Re: [glob2-devel] further direction for mars and months to come

[Kai Antweiler]

> Yeah, I am also afraid that is not possible. But that won't stop me from 
> trying :)

One pitfall I remember is that gradients are currently computed every
fifth turn only.

Second pitfall, the non-and-semi-path-finding-movements. Globs behave
in a special
way when they get near other globs, or reach the flag where they
should be around,
or when they don't have any targets.

One feature that we would like to have is three-point-path-finding:
>From where I am to target number one and from their to target number
two. We currently
don't get short(est) paths for this problem. This would be good to
have for globs, that are
looking for a job.

This could also be important for you to improve speed: Only globs that
look for a job have
to check multiple target types. Globs that are assigned just have to
care about the job they
are assigned to.
-- 
Kai Antweiler