[Freeciv-Dev] Re: (PR#10177) Workers has activity, which it can't continue

["Jason Short" <jdorje@xxxxxxxxxxxxxxxxxxxxx>]

<URL: http://rt.freeciv.org/Ticket/Display.html?id=10177 >

Marko Lindqvist wrote:
> <URL: http://rt.freeciv.org/Ticket/Display.html?id=10177 >
> 
>   Activity mentioned in that error message always seems to be 
> "Irrigation". This is reproducible from stable branch with attached 
> autogame.
>   I put breakpoint to the line with that freelog() -call.
> 
> 
> (gdb) print punit->type
> $13 = 2
> 
> (gdb) print game->year
> $14 = 1460
> 
> (gdb) print tile_has_special(punit->tile, S_IRRIGATION)
> $15 = false
> 
> (gdb) print punit->tile->terrain
> $18 = 8
> 
> (gdb) print is_water_adjacent_to_tile(punit->tile)
> $19 = false
> 
> (gdb) print get_tile_type(punit->tile->terrain)->irrigation_result
> $20 = 8
> 
>   So we are trying to irrigate tile that is not adjacent to water.

Presumably the adjacent irrigate tile that provided the water has just 
been changed to forest.

There is a problem with the current autosettlers.  Because any 
transformations are now considered we can end up doing some pretty 
stupid things.  Really this should be pretty simple to avoid; it's 
pretty obvious what terrains are better.  The problem I think is that 
the weights used for calculating the value of the food/shield/trade of 
the tile come from the city AI code and are not consistent.  I think 
this is a case of trying to model too much.  It would be better to 
assign simple weights like 10/5/4 that can be provided by the experts 
(10/5/4 is, more or less, what they recommended for the CMA weights). 
Although note with weights like this we will move toward entirely 
grassland and plains.

jason