Age | Commit message (Collapse) | Author |
|
location
|
|
the map, which is now done by a manual key sorter on SmallMap. This fixes engine ID list sorting.
|
|
code :(
|
|
time and binary size
|
|
|
|
|
|
|
|
|
|
client so it is immediatelly clear which one you are working with.
|
|
|
|
autoreplace kernel.
|
|
|
|
NewGRF window. Presets are saved in the config file.
|
|
|
|
encoding so it uses similiar scheme as 8bpp-optimized
All zoom levels are stored and a kind of RLE is used. Together with further changes and reducing number of variables, drawing is ~50% faster in average.
|
|
got/stayed broken
|
|
for core functions (big functions use just inline instead)
|
|
there
-Codechange: convert smallvector from struct to class
|
|
TTD_[BIG/LITTLE]_ENDIAN is defined
|
|
from <string.h>
|
|
-Codechange: Split the sorting code from the sortlist to an appropriate header
|
|
|
|
|
|
something more than whitespace as description of files that don't have a description.
|
|
|
|
that bytes should be cast to ints.
|
|
|
|
operators.
|
|
have a viewport instead of one global array with a viewport for each window, even when they do not use the viewport.
|
|
a patch by Alberth.
|
|
math_func.hpp
|
|
values didn't work
|
|
use the SmallStackSafeStackAlloc wrapper. Allocate on the stack by default and on the heap for NDS (or other devices that have a very small stack).
|
|
won't be used in the future
|
|
|
|
|
|
the value
|
|
nearest now
|
|
file to spare 4-8kB of binary size
|
|
|
|
duplication.
|
|
stacks by allocating it on the heap or pushing a few kB of data onto the stack when there is a large stack.
|
|
|
|
|
|
|
|
-Codechange: Adapt the md5 algorithm to the OpenTTD source
|
|
|
|
it is really needed.
|
|
|
|
|