This will allow compilation using the older SDK until the buildbot
PS2 toolchain SDK can be upgraded and we can confirm the newer SDK
builds are working correctly.
As the GCC manual notes, "On most systems that use stabs format, -g
enables use of extra debugging information that only GDB can use;
this extra information makes debugging work better in GDB but will
probably make other debuggers crash or refuse to read the program."
-gstabs produces stabs without GDB extensions, and thus the AmigaOS
debugger will work.
The host targets for Android ARM builds are changed to "android-arm"
and "android-arm-v7a", from "android" and "android-v7a", and two new
targets are added of "android-mips" for MIPS and "android-x86" for
x86.
The older "android" and "android-v7a" targets are still supported,
but are deprecated.
This implements count badge, progress bar, and icon overlay.
It uses the NSDockTile API which is available since OS X 10.5.
The code compiles and run on older system but without doing
anything.
Formerly the taskbar support was *always* enabled except when the backend or
user specified that it should be disabled. This causes nasty crashes for
backends which do not have any taskbar support (like DC, Tizen and probably
more which simply did not disable it so far) when defaultErrorHandler was
called for example (Mass Add is also broken for those).
The SDL (and derived backends) worked around missing taskbar support by
simply faking a dummy taskbar implementation (but still claiming in configure
that we feature taskbar integration, ouch).
To avoid all non-SDL backends from manually specifying _taskbar=no I added
some auto detection code which simply only enables taskbar support in case
ScummVM is built on Win32 or libunity is present.
This changes makes ScummVM compilable with newer versions of DevKitPPC. ScummVM can be linked against the original libogc and libfat. That makes some newer WiiMotes work, improves audio-/video-playback and contains various improvements.
This caused an exception if a error() call occurs, rather than a
clean exit. This occurred in the defaultErrorHandler() function of
engines/engine.cpp, probably due to g_system->getTaskbarManager()
returning a null pointer.
This is mainly "cosmetic" to keep the SCUMM engine and subengines at
the top of the various files, but probably a good idea to prevent any
subtle regressions associated with changing the order.
Each engine now only has to provide a single configure.engine file
adding the engine into the configure script, which then produces the
required other files automatically.
This is the third and final commit enabling fully pluggable engines.
Now providing an engine folder contains a configure.engine, engine.mk
and engine-plugin.h file, it will be picked up automatically by the
configure script.
This is the first part of allowing engines to be added dynamically.
They are placed into a folder in engines/ which must contain a file
named "configure.engine" to add the engine, which is pulled into the
top level configure script automatically.
This was breaking AmigaOS4 builds for the porter as his libpng was built
as a shared library depending on zlib. This should be safe for static
builds as well.
Instead of always assuming /sw (i.e. fink) it should now also pick up
the macports prefix or /usr/local depending on which one exists.
Of course this can still be overwritten with --with-staticlib-prefix
These now restore the original release build logic to ensure that we
don't have issues with releases. Missing something here as _debug_build
doesn't seem to work as I expected...
After this change, release builds will still have serial debug disabled,
but debug builds will now not provide this unless --enable-debug is
passed explicitly. This could have been another option or environment
variable, but other embedded ports do similar things with _debug_build
and this looks like the cleanest way to deal with this.
This allows configure to work fine on Solaris 10 again. This fixes bug
#3614513 "A few posixisms have crept into configure (w/ patch)".
Thanks to lblume for his patch.
This fixes compatibility issues with clang when using -enable-Werror:
clang generates a warning when being passed both "-c" and linker-only
flags like "-L" or "-l". Combined with -Werror, this causes a spurious
configure failure.
This is triggered by the scummvm-conf.exe binary produced from the OpenGL test.
It is detected by a generic signature in AVG, Malware-bytes and several other
antivirus programs, so not easy to get this fixed.
This workaround avoids this by adding some redundant output code to the binary
which changes it enough to avoid the signature.
Freetype-config reports a broken include dir
when et_EE locale is used. We set LC_ALL
and LANGUAGE to "C" explicitly like autoconf
does to force consistent behavior.
This changes the default for Caanoo, GP2x, GP2xWiz, OpenPandora and PS2.
For those now we only disable debug symbols and enable optimizations by
default.
This allows compilation of plugins on PowerPC. This is done similar to what
was proposed in patch #3575572 "Fails to build on Linux PowerPC when using plugins".
I chose to use -fPIC unconditionally, since it will only make a difference on
platforms, which will require -fPIC, like PowerPC. On amd64 for example gcc
produces the same code.
The issue was that it cannot find a 64 bits type because 'long int' is
32 bits and compilation of the test program with 'long long' failed
due to the use of -pedantic flag (and 'long long' is not part of
ISO C++ 1998 standard). It nevertheless defined _UINT64
which subsequently told systems headers to not redefine uint64 and
caused many compilation error in system headers.
This fixes Tremor autodetection.
tremor/config_types.h typedefs ogg_int64_t, which errors out with
"ISO C++ 1998 does not support ‘long long’" when using -pedantic.
This adapts the check to be similar to the find_type_with_size check, i.e. it
assigns a value to test_array and also returns 0. Should not be required, at
least builtbot went fine, but better be safe and sorry.
Except for the PSP port, nothing is actually using libpng symbols.
It is being linked against if it is available at compile time.
This needlessly creates a dependency.
This changes the set of engine options to ./configure to:
--enable-all-engines
--disable-all-engines
--enable-engine=<engine name>[,<engine name>...]
And if plugins are enabled:
--enable-engine-static=<engine name>[,<engine name>...]
--enable-engine-dynamic=<engine name>[,<engine name>...]
... level is specified
This preserves the old behavior of ports that used to always pass in -Os or -O3
It also allows --disable-release-optimization for those ports to function
This is a holdover from when it was going to warn on a per-engine basis.
This warning is now handled on a per-game basis with ADGF_UNSTABLE and
ADGF_TESTING.
It resulted in a lot of spam in config.h: a define for every single
engine and subengine that looks like this:
define ENGINE_WIP_SCUMM_7_8 false
It was not being used in-tree and I don't see it being used in the near future.
The improvements in this commit are mostly code formatting, variable
abstraction, and in one case, a performance enhancement as calculations
were made a one-time fire rather than being run with every mouse move
event.
On Mac OS X (Darwin), there are minor differences in the 'sed'
and 'install' tools that require slightly different usage. This
distinction has been made in webos.mk, made possible by an
additional flag in ./configure.
