From 9d7c63add11fbeeffc9770654664701abe3f9e70 Mon Sep 17 00:00:00 2001 From: Antonio Linares Date: Thu, 11 Nov 1999 11:52:34 +0000 Subject: [PATCH] *** empty log message *** --- harbour/ChangeLog | 60 +++++++++++++++++++++++++++-------------------- 1 file changed, 35 insertions(+), 25 deletions(-) diff --git a/harbour/ChangeLog b/harbour/ChangeLog index 66e0adb312..adeadc09cc 100644 --- a/harbour/ChangeLog +++ b/harbour/ChangeLog @@ -1,3 +1,13 @@ +19991111-11:19 GMT+1 Antonio Linares + * source/vm/hvm.c + hb_stack.pItems initialized to NULL (required by fm.c module) + * makefile.b32 + + some files added. + * Notice I have left -v as a bcc32.exe compiler option + to include debugging information. + * source/rtl/classes.c + Improved memory management. + 19991111-12:02 GMT+1 Victor Szel * source/vm/dymsym.c % hb_dynsymGet(), hb_dynsymFindName() - Optimized and fixed the uppercase @@ -22,16 +32,16 @@ ! hb_xalloc() chaining bug fixed. ! hb_xrealloc() chaining bug fixed. - Removed the faulty handling of the hb_xrealloc() call with zero size, - previously the memory block was considered freed but free() was never - called. I think that until a free() call is not issued for an allocated - block we should consider it allocated regardless of it's size. This way + previously the memory block was considered freed but free() was never + called. I think that until a free() call is not issued for an allocated + block we should consider it allocated regardless of it's size. This way it's also faster and cleaner. % DeleteNode() moved to hb_xfree() since it's only used from one place now. % Casts removed from the non-FM_STATISTICS branches. 19991111-04:41 GMT+1 Victor Szel * source/rtl/fm.c - ! hb_xrealloc() several things fixed (access to the pointer after + ! hb_xrealloc() several things fixed (access to the pointer after reallocating it, uneccessary structure initializations, access to NULL pointers of failed or zero size realloc), RTL_TEST GPF fixed. ! Simplified, optimized and fixed the dump part. @@ -45,8 +55,8 @@ ! hb_xrealloc() signature check moved before any MEMINFO usage. ! Explicit \n changed to hb_consoleGetNewLine() call. ! 256 buffer size for symbol names changed to HB_SYMBOL_NAME_LEN constant. - ! Now the hb_xrealloc() is returning a NULL pointer (instead of trash) - when the passed length is zero, and it's also freeing the pointer with + ! Now the hb_xrealloc() is returning a NULL pointer (instead of trash) + when the passed length is zero, and it's also freeing the pointer with free(). % hb_xrealloc() - pResult variable optimized out. * source/rtl/hbsetup.h @@ -60,12 +70,12 @@ * include/hbtrace.h ! Every macro names prefixed with HB_ * source/vm/hvm.c - ! Explicit buffer sizes are now determined using the HB_SYMBOL_NAME_LEN + ! Explicit buffer sizes are now determined using the HB_SYMBOL_NAME_LEN constant. They can't overrun anymore. * source/vm/dynsym.c % hb_dynsymGet(), hb_dynsymFindName() - Using local stack for temp symbol name buffer instead of hb_xgrab()/hb_xfree(). Some optimalizations. - Now they will cut the symbol name to the HB_SYMBOL_NAME_LEN size, so + Now they will cut the symbol name to the HB_SYMBOL_NAME_LEN size, so by setting this to 10, we're 100% Cl*pper compatible from this respect. These functions are now faster. @@ -73,14 +83,14 @@ * include/hbtrace.h % Excluded trace calls are converted to nohting instead of "do {} while 0" since the latter was generating jumps for MINGW32. This is now violating - a -pedantic warning in GCC (duplicated ";"), so this may not be the + a -pedantic warning in GCC (duplicated ";"), so this may not be the perfect solution, but at least it doesn't generate any code. * source/rtl/fm.c ! Casts added. ! ulProcLine changed to USHORT. + Added signature support. + The whole mem tracing stuff can be excluded by a #define. - + The memory trace info level is HB_TR_ERROR now, so that it gets logged + + The memory trace info level is HB_TR_ERROR now, so that it gets logged by default. * tests/Makefile + fsplit.prg added. @@ -94,7 +104,7 @@ - Removed #defines which were previously moved to hbsetup.ch. * include/hbextern.ch source/runner/stdalone/external.prg - + __PREPROCESS added as EXTERNAL + + __PREPROCESS added as EXTERNAL 19991110-11:08 GMT+1 Bruno Cantero * include/rddapi.h @@ -123,30 +133,30 @@ Tue Nov 09 11:56:05 1999 Gonzalo A. Diethelm - * include/hbsetup.h: - * source/rtl/gt/gtsln.c: - * source/rtl/gtxxx.c: - * source/rtl/mousexxx.c: + * include/hbsetup.h: + * source/rtl/gt/gtsln.c: + * source/rtl/gtxxx.c: + * source/rtl/mousexxx.c: Added an initial implementation of GT over S-lang. - - * source/rtl/inkey.c: + + * source/rtl/inkey.c: Changed the order in which the #ifdef/#endif checks are done, to make sure the curses and S-lang implementations are picked up correctly. - - * config/linux/gcc.cf: - * config/w32/gcc.cf: - * config/w32/mingw32.cf: + + * config/linux/gcc.cf: + * config/w32/gcc.cf: + * config/w32/mingw32.cf: Added initial support for easily switching between curses and slang. This could be vastly improved. - - * tests/testbrw.prg: + + * tests/testbrw.prg: Added a piece of code bound to K_TAB, just for the fun of it... 19991109-01:58 GMT+1 Victor Szel * samples/Makefile samples/cccppc/*.* - + Fully CA-Clipper 5.x compatible PP written in Clipper added to the + + Fully CA-Clipper 5.x compatible PP written in Clipper added to the samples directory. It also has a special object system (SmallTalk like if I'm not mistaken), and it's a good test for the Harbour compiler. Note that Harbour can't yet compile it, due to hangs in several .PRG. @@ -165,7 +175,7 @@ Tue Nov 09 11:56:05 1999 Gonzalo A. Diethelm 19991108-23:41 GMT+1 Victor Szel * source/rtl/filesys.c - % hb_fsWrite(), hb_fsReadLarge(), hb_fsWriteLarge() double var + % hb_fsWrite(), hb_fsReadLarge(), hb_fsWriteLarge() double var assigments optimized. * Some small cleanups. * tests/rtl_test.prg