* utils/hbmk2/hbmk2.prg
+ Added preliminary support for autosetup for several more
compilers: win/watcom, win/pocc, win/pocc64, wce/poccarm,
dos/djgpp, dos/watcom.
Only these were tested: dos/djgpp, win/watcom, win/pocc.
This means it's now possible to "bundle" DJGPP, watcom or
pocc _compiler tools_ with Harbour and hbmk2 will be able
to detect them, or user can choose between them using
-arch/-comp switches, then hbmk2 will just be able to use
them "as is", without the need to change anything on the
environment. zero-conf usage in essence, just like we
already had for mingw family.
; NOTE: Before the release I'll rethink placement of these
tools inside the Harbour directory tree. Currently
they are detected in the Harbour root dir, but with
multiple compilers this is not ideal.
* source/vm/thread.c
! Fixed compile error after 2009-06-16 08:55 UTC+0200.
* harbour/contrib/hbct/token1.c
% use explicit constant value instead of HB_MKULONG() macro
* harbour/source/compiler/cmdcheck.c
! fixed few typos in bit manipulation used to calculate result in
PackDateTime() function by rewriting it to use bitfield union.
* include/hbextern.ch
* source/vm/thread.c
+ Added HB_MUTEXWAITERSCOUNT() to core.
(slightly rewritten using local functions)
- examples/uhttpd2/umutex.c
- No longer needed.
* harbour/source/vm/mainstd.c
* forced '\' in argv[0] instead of '/' in DJGPP builds
* disabled using SHELL envvar and forced COMSPEC in system() function
in DJGPP builds (it effects __run() and hb_run() .prg functions)
* harbour/source/pp/Makefile
! fixed / used as directory separator in $(RM) command with non *nix like
shells
* restored wildcard hack for some dynamic shortcut redirectors, f.e.
DOSEMU
* INSTALL
* Updated.
* source/pp/Makefile
! Fixed to ignore clean command result which is a TOFIX.
(anyone to help here?)
! Changed ChangeLog file detection to specifically look
for CHANGE~1 and to not look for CHANGE~* and CHANG~*
where the latter cannot exist and the former may
accidentally pick up wrong file.
* harbour/contrib/gtwvg/wvgwin.c
! fixed wrongly used & operator
* harbour/source/rtl/hbproces.c
* added TODO warning in OpenWatcom Linux builds
* harbour/make_gnu.sh
! fixed test_param() function definition - it has to be defined it's
before use
* harbour/include/hbexpra.c
* harbour/include/hbexprb.c
* harbour/source/common/expropt1.c
* harbour/source/rtl/round.c
* removed unnecessary <math.h>
* harbour/source/vm/hvm.c
* harbour/source/common/expropt2.c
* harbour/contrib/hbct/ctmath.h
* harbour/contrib/hbclipsm/num.c
* harbour/contrib/hbclipsm/numfloor.c
* harbour/contrib/hbclipsm/numceil.c
* use "hbmath.h" instead of <math.h> to avoid possible desynchronization
between used math libraries and header files
* harbour/source/pp/Makefile
! fixed real DOS compilation - replaced Windows only hack by wildcard
function usage
* harbour/config/dos/dir.cf
* removed intermediate sh call from executed commands
* harbour/source/vm/Makefile
* disabled HB_HVM_ALL in all Watcom builds - it works but the compilation
time is very huge so user who wants to use it should set envvar
HB_HVM_ALL=yes
* enabled HB_HVM_ALL for DJGPP builds
* source/pp/Makefile
! Fixed (almost) to delete include/hbverbld.h on clean.
Keeping this file there hid DOS build problems when using
Windows host.
; TOFIX: Filename in delete command for all platforms.
; TOFIX: HB_CHANGELOG value for DOS builds. It's currently
broken for MS-DOS hosts. To avoid these problems
we should use 'CHANGES' instead of 'ChangeLog'...
; TOFIX: I applied Przemek's patch to make_gnu.sh wrongly,
and have no idea how to fix it.
* include/hbextern.ch
* source/vm/hashfunc.c
+ Added HB_HGETDEF( <hHash>, <xKey>[, <xDefault> ] ) -> <value>
This function will return <xDefault> when <xKey> isn't
found in the hash table. If <xDefault> isn't passed, it
will return NIL.
