Previously, the Micro AI behavior was inconsistent at best when it came
to dealing with these units and could even result in AI errors when an
AI unit was ambushed or a petrified unit was in the way of a move. Now,
both types of units are properly "ignored" and the AI moves have been
made robust against unexpected events such as ambushes. Incidentally,
the latter also makes the AI more robust against WML events doing
things the AI cannot know about (such as removing units).
Don’t check whether rating equals (or doesn’t equal) -9e99 or the like,
verify that a valid move was found instead. Also, avoid setting
variables to empty tables if not needed.
It’s theoretically possible that different Micro AIs’ evaluation
functions store information in the same variable within self.data.
This would only happen if the CAs have the same score and while this
should generally be avoided when setting up a scenario, it is better to
ensure that it cannot cause conflicts (not in the stored data at least,
the MAIs might still interfere with each other in other respects).
It’s slow for finding units with moves or attacks left. The
alternative method of getting all units and then looping over the table
with a condition is much faster.
It’s slow. The inverse logic using table.insert is much faster,
especially for large tables. Only kept table.remove in a couple places
where it doesn’t matter.
If a WML event takes a unit off the map, many of the AIs would
previously have run into problems by trying to access a unit that is
then not there any more. Thus, the existence of the unit needs to be
verified before continuing in cases when this would cause errors.
I had not realized that TextWrangler changes spaces to tabs when
shifting blocks of text to the left, which had been done for all CA
files when converting the engine files.
