| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
|
|
|
|
outside of the bounds of a 32 bits integer
|
|
|
|
do not have a strings.h at all, most define the functions in string.h as well.
|
|
|
|
platforms having wchar for SQChar and others just char; always use char (and UTF-8) like in the rest of (internal) OpenTTD
|
|
the return is not NULL)
|
|
|
|
|
|
variables in the SSE blitters (MJP)
|
|
XP and we can safely assume it as the target platform.
|
|
not the normal ANSI code page and definitely not UTF-8 either.
|
|
|
|
define them if missing.
|
|
a base member initializer list as we don't use it in any problematic way.
|
|
Macros and side effects don't mix, especially if there's some obscure '#define min' in a windows header that nobody thinks of.
|
|
others were still regular inline), but make sure inline is always a 'forced' inline (I am looking at you MSVC)
|
|
out some indirect calls
|
|
don't need to cast to void/non-const before being able to free
|
|
|
|
overflow of the argument to malloc. (monoid)
|
|
|
|
PoolItem::CanAllocateItem() before actually allocating it
|
|
|
|
with the same effect
|
|
|
|
|
|
(Windows/OpenTTD) if it exists
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
with gcc3.3 and removes the need of assert_tcompile()
|
|
"non-placement deallocation function [is] selected for placement delete", or in other words delete(void *, size_t) is 'magic'.
We implemented these delete(void *, size_t) operator functions because MSVC warned that "no matching operator delete found; memory will not be freed if initialization throws an exception" for new(size_t, size_t).
This disables MSVC warning about this because we do not use exceptions in the (constructors that use the) overridden allocation functions, as such they will never be called; delete(void *) remains necessary though.
|
|
OTTD's traditional assert_compile() does not work inside templates for gcc compilers, static_assert() does though.
The new assert_tcompile() resolves into static_assert() if present, or into a runtime assert() else.
|
|
CrashLog infrastructure.
|
|
and libpng compiles and links (for me). Something's fishy with the network so it doesn't actually work (yet)
|
|
if available
|
|
I don't have all exotic machines so it might break compilation on the more exotic configurations
|
|
can be reused by other crash logger implementations
|
|
windows does crash reports instead of using several that aren't necessarily equal
|
|
|
|
|