* source/rtl/achoice.prg
! Partially reverted this commit:
https://sourceforge.net/apps/trac/harbour-project/changeset/10605
After this change, activating DBU menus with all items disabled
(f.e. F7/F8) caused app hang.
This line:
nUserFunc := iif( nMode != AC_NOITEM,;
Do( xUserFunc, nMode, nPos, nPos - nAtTop ), NIL )
Was now changed back to this:
nUserFunc := Do( xUserFunc, nMode, nPos, nPos - nAtTop )
ac_test2.prg still seem to function fine after this, but
I'd kindly like to ask Vladislav Lavrecky to make tests
again, maybe we need another fix here.
* source/vm/evalhb.c
* Minor formatting.
* contrib/hbtip/client.prg
* Minor formatting.
! :ReadToFile() nMode param check.
* utils/hbmk2/hbmk2.prg
* Cleanup to variable name prefixes.
* contrib/xhb/Makefile
+ contrib/xhb/xthrow.prg
+ Added THROW() function for xhb compatibility.
This is written in Harbour, based on hbcompat.ch
and has Harbour license + exception.
+ contrib/xhb/xcstr.prg
+ contrib/xhb/xdbmodst.prg
* contrib/xhb/Makefile
+ Ported following old functions from xhb:
CSTRTOVAL(), STRINGTOLITERAL(), VALTOPRG(), PRGEXPTOVAL()
+ Ported following new functions from xhb:
VALTOARRAY(), VALTODATE(), VALTOHASH(), VALTOLOGICAL()
VALTONUMBER(), VALTOARRAY(), VALTOTYPE(),
DBMODIFYSTRUCTURE(), DBIMPORT(), DBMERGE()
(borrowed from xhb, work of Ron Pinkas)
* source/debug/dbgentry.c
* contrib/xhb/hbxml.c
* contrib/hbbtree/hb_btree.c
% hb_itemRelease( hb_itemReturn() ) -> hb_itemReturnRelease()
* contrib/xhb/hbxml.c
! Applied this fix from xhb:
2008-11-06 20:00 UTC+0100 Jose F. Gimenez <jfgimenez/at/wanadoo.es>
! 'Escape' characters, such & or " were not written correctly in node attribute values
! 'Escape' characters in the form &#NN; and &#xHH; were not allowed when reading
* INSTALL
* Stripped "\devl" from all examples dirs.
+ Adjusted all examples dir to use default installation
paths where possible.
+ Adjusted examples dir to not include a version number.
+ Added OS/2 Watcom example (untested)
; These makes example much more readily usable and
avoid some more potential confusion.
* mpkg_win.bat
! Handling spaces in HB_DIR_NSIS.
* make_gnu.bat
+ Added beeps if errors were detected.
(step towards error handling)
* source/common/hbver.c
* Changed OS version for DOS to show "(Windows NT)
under NT systems (instead of former "(Windows NT/2000)").
; TOFIX: dos/owatcom builds can now created successfully
(using standard Open Watcom 1.8 installation),
the bad news is that f.e. hbmk2 will crash the
NTVDM when trying a simple make.
* INSTALL
! Minor fix to DOS instructions.
+ Added DOS build examples.
* mpkg_dos.bat
! Fixed target dir and archive name to be different
for djgpp and owatcom builds.
* make_gnu.bat
+ config/djg-make.exe
+ Added embedded copy of DOS build of GNU Make.
This is to make DOS builds, specifically dos/owatcom
builds simpler. Original filename was 'make.exe'.
This can probably be removed once we move DOS
platform status to a 'target-only' one, like WinCE/ARM.
* make_gnu.bat
! Ugly hack added to create doc/en-EN dir. This still
won't solve the DOS copy problem unless I remove
the hack committed previously. If I remove it
the make process cannot be anymore started without
this batch file as it would choke on xcopy if
above dir doesn't exist.
* make_gnu.sh
! Applied fix to only create install dirs if called
with 'install' option. At the same time enabled
this feature for all platforms. It will also create
doc/en-EN dir.
