[Freeciv-Dev] Re: (PR#10835) Trieme navigation

["Jason Short" <jdorje@xxxxxxxxxxxxxxxxxxxxx>]

<URL: http://rt.freeciv.org/Ticket/Display.html?id=10835 >

Vasco Alexandre da Silva Costa wrote:

> AFAIK the issue is that goto is now done client side, so you can have that
> fancy line telling you the path and so on. The problem, of course, is that
> the client does not necessarily know the full map at all times. Only the
> server does. So you get this, hm, interesting behaviour.

Goto has been done client-side for a long time.  But now that it's done 
with PF instead of with warmap code it is "smarter", and won't let the 
user do something "stupid".

Also the client no longer cheats by knowing unknown tiles that are 
adjacent to known ones.  I imagine PF is smart enough that it wouldn't 
cheat anyway (unlike the old warmap code which would send units on 
intelligent paths through seemingly unknown territory).

> Possible fixes:
> a) kludge the client goto to do a unit move on one-step gotos.
> b) queue a linear goto across unknown territory anyway. so it may not be
> the optimal path, who cares. if the unit can't walk inside the requested
> path the goto will abort anyway. if you really are being picky about this,
> show the non-optimal path using a different line colour.
> 
> I guess b) is more interesting.

For dangerous positions, it shouldn't be too hard.  We should just allow 
dangerous destinations (although paths should still avoid resting on 
dangerous positions).

For unknown territory, it's easy enough to allow paths to go through it. 
  The problem is then we'll get just as many bug reports complaining 
about the stupidity of the new behavior as we now get complaining about 
the stupidity of the current behavior...

jason