Read 4592 times)

Offline Shen

  • Bot Destroyer
  • ***
  • Posts: 111
    • View Profile
Another interesting mutation bug
« on: April 20, 2005, 08:50:03 AM »
Another interesting yet slightly worrying mutation bug.

It appears that the start has been mutated out of the first gene, messing up the syntax. Dunno if this is intended but it could crash the sim if you start getting rid of stop and end. Bot attached

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Another interesting mutation bug
« Reply #1 on: April 20, 2005, 09:23:52 AM »
You're right, that's very worrying.  Unfortunately, I have no idea how to track it down.

I fixed a couple of bugs in 2.36.5 with weird DNA, so it shouldn't crash your program anymore (hopefully), but it's still definately not a good thing.

Offline PurpleYouko

  • Bot God
  • *****
  • Posts: 2556
    • View Profile
Another interesting mutation bug
« Reply #2 on: April 20, 2005, 09:34:40 AM »
It would certainly appear to have lost its start wouldn't it.

That needs to be looked at. Trouble is how to replicate it  :wacko:
There are 10 kinds of people in the world
Those who understand binary.
and those who don't

:D PY :D

Offline Shen

  • Bot Destroyer
  • ***
  • Posts: 111
    • View Profile
Another interesting mutation bug
« Reply #3 on: April 20, 2005, 09:44:16 AM »
Well cond/start/stop are counted as lines in the genome arnt they? It could have just said like 'delete line 5' which happens to be start rather than an actual part of the gene.

Offline Old Henk

  • Bot Destroyer
  • ***
  • Posts: 229
    • View Profile
Another interesting mutation bug
« Reply #4 on: April 20, 2005, 09:46:21 AM »
Some real life mutations are deadly, and make organisms infertile. Why not in DB?

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Another interesting mutation bug
« Reply #5 on: April 20, 2005, 09:50:08 AM »
Not quite that simple.

There are something like 14 or so functions that change the DNA through mutations.  Each can potentially delete, move around, or add lines.  Also, alot of their comments are in italian.  Makes figuring out what the heck is going on a huge chore.

Also: the DNA is an array.  Think of it like a row of houses.  To add or delete DNA sequences it's not just 'insert here'. You have to come in and physically move every house downstream of the change.  That movement process can potentally be bug ridden.  It's just not an elegant proposition.

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Another interesting mutation bug
« Reply #6 on: April 20, 2005, 09:50:59 AM »
Quote
Some real life mutations are deadly, and make organisms infertile. Why not in DB?
Well, aside from the fact that the DNA shouldn't be doing that, I'm not against bad mutations occurring.  As long as it doesn't crash the sim.

Offline PurpleYouko

  • Bot God
  • *****
  • Posts: 2556
    • View Profile
Another interesting mutation bug
« Reply #7 on: April 20, 2005, 10:05:40 AM »
Quote
Some real life mutations are deadly, and make organisms infertile. Why not in DB?

Not a very good analagy, I'm afraid.

A more suitable one would be if a single muation in a single organism were able to unravel all the physical laws of the entire universe, therby making the whole universe cease to exist.

That is what happens in DB some of the time.
Not a happy situation  :(
There are 10 kinds of people in the world
Those who understand binary.
and those who don't

:D PY :D

Offline Old Henk

  • Bot Destroyer
  • ***
  • Posts: 229
    • View Profile
Another interesting mutation bug
« Reply #8 on: April 21, 2005, 12:57:48 PM »
Ok, the it's ok. :)
so... As long as it doesn't crash the program all mutations are allowed?

Offline PurpleYouko

  • Bot God
  • *****
  • Posts: 2556
    • View Profile
Another interesting mutation bug
« Reply #9 on: April 21, 2005, 01:52:30 PM »
Quote
Ok, the it's ok. :)
so... As long as it doesn't crash the program all mutations are allowed?
Exactly.

As long as the program doesn't crash then who cares what happens in the DNA. Some really cool stuff might eventually come out of it.
There are 10 kinds of people in the world
Those who understand binary.
and those who don't

:D PY :D

Offline PurpleYouko

  • Bot God
  • *****
  • Posts: 2556
    • View Profile
Another interesting mutation bug
« Reply #10 on: April 21, 2005, 01:56:08 PM »
Quote
Well cond/start/stop are counted as lines in the genome arnt they? It could have just said like 'delete line 5' which happens to be start rather than an actual part of the gene.

As num said above, Not quite that simple.

What actually happens is that values in the DNA are split into subgroups. Start, stop, cond and end are the only members of their particular subgroup and they are supposed to be protected from any and all changes, deletions etc. except where an entire gene is duplicated or deleted.

The fact that a "start" can be removed somehow is a little concerning as it means that one of the failsafes has actually failed  :(
There are 10 kinds of people in the world
Those who understand binary.
and those who don't

:D PY :D

Offline Shen

  • Bot Destroyer
  • ***
  • Posts: 111
    • View Profile
Another interesting mutation bug
« Reply #11 on: April 21, 2005, 02:55:11 PM »
It might be something to do with just updating the dna with mutations. Im getting a lot of mutations that dont show any difference in the info panel, as well as blank space where it says its inserted something etc.

It might have just added a mutation and not updated its view of the dna. Or something. I dunno its a minefield this mutation business :(