PLEASE TEST/REVIEW, I didn't do any testing.
; TOFIX: Much or all of these should IMO be done directly
from make files, it would be much self-contained
and consistent.
* source/pp/Makefile
! Fixed to find ChangeLog when building for DOS targets.
This was a problem when trying to build dos/owatcom.
Now in DOS mode it will refer to the ChangeLog by its
8.3 name.
* utils/hbmk2/hbmk2.prg
* Minor formatting.
* mpkg_deb.sh
* harbour.spec
- Deleted hbmsql.
; TOFIX: hbqt is missing from some Linux make files.
; TOFIX: (in next major release) We should try to
make Linux build files less-maintenance intensive.
* config/dos/global.cf
! Fixed to use xcopy instead of copy to 'install' files.
This means long filenames (and also casing) will be retained.
This also means that now 'install' actually works.
It may also mean that creating a DJGPP build isn't anymore
possible on a pure DOS system, I can't remember when xcopy
was introduced. Anyway I think it's better to have this
requirement than not having a proper DOS build at all
(since long years).
This also means that our DOS release cannot even be *used*
in pure DOS environment, since our long filenames will
be chopped when extracting the distro .zip. Only final
.exes are able to run under DOS.
We still have the option to implement full short name
support, but I'm not sure this has any importance anymore.
If we don't want to do this, we can finally lift some
restrictions which tied our hands so far, based on the
fact that DOS builds cannot even be created and run on
real DOS. So, this means:
- We may use long filenames as distro name.
- We may convert DOS platform as a target only platform,
similar to WinCE/ARM.
- If we do the above, we may even embed our DOS
libs in Windows distro, and build DOS apps using
hbmk2 hello.prg -arch=dos
- We may use long filename anywhere in our tree.
(but we must check first if long names are also
supported for headers, dirnames and source names)
If you have an opinion, please tell.
* config/dos/global.cf
! Adding ending pathsep to DOC_SUBDIR to make xcopy happy
and not ask if the target is a file or directory.
; TOFIX: Anyhow, DOC_SUBDIR feature simply doesn't work since
very long, so we should do something about it in the future.
* debian/changelog
+ Added new version entry. (Thanks Guy)
* bin/postinst.bat
! Added workaround to make it work under DOS.
(avoiding 'Out of environment space' error)
+ mpkg_dos.bat
+ Added DOS package generator batch. Very similar
to mpkg_win.bat, maybe we should merge them in
the future.
* mpkg_win.bat
* Using short name for main dir, just to sync it with
DOS version.
* utils/hbformat/hbformat.prg
+ Changed default GT to GTCGI.
* utils/hbmk2/hbmk2.prg
! Added workaround for dos/djgpp anomaly in hb_DirBase()
(and also hb_ProgName()) where path contains forward
slashed instead of backslashes.
; TOFIX: In core.
* source/vm/Makefile
! Disabled hbvmall for dos build because of thid djgpp error:
---
gcc -I. -I../../../../include -Wall -W -O3 -c ../../hvmall.c -ohvmall.o
In file included from ../../../../include/hbmath.h:60,
from ../../itemapi.c:97,
from ../../hvmall.c:67:
c:/devl/djgpp/include/libm/math.h:97: error: redefinition of 'struct exception'
---
* contrib/hbblat/hbblat.hbc
* contrib/hbhpdf/hbhpdf.hbc
* Minor formatting.
* INSTALL
+ Added comp.lang.clipper as public forum.
* make_gnu.bat
! Minor.
* source/pp/hbpp.c
! Added version status to banner.
* examples/superlib/hbsuper.hbp
* examples/superlib/hbsuper.dif
* examples/superlib/readme.txt
* examples/dbu/dbu.hbp
- examples/dbu/dbu.dif
+ examples/dbu/dbu52.dif
+ examples/dbu/dbu53.dif
* examples/dbu/readme.txt
* examples/rl/rl.hbp
* examples/rl/rl.dif
* examples/rl/readme.txt
* Updated instructions: New item for *nix users to convert
to lowercase and native EOL format, added hbmk2 command
for all projects, separate 5.2/5.3 DBU patches.
+ Added separate patch for C53 version of DBU sources.
