- use hex-masked images make more sense, even if it's just few pixels
difference
- prevent to add image in the UNSCALED cache, since we now don't use
it for the main map
- when starting a unshrouded map, it will load in the hexed cache the
few terrains used on the minimap, so these ones will be already hexed
when scrolling the main map
...for standard terrain at default zoom level:
Previously the image type UNMASKED was a cache for hex-masking and
scaling. added a type HEXED for caching the hex-masking, when using
it skip the UNSCALED, since these hexed images will normally never be
used non-hexed.
So this new cache doesn't cost memory and hex-masking doesn't need to
be redone when zooming.
At default zoom level, we can now skip the UNMASKED cache, so will
spare a cache level in memory
In theory, we win 25%-33% of memory space (depend of the use of
BRIGHTENED) for standard terrain hexes.
Note: the hex-masking and scaling operations are now inverted, it
doesn't seems to have a visible effect.
Moving Delf away is the right thing, but he only needs to move across 1.
Also scroll away from Delf before we kill him: it still looks weird to
see him stop, regain his bars then we scroll away.
Ideally we'd have a scroll=no option to move_unit_fake...
...and sfi was supposed to be replaced by ffi throughout the block,
but wasn't. For gcc this produces a scoping error which points out the
problem. Windows compilers seem to have different scoping rules, so
this may have compiled for whoever submitted the patch.
You'll only see "Victory" if you're a player, and you'll only see the
end-of-scenario statistics when they're needed (the test has not
changed).
Count of pop-ups we see on a winning end of-scenario before going to
the next scenario is now just one (1).
- better fix for the bug "briefly empty hex if we must scroll to a
moving unit" than commit 18093.
- avoid the problem which occur when trying to draw orb of moving
temporary unit (see 2007-06-06T09:16:20Z!alinkmaze@gmail.com)
- now the orb+bars are hidden after the end of the inital scrolling
and the start of the move
- add a check for strange 1-hex path
...breaks an assert when do the first fake move in IE1.
I remove it for the moment, because i don't see the error. There is
still a bug with the lich not caused by this commit but maybe related
to the problem.