Age | Commit message (Collapse) | Author |
|
|
|
|
|
ALL: Avoid using is* macros from ctype.h
|
|
|
|
|
|
|
|
On some systems, passing signed chars to macros like isspace() etc. lead
to a runtime error. Hence, mark these macros as forbidden by default,
and introduce otherwise equivalent alternatives for them.
|
|
|
|
|
|
Thanks to Yotam for confirming that this isn't needed anymore.
|
|
|
|
|
|
This reverts commit 1e0de79cf54f2483c819584033ae3405ea7e8011.
The only available constructor for PluginManagerUncached is
protected, so this broke the builds using it.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This reverts commit ac85d134b31f770f27d085c0fe4c307e278c1875.
|
|
Keymapper now has a global keymap outside the active keymap stack.
That global keymap is always checked after the active stack has been checked.
|
|
|
|
|
|
|
|
This was removed by accident in 59739a7a0e3e4826ba7b27d5270a8d7a26b787ef.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
This doesn't make any practical difference, since the search
manager already guards against duplicate paths, but it does get rid
of a slightly confusing warning message when you have a global
extra path configured.
|
|
It was badly broken after refactoring into EventObserver.
Fitst, deinit() method was never called which lead to bad record
files. Then, the concept of counting pollEvent() calls was ignored.
Introduced dispatchPoll() method of EventObserver which is implemented
in EventRecorder. It counts calls so is able to inject events at
more proper time.
Additionally now event times are recorded.
|
|
|
|
|
|
|
|
|
|
Now SaveStateDescriptor no longer subclasses HashMap. Instead all possible
saved meta data is included directly into SaveStateDescriptor.
This is slightly less flexible, but we never needed that flexibility so far.
On the other hand it should reduce the memory usage. At least on my system
(Linux/amd64) the old SaveStateDescriptor had a size of 928 and the new
SaveStateDescriptor has a size of 200.
|
|
Silences the clang warning:
static data member specialization of '_singleton' must
originally be declared in namespace 'Common'; accepted as a C++0x
extension [-Wc++0x-extensions]
Wrapping "namespace Common {}" around the macro assignment causes clang
to complain about a spurious semicolon, and removing the semicolon at
the end of the macro causes some editors to misbehave.
Changing the requirement of using the macro in one namespace (the
global) to another (Common) seems a small price to pay to
silence a warning.
|
|
of enabling the console window.
|
|
|
|
WinCE's strtol can't actually do proper error handling because it is
lacking errno, so just check the bounds instead. If you really have
some need to pass LONG_MIN/LONG_MAX as command-line parameters then
it could always be #ifdeffed, but I'm assuming no-one cares.
|
|
- endptr needs not be initialised before calling strtol(), it is always set by strtol().
- endptr is never a null pointer after calling strtol().
- There is no need to rely on strtol() setting EINVAL (which is not guaranteed).
Testing whether strtol() parsed till the end of the string is sufficient (it is guaranteed that the string is not empty at this point).
- It is sufficient to just test for ERANGE without inspecting the return value, because a successful call to strtol() does not change errno.
Signed-off-by: Alyssa Milburn <fuzzie@fuzzie.org>
|
|
This fixes a potential problem with passing char values that would be sign-extended and yield unexpected results.
See http://msdn.microsoft.com/en-us/library/ms245348.aspx
|
|
This fixes the logic of 979fc29b (I hope) and includes some missing
headers.
|
|
|
|
This tries to make our code a bit more compliant with our code formatting
conventions. For future use, this is the command I used:
git ls-files "*.cpp" "*.h" | xargs sed -i -e 's/[ \t]*$//'
|
|
ADGF_UNSTABLE is always warned about.
ADGF_TESTING is only warned about when running
configure with --enable-relase.
Both warnings are subject to the enable_wip_game_warning
config option.
|