* Converted all original filename references to lowercase.
! Fixed a few bugs in previous patches.
! Fixed .dif files to work on both *nix and non-*nix systems
(had to create them on *nix then convert them to have consistent EOLs)
* harbour/source/vm/eval.c
* update number of parameters in HB_EXECMSG() to keep the HVM stack
clean for any external code which can try to make some HVM stack
tracing/scanning
* harbour/source/vm/evalhb.c
! use hb_vmSend() instead of hb_vmDo()
* harbour/source/pp/pplib.c
* harbour/source/pp/pplib3.c
* harbour/source/vm/runner.c
* harbour/source/vm/hvm.c
* harbour/source/vm/thread.c
* harbour/source/vm/eval.c
* harbour/source/vm/dynlibhb.c
* harbour/source/vm/classes.c
* harbour/source/rtl/do.c
* harbour/source/rtl/xhelp.c
* harbour/source/rdd/workarea.c
* harbour/source/rdd/dbf1.c
% use hb_vmProc() instead of hb_vmDo() in all places where context is
well known as function or procedure call
* harbour/include/hbvm.h
* harbour/source/vm/hvm.c
+ added hb_vmProc() function - it's hb_vmDo() but without hack to send
messages.
TODO: remove from hb_vmDo() support for messages so it can replace
new hb_vmProc() function or if we really need such mixed common
execution function then replace in HVM calls to hb_vmDo() with
hb_vmProc()
* harbour/source/vm/classes.c
* replaced hb_vmFunction() call by hb_vmDo()
* harbour/source/vm/arrays.c
* harbour/source/vm/thread.c
! cleaned hb_vmDo()/hb_vmSend() usage
* harbour/source/vm/eval.c
! replaced hb_vmFunction() calls by hb_vmSend()
+ added .prg function hb_execMsg( <sFuncSym>, <object>, [<params,...>] )
which executes <sFuncSym> with <object> set as QSELF() value.
Mindaugas, you can use it in your HashObject to execute functions
like pseudo messages.
* harbour/source/rtl/errintlo.c
! fixed GPF inside internal error handler for errors without default
description and error codes not in the form 9000 + <n>
* harbour/source/rtl/errapiu.c
* minor formatting
* harbour/source/rdd/dbfnsx/dbfnsx1.c
! fixed stupid typo in C else block scope which caused false alarm
about corrupted index. Many thanks to Vitor Mangabeira for information
and self contain example.
* utils/hbmk2/hbmk2.prg
+ In -rebuild mode now targets will be deleted by hbmk2
before calling compilers to create them. This will resolve
some potential errors when compiler tools get confused about
an existing file with the same name but perhaps created with
another compiler. It will also make sure no attempts will be
made to create an incremental build. If target cannot be
deleted, a warning message will be shown by hbmk2.
* harbour-win-spec
* harbour-wce-spec
* harbour.spec
* mpkg_win.bat
* include/hbver.h
* doc/whatsnew.txt
* source/vm/maindllp.c
* 1.1.0dev -> 2.0.0dev
; Preparation for next rc1.
* INSTALL
* Long lines reflowed.
* harbour/include/hbclass.ch
+ extended support for xbase++ compatible method declaration:
[SYNC] METHOD <methodName1>[([<params>])] ;
[ , <methodNameN>[([<params>])] ]
* harbour/include/hbthread.h
* harbour/source/vm/thread.c
* harbour/source/vm/classes.c
+ added alternative support for SYNC method which does not use
sync mutexes. The old method is still present. I'll remove it
when xbase++ users confirm that the new one exactly emulates
xbase++ behavior.
* harbour/source/vm/classes.c
* harbour/source/rtl/tthreadx.prg
* switch to alternative SYNC method implementation which seems to be
xbase++ compatible - please test
! fixed missing thread interrupt in recent modification for ::setInterval
* utils/hbmk2/hbmk2.prg
+ Added support for hbcs= lins in .hbc files. This means
they can now be embedded. Filters and macros are accepted,
extension optional, .hbc will be added.
! Fixed wrongly calculated embedding level, counting now made
simpler.
+ Added warning message if embedding level reached the maximum,
instead of just ignoring the to be embedded files.
