Age | Commit message (Collapse) | Author |
|
|
|
|
|
|
|
|
|
|
|
|
|
with version numbering as boost autolinks them)
|
|
|
|
* topfs2/remove_jsoncpp: (21 commits)
Fix. Added explicit typecast to platform int to quench warnings for variant use in jsonrpc
[win32] replace jsoncpp with yajl
Removed jsoncpp
Added pragma and license to IClient.h
Removed the dependency of jsoncpp in CVariant
Switched to use CVariant instead of jsoncpp values in JSON-RPC
Added yajl writer for Variant
Added yajl parser for Variant
Added c_str and have size of variant work with strings
Added check for libyajl to configure
Added CVariant::type which returns the type of the current variant (easier than multiple isFoo queries)
Switched Variant to use double internally, still accepts floats
Added swap method to CVariant
Moved to use explicit type in CVariant instead of isFoo
Added operator== to CVariant
Added array and map iterators to CVariant
Added CVariant constructor which takes a string pointer and length
Added append to CVariant
Added isMember to CVariant
Removed debug in CVariant
...
Conflicts:
Makefile.in
configure.in
project/VS2010Express/XBMC.vcxproj
xbmc/interfaces/json-rpc/AudioLibrary.cpp
xbmc/interfaces/json-rpc/FileItemHandler.cpp
xbmc/interfaces/json-rpc/FileItemHandler.h
xbmc/interfaces/json-rpc/VideoLibrary.cpp
|
|
|
|
|
|
downloader
|
|
|
|
and debug libs are available but not downloaded yet (see boost_d.txt)
|
|
|
|
freetype6.dll so update there to version 2.4.4 as well. might be nice to switch to static.
|
|
|
|
|
|
solution
|
|
|
|
|
|
part of the pysqlite BW Compatiblity PR (106) but was accidentally deleted by me with a forced push.
|
|
by sqlite3 included in python 2.6 but scripts need to change the import line.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Fixed the external python to work without the Dll proxies. Fixed XBPython.cpp so that even though it's directly linked, it works correctly allowing scripts to run multiple times. This included removing the Py_NO_SHARED_ENABLED and adding the USE_EXTERNAL_PYTHON. We now need to: set PYTHONPATH in the code, fix the debug linking, move the preprocessor changes to other builds besides Debug OpenGL.
[WIN32] renamed python download deps.
[WIN32] renamed python download deps.
[WIN32] changed: we don't have debug libs for vanilla python. credit WiSo
[WIN32] copy python dll to the debug directory - credit WiSo
[WIN32] copy sdl.dll to the debug directory
[WIN32] copy python dll to dependencies. Credit WiSo
[WIN32] fixed dep script file
[WIN32] download stripped down version of python 2.6.6
[WIN32] enable external python (missing commit)
fixed some settings on the two windows release configurations
|
|
credit goes to mostly WiSo and blinkseb.
[WIN32] added include/python to the include path and removed lib/python/PC
[WIN32] added python27.lib to the linker
[WIN32] removed Py_NO_ENABLE_SHARED
Still need the python directory that was under system because linux internal python build still put things there.
There is a bug in the Windows external python because on Windows, the DLL that python is linked against may not be the DLL that xbmc is linked against. So passing a FILE* to python from an fopen has the potential to crash. This resolves that problem. Credit to blinkseb.
added: download python dependencies with our script.
changed: use Python 2.6.6 on Windows. Debug build is included in the archive, but not copied by default. Set DEBUG to true to copy these.
[WIN32] changed: in case of debug python is used, be sure to exclude it from our installer
[WIN32] fixed: get debug python working. The following has to be defined:
- Py_NO_ENABLE_SHARED
- Py_TRACE_REFS
[WIN32] fixed: oups...
[WIN32] changed: got external python work. It needs more testing to be sure. We don't link against a .lib to be able to load python from where we want (that's why Py_NO_ENABLE_SHARED is defined). We don't define USE_EXTERNAL_PYTHON because we need to use our dll loader.
Fixed a couple of problems: 1) The Makefile.in file was merged incorrectly. 2) We cannot call PyEval_AcquireLock prior to the threads being initialized in linux.
OSX shouldn't build the internal python ... ever ... but especially when USE_EXTERNAL_PYTHON is set.
|
|
|
|
|
|
|
|
|
|
the code.
|
|
|
|
|
|
|
|
|
|
|
|
instead of mingw automake. Dunno what's the difference there and why one lib compiles with the first and another with the latter only.
Now we have both. Hope they won't interfere.
|
|
(will move to our mirror later)
|
|
|
|
without version numbers (would be a symlink on Linux)
|
|
|
|
- initial make distclean in a freshly checked out tree fails because of missing Makefile/config.mak
- make distclean needs the xargs command
|
|
|
|
|