Re[2]: [Gnash-commit] gnash configure.ac Makefile.am autogen.sh Chang...

[Udo Giacomozzi]

Hello Rob,

Wednesday, May 30, 2007, 5:51:49 PM, you wrote:

RS> Udo Giacomozzi wrote:

>>       * backend/sound_handler_sdl.cpp: Rearranage header order for
>>       braindead platforms.
>>       * server/URLAccessManager.cpp: Include iconv.h. Use lower case for
>>       winsock2.h.
>>       * server/asobj/NetStreamFfmpeg.cpp: Use lower case for winsock2.h.
>>       * server/asobj/xmlsocket.cpp: Use lower case for winsock2.h.
>>       * utilities/parser.cpp, processor.cpp: Include libintl.h and
>>       locale.h for braindead platforms.

RS>   Right, I'd assume this is more likely the culprit.

I did a CVS checkout from 2007-05-15 16:00 UTC (prior to your commit)
and patched all *.cpp files according to your commit.

patch ./gnash/server/asobj/NetStreamFfmpeg.cpp <NetStreamFfmpeg.cpp.patch
patch ./gnash/utilities/parser.cpp < parser.cpp.patch
patch ./gnash/utilities/processor.cpp <processor.cpp.patch
patch ./gnash/server/URLAccessManager.cpp <URLAccessManager.cpp.patch
patch ./gnash/server/asobj/xmlsocket.cpp <xmlsocket.cpp.patch


None of these patches slow down Gnash. So, it's not related to C++
code.



RS> The winsock changes
RS> we can ignore, they only effect win32 builds.

Agree.


RS> I'd be suprised if adding
RS> libintl.h, locale.h, or iconv.h would cause a performance problem, but
RS> it's more possible than the Makefile change. I needed all those changes
RS> so Gnash would build with GCC 4.3, and with Mingw32. You should be able
RS> to remove them and still build for a Linux/BSD platform with an older GCC.

It's difficult to find the patches that need each other... Hints?


Udo