Hi Phil Any change to you post gtwvw to new GT?
Regards Luiz ----- Original Message ----- From: "Phil Krylov" <[email protected]> To: "xHarbour-Developers" <[email protected]> Sent: Tuesday, December 23, 2008 4:08 PM Subject: [xHarbour-developers] ChangeLog 2008-12-23 21:05 UTC+0300 PhilKrylov <phil a t newstar.rinet.ru> > 2008-12-23 21:05 UTC+0300 Phil Krylov <phil a t newstar.rinet.ru> > * xharbour/config/global.cf > * formatting > > * xharbour/config/w32/bcc32.cf > * xharbour/config/w32/xcc.cf > ! updated for new GT and zlib libraries > > * xharbour/contrib/gtwvg/wvtutils.c > * use HB_PTRDIFF instead of UINT_PTR which is not available in some > C compilers > > * xharbour/include/hbapi.h > * xharbour/source/vm/garbage.c > * changed ULONG hb_gcIncRef( void * ) to void hb_gcIncRef( void * ). > The return value is not necessary and it may be impossible to write > machine code without external locking (mutexes) which will make > such operation in atomic MT safe way on some CPU. F.e. it's possible > to write BOOL hb_atomic_dec( int * ) in assembler for i386 but not > int hb_atomic_inc( int * ). Only void hb_atomic_inc( int * ) can be > written without additional synchronization mechanism. > > * xharbour/include/hbcompat.ch > ! fixed included header files and synced with Harbour version > If possible keep this file synced in both projects. > > * xharbour/include/hbdefs.h > * disabled HB_FORCE_EXPORT in MinGW builds - it breaks xharbour.dll > creation when GNU make is used so also -shared parameter do not work > in xhb* scripts. > > * xharbour/include/thread.h > + added assembler code for hb_atomic_inc32() and hb_atomic_dec32() > for GCC and 32bit builds. It gives as least basic support for > MT safe readonly access to complex variables in Linux 32bit builds. > > * xharbour/source/common/hbstr.c > ! fixed function name in HB_TRACE message > > * xharbour/source/compiler/cmdcheck.c > ! fixed parameter parsing - argv[0] is program name with path and > it was used to parse parameters so any '-' and '/' character in > harbour binaries name and/or paths were decoded as parameters. > > * xharbour/source/pp/ppcore.c > ! fixed accessing freed memory in line number updating > > * xharbour/source/pp/ppgen.c > ! added missing #undef HB_VER_L_USR in generated hbverbld.h file > > * xharbour/include/thread.h > * xharbour/source/vm/thread.c > * xharbour/source/rtl/hbserv.c > ! fixed support for HB_THREAD_TLS_KEYWORD - if platform supports it > then it gives performance improvement. > > * xharbour/source/rtl/pcre/Makefile > ! added include path necessary for POCC/XCC builds with GNU make > > * xharbour/source/vm/classes.c > ! fixed casting for memcpy and pClass->pFunError - by default it's > declared as function address and passing code segment pointer to > memcpy is illegal in some OSes so without valid casting such code > cannot be compiled > + added __CLS_CNTSHRDATA() > > * xharbour/include/hbtypes.h > * xharbour/source/vm/maindllp.c > * updated function declarations to current CVS > ! fixed wrong default initialization for FHANDLE. It should be > ( FHANDLE ) -1 instead of NULL > > * xharbour/utils/hbrun/Makefile > ! removed unnecessary zlib from library list - it should be set > automatically like pcrepos by *.cf files when rtl is used. > > * xharbour/tests/speedtst.prg > * updated speedtst to current Harbour code. This version allows to > test also MT performance it's quite good MT stress test - better > then dedicated ones > > (by Przemyslaw Czerpak. Thanks!) > > ------------------------------------------------------------------------------ > _______________________________________________ > xHarbour-developers mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/xharbour-developers > ------------------------------------------------------------------------------ _______________________________________________ xHarbour-developers mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/xharbour-developers