* COPYING
+ Added utils/hbmk2 to the COMPILER LICENSE section.
- Deleted OLD HARBOUR LICENSE section.
* source/lang/msgeu.c
* source/lang/msgsr852.c
* source/lang/msgsriso.c
* source/lang/msgsliso.c
* source/lang/msgslwin.c
* source/lang/msgsl437.c
* source/lang/msgsl852.c
* Changed copyright to new Harbour license by group (silent)
consent. Copyright holders kept intact, if that's a problem
for anyone in the future, we shall delete these files from
the repository and retranslate (or do whatever else we
decide), notice however that msgsr*.c and msgsl*.c files
were only containing translations for month and day names.
msgeu.c contained translation work spanning to 40 strings.
* source/compiler/hbusage.c
* Updated Lorenzo Fiorini's e-mail address in credit list.
* examples/rl/rl.hbp
+ Added options to create running rl executable.
* source/rtl/achoice.prg
! Fixed strange and undocumented ACHOICE() calling convention
used in RLDIALG.PRG from RL tool supplied with Clipper 5.x
versions. In this case the user function is passed as 6th
parameter.
* INSTALL
+ Added note that when compiling for mingw without msys, sh.exe
of msys musn't be in the PATH, otherwise the build won't work.
* include/hbextern.ch
* source/rtl/idlex.c
* Renamed XPP compatible SLEEP() function to XPP_SLEEP() to
avoid collision with FlagShip and CLIP implementations.
* contrib/hbwin/wapi_winbase.c
+ Added WAPI_SLEEP() function as a wrapper to Windows Sleep()
function.
* harbour/include/hbsetup.h
+ added HB_RESTRICT macro which can be used in some declarations
to reach better code optimization
* harbour/include/hbclass.ch
+ added support for xbase++ compatible method declaration by:
METHOD <methodName1> [ , <methodNameN> ]
* harbour/source/rtl/tthreadx.prg
* redirect parameters passed to ::new() method to ::init() method
as emulation for class object new method
+ added basic support for ::setInterval in xbasee++ THREAD class
* harbour/tests/speedtst.prg
! removed unused variable cDir
* contrib/xhb/Makefile
+ contrib/xhb/bkgtsks.c
+ Added background tasks functions. My only current goal
was to make it compile under Harbour, plus did some
optimizations along the way.
(borrowed from xhb, work of Francesco Saverio Giudice)
; TODO: Adapt to Harbour: MT mode, Set() replacement,
attach it to VM possibly without modifying
the VM itself.
* source/rtl/mlcfunc.c
+ Reenabled for default builds the parameter extension
in memo functions (now guarded with HB_C52_STRICT).
This breaks our effort to not introduce such extensions,
but since was no better idea I've made an exception,
but I personally still don't like it and this practice
stays forbidden.
Talking about the last parameter of these functions:
MEMOLINE(), MLCOUNT(), MLPOS(), MLCTOPOS(), MPOSTOLC()
* source/codepage/cphr437.c
* source/codepage/cphr852.c
* source/codepage/cphrwin.c
* Fixed wrongly ordered accented chars.
* Fixed missing accented chars in Windows CP.
; Submitted by Vladimir Miholic on sf.net.
* contrib/xhb/xhb.h
+ Added hb_seconds() compatibility macro.
* tests/speedtst.prg
+ Added SET DATE ANSI and SET CENTURY ON.
* tests/Makefile
+ tests/brwpos.prg
+ Added TBrowse() :rowPos test submitted by Xavi to the list.
* utils/hbmk2/hbmk2.hbp
* utils/hbrun/hbrun.hbp
% Minor opt to prev.
* tests/rto_tb.prg
+ Added hb_SToD() compatibility function to make it link
with xhb.
+ Added forced define of HB_COMPAT_C53 in xhb mode, but
commented because it makes it RTE.
* INSTALL
! Term changed to be consistent with rest of doc.
* include/hbapi.h
! Typo in comment.
* utils/hbmk2/hbmk2.prg
+ Added -echo= command line parameter. Similar to echo= in .hbc files.
! Fixed filename expansion behaviour on empty strings.
* Tweaked a little the filter parser to recognize keywords using
similar rules to Harbour compiler.
