* include/hbexpra.c
* source/compiler/expropta.c
* source/compiler/harbour.c
* source/macro/macroa.c
* source/rtl/tget.prg
* source/rtl/tgetint.prg
* fixed to be more Clipper compatible
* source/rtl/persist.prg
* fixed to not use xBase extensions
* source/compiler/harbour.sly
! Optimized __GETA() support, and changed it to be implementated identical to Clipper.
/* Clipper passed a bGet Block which refrences the Base Array Variable and not the Variable itself.
Clipper passes NIL instead of bGet block if the Base Array is MACROVAR or MACROTEXT.
Clipper does not optimize "&Macro" to Macro, but Harbour does! */
* source/rtl/tgetint.prg
* __GETA() Changed 1st parameter aVar to bGetArray and logic to get the Base Array if bGetArray is NIL
/* Optimized to not use macro if possible (most cases) even if the GetArray is a macro. */
* tests/testget.prg
+ Added additional test.
* include/hbmacro.h
* source/macro/macro.y
* source/vm/macro.c
! Renamed hb_compParse() to hb_macroYYParse(). We already have hb_macroParse() and hcompparse()
* source/compiler/harbour.y
* Added token GET (not used) just so that harboury.h is compatible with harbour.sly
* source/rtl/tgetint.prg
* Reverted to use _1 and == NIL to not break Clipper compatibility.
/* Ryszard, if you want to re-introduce PCount() and HB_PVALUE(), I'll ask (since I wrote this original code)
that you protect it with #ifdef FLEX etc. If you do I would suggest you further protect it with #ifdef STRICT...
becuase this will defintly break strict compatability. */
* source/compiler/harbour.c
* Exported: hb_compFieldGetPos() and hb_compMemvarGetPos()
* source/compiler/harbour.sly
! Changed __GET() support to be parameter compatible with Clipper
* include/hberrors.h
* source/compiler/hbgenerr.c
+ Added Error: "GET contains complex macro"
* source/pp/pptable.c
- Removed bSetGet from rule of GET, PP output now Clipper compatible.
* source/vm/memvars.c
+ Added HB_FUNC( __ISMV ) // Return .T. if passed string as a Memory Variable.
* source/rtl/tgetint.prg
! Fixed __GET() to be 100% parameter compatible with Clipper.
! When 1st parameter (bSetGet) is NIL the bSetGet will be built internaly,
not using macro in most cases, even if the Get Variable itslef is a macro :-)
/* Ryszard, this will further break Flex support for GET, but makes __GET() 100% Clipper compatible as you suggested.
Please note, that Clipper does *NOT* pass a bSetGet *only* when the Get Var is a *simple* *non* declared Variable.
Declared Variables in this context are: MEMVAR, FIELD, LOCAL, and STATIC. For all of those, bSetGet *is* generated!
Clipper also generates bSetGet for "Complex Variables", i.e. Aliased Variables, Object Data, etc.! */
* makefile.bc
Added missing dependancy of source/rtl/tgetint.prg
* source/rtl/tgetint.prg
! Improved __GETA() - No longer uses slow function calls to traverse array.
bSetGet much faster because it no longer traverses base array every time bSetGet is evaluated,
because final array dimension is now compiled into the bSetGet :-)
- Removed no longer needed __GetAValue(
* tests/testread.prg
+ Added multi dimension GET test.