! Fixed filter parser not to RTE on malformed filters.
- examples/hbapollo/Makefile
+ examples/hbapollo/hbapollo.hbm
- examples/hboleold/Makefile
+ examples/hboleold/hboleold.hbm
- examples/hbsqlit2/Makefile
+ examples/hbsqlit2/hbsqlit2.hbm
- examples/hbgf/hbgfwin/Makefile
+ examples/hbgf/hbgfwin/hbgfwin.hbm
- examples/hbgf/hbgfos2/Makefile
+ examples/hbgf/hbgfos2/hbgfos2.hbm
- examples/hbgf/hbgfgtk/Makefile
+ examples/hbgf/hbgfgtk/hbgfgtk.hbm
- examples/hbwhat/Makefile
+ examples/hbwhat/hbwhat.hbm
* Replaced GNU Make files with hbmk2 make files. These should
now work outside the source tree. Please test them, they are
also good examples for .hbm files.
- examples/pe/Makefile
- examples/pp/Makefile
- examples/hbdoc/Makefile
- examples/guestbk/Makefile
- examples/misc/Makefile
- examples/dbu/Makefile
- examples/hscript/Makefile
- examples/rl/Makefile
- examples/Makefile
- examples/hbgf/Makefile
- Deleted unnecessary make files belonging to GNU Make system.
(these wouldn't work outside the source tree, f.e in a binary
distro which includes these examples in source form)
- examples/rdddbt
- Deleted this example as it doesn't build anymore.
* examples/hbsuper/hbsuper.hbm
! Added SVN header.
* source/rtl/Makefile
* source/rtl/errint.c
+ source/rtl/errintlo.c
* Split hb_errInternal() and hb_errInternalRaw() for
easier user override.
* harbour/harbour.spec
* harbour/mpkg_tgz.sh
+ added hbformat
* harbour/include/hbdefs.h
* harbour/include/hbapi.h
* harbour/source/common/hbarch.c
* redefined macros used to store/retrieve integer and double values
to/from byte arrays in little and big endian order.
Now GCC uses inline functions which are hardly optimized by compiler
giving on x86 machines the same code as direct casting (with the
exception to march=i686 where putting double value forces strict
alignment to avoid hidden CPU exception) but they respect strict
aliasing rules. It also means that now GCC builds always respect
strict alignment even if HB_STRICT_ALIGNMENT macro is not set.
* include/hbsetup.h
* source/common/hbver.c
+ Moved CPU detection to core header. Introduced
HB_CPU_* macros. Now they can be used to branch code
by CPU type.
* contrib/rddads/adsmgmnt.c
+ Added my copyright.
* contrib/hbwin/wapi_commctrl.c
* contrib/hbwin/wapi_winbase.c
* contrib/hbwin/wapi_winuser.c
! Fixed remaining hb_parc() vs hb_parcx() usages.
% WAPI_TABCTRL_CREATE() further simplified.
Please review, it may be wrong.
* utils/hbmk2/hbmk2.prg
* Minor internal change.
* include/hbver.ch
* source/rtl/version.c
+ Added hb_Version( HB_VER_COMPILER_CPP ) which
returns if the C compiler was used in C++ mode.
Mainly added for HB_BUILDINFO() emulation, maybe it
can be useful to set defaults in hbmk2, we will see.
* contrib/xhb/Makefile
+ contrib/xhb/xhbver.prg
+ Added emulation of HB_BUILDINFO() xhb function.
+ contrib/xhb/xhbver.ch
+ Copied xhb's hbver.ch under this name.
(borrowed from xhb, work of Andi Jahja)
* contrib/hbct/print.c
+ added PRINTSEND() support for Windows platform.
(Borrowed from xhb, work of Pavel Tsarenko, with few minor changes)
* contrib/gtwvg/wvgwin.c
* contrib/gtwvg/wvgsink.c
! Fixed two new warnings after my recent commits.
* contrib/hbwin/hbwapi.h
* Tweaked previous commit. COLORREF and HRESULT now using
hb_retnint() again to fix 32-bit numbers appearing as signed ones
at .prg level.
* source/rdd/usrrdd/rdds/arrayrdd.prg
; Added some patches recently committed to xhb:
! few fixes provided by Quique
+ Added AR_LOCK() function, but commented.
; NOTE: Francesco, would you mind reviewing these changes?
* utils/hbmk2/hbmk2.prg
* Minor formatting, comment added.
* contrib/xhb/fparse.c
* contrib/rddsql/sddmy/mysqldd.c
* contrib/rddsql/sqlbase.c
* contrib/rddsql/sqlmix.c
* contrib/rddads/adsx.c
* contrib/examples/terminal/trm_client.prg
* source/rdd/dbfnsx/dbfnsx1.c
% "" -> NULL where applicable (hb_retc*(), hb_retds(), hb_itemPutC*(),
hb_arraySetC*(), hb_errInternal(), hb_itemPutDS(), hb_arraySetDS())
* contrib/hbqt/hbqt_qtextdocument.cpp
* contrib/hbqt/hbqt_qgridlayout.cpp
* contrib/hbqt/hbqt_qpainter.cpp
* contrib/hbqt/hbqt_qeventloop.cpp
* contrib/hbqt/hbqt_qicon.cpp
* contrib/hbqt/hbqt_qtreeview.cpp
* contrib/hbqt/hbqt_qtextline.cpp
* contrib/hbqt/hbqt_qdesktopwidget.cpp
* contrib/hbqt/hbqt_qhttp.cpp
* contrib/hbqt/hbqt_qlineedit.cpp
* contrib/hbqt/hbqt_qcombobox.cpp
* contrib/hbqt/hbqt_qlistwidget.cpp
* contrib/hbqt/hbqt_qtextcursor.cpp
* contrib/hbqt/hbqt_qtextlayout.cpp
* contrib/hbqt/hbqt_qftp.cpp
* contrib/hbqt/hbqt_qclipboard.cpp
* contrib/hbqt/hbqt_qurl.cpp
* contrib/hbqt/hbqt_qwebpage.cpp
* contrib/hbqt/hbqt_qtextedit.cpp
* contrib/hbqt/hbqt_qpixmap.cpp
* contrib/hbqt/hbqt_qsplashscreen.cpp
* contrib/hbqt/hbqt_qtablewidget.cpp
* contrib/hbqt/hbqt_qtreewidget.cpp
* contrib/hbqt/generator/hbqtgen.prg
* contrib/hbqt/hbqt_qfont.cpp
* contrib/hbqt/hbqt_qmainwindow.cpp
* contrib/hbqt/hbqt_qdir.cpp
* contrib/hbqt/hbqt_qwidget.cpp
* contrib/hbqt/hbqt_qstyle.cpp
* contrib/hbqt/hbqt_qabstractbutton.cpp
* contrib/hbqt/hbqt_qobject.cpp
* contrib/hbqt/hbqt_qcolor.cpp
* contrib/hbqt/hbqt_qtextstream.cpp
* contrib/hbqt/hbqt_qwebview.cpp
* contrib/hbqt/hbqt_qboxlayout.cpp
* contrib/hbqt/hbqt_qstatusbar.cpp
* contrib/hbqt/hbqt_qtabwidget.cpp
* contrib/hbqt/hbqt_qscrollarea.cpp
* contrib/hbqt/hbqt_qtextcharformat.cpp
! Fixed to use HB_ISNUM() instead of HB_ISNIL().
% Don't call HB_ISNUM() where the default value is
zero anyway. hb_parn*() functions will return zero
anyway. (well, maybe this exploits the array-type
bug in our Extend API.)
* contrib/gtwvg/wvgwin.c
* contrib/gtwvg/wvgsink.c
* Avoiding ISNIL(). Corrected just a few occurences.
This practice is still wrong in most of the cases in
GTWVG code, so if the goal is GPF-free and consitently
behaving apps, these would better be fixed. With
string params ISNIL() is either superfluous or simply
don't protect agains GPFs, for rest of the cases it
either causes unexpected defaults (zero), or simply
superfluous for the rest of the cases. IOW: Avoid it,
use HB_ISNUM(), HB_ISCHAR() / parcx() instead.
* contrib/hbgd/gdwrp.c
* Formatting.
