* config/postinst.prg
! Added hack to workaround the exception where mysql .dll doesn't
work with Harbour if copied to .a file. Instead I copy over the
import .lib to .a as before. If anyone knows a better solution,
pls speak up.
* src/common/hbtrace.c
! Fixed another typo in recent modifications.
[Too bad syslog() is missing from watcom. Makes testing
much painful.]
Please test.
* harbour/contrib/rddads/ads1.c
! Fixed date handling in area putValue method for .dbf tables.
ADS does not support dates smaller than 0001-01-01 and stores
invalid values in .dbf without any error. F.e., 0000-01-01 is
stored as '-0011231', CTOD("")+5 is stored as '****1128'. Such
invalid filed values generates ADS runtime error 5095 later on
getValue area method.
DBF RDD stores empty dates in case of unsupported date range.
This fix makes ADS RDD behaviour more compatible to DBF RDD.
ADSADT behaviour is not changed because ADS allows to store
and retrieve field values for dates before 0001-01-01.
* contrib/Makefile
+ contrib/hbcups
+ contrib/hbcups/Makefile
+ contrib/hbcups/hbcups.c
+ contrib/hbcups/hbcups.hbc
+ contrib/hbcups/tests
+ contrib/hbcups/tests/hbmk.hbm
+ contrib/hbcups/tests/test.prg
+ Added basic CUPS wrapper lib, based on code idea posted on the
list by Doug.
* src/common/hbtrace.c
* INSTALL
+ Added trace output support to syslog() for *nix systems
(except watcom builds).
It's enabled by HB_TR_SYSOUT=yes envvar, which now
supercedes previous HB_TR_WINOUT setting.
Windows users pls change HB_TR_WINOUT to HB_TR_SYSOUT.
! Fixed potential problem in HB_TR_WINOUT code, where
the same vararg holder variable was used multiple
times. On *nix this caused GPF. The safest is to make
a copy of it for each usage.
* contrib/hbwin/hbwin.ch
* contrib/hbwin/Makefile
+ contrib/hbwin/tests/testevnt.prg
+ contrib/hbwin/win_evnt.c
+ Added function to add a new event to the Windows event log:
WIN_REPORTEVENT( [<cServerName]>, <cEventLog>,
<nType>, [<nCategory>], [<nEventID>],
<cString> | <acString>, [<cRawData>] ) -> <lSuccess>
* contrib/hbwin/win_prn2.c
! Fixed wce warning.
* utils/hbmk2/examples/contribf.hbc
* utils/hbmk2/examples/contrib.hbc
+ Added minizip lib.
- external/libpng
+ external/png
* external/png/Makefile
* external/libhpdf/Makefile
* external/Makefile
* harbour.spec
* contrib/hbhpdf/hbhpdf.hbc
* contrib/hbwin/hbwin.hbc
* Renamed locally hosted libpng lib name from 'libpng'
to 'png'.
This way Harbour shifts to the *nix naming and syncs
better with these systems. On Windows the "most official"
binary builds still use 'libpng', so if someone uses them
instead of locally hosted version (which is not very likely)
the .hbc files shipped with Harbour should be edited
accordingly. This change also satisfies past request from
some users.
; NOTE: INCOMPATIBLE, pls change lib name 'libpng' to 'png'
for all platforms in your make files.
* contrib/hbhpdf/tests/harupdf.prg
* Indentation.
* Formatting.
* external/minizip/readme.txt
* Deleted paths from filenames.
* src/rtl/achoice.prg
! Fixed to be more compatible with Clipper when passed wrong
parameter types to ACHOICE(). (replacing DEFAULT TO usage
with foolproof method)
! ACHOCIE() fixed to default dimensions to 0,0,0,0 like in Clipper.
! Fixed typo in ACHOICE() K_UP handling, which caused a bogus line
to appear below the bottom when the height of the achoice area
was only one line. Pls verify me for regressions.
! ACHOICE() fixed to immediately return zero (and position
cursor) when wrong type or empty array was received as
value array.
! Fixed minor inefficiency caused by wrongly calculated display
area width when rightmost colums was out of screen in ACHOICE()
call.
* src/rtl/alert.prg
+ HB_ALERT() extended to support all variables types (not
just strings) when an array is passed as first parameter:
F.e.: HB_ALERT( { "hello", 100, .T. } )
% Optimized out two local variables along the way.
* contrib/hbide/ideeditor.prg
* contrib/hbide/idemisc.prg
* contrib/hbide/idesaveload.prg
* contrib/hbide/idesources.prg
+ Implemented: save/restrore of bookmarks.
Now editing instances will be opened containing bookmarks
which were placed in previous successful exit. This is
valid for bookmarks in primarary window of editing instance, i.e.,
no split window is taken account of.
* INSTALL
* Minor updates to tool links and information.
* INSTALL
* external/Makefile
+ external/minizip
+ external/minizip/Makefile
+ external/minizip/readme.txt
+ external/minizip/minizip.dif
+ external/minizip/ioapi.c
+ external/minizip/zip.c
+ external/minizip/unzip.c
+ external/minizip/ioapi.h
+ external/minizip/crypt.h
+ external/minizip/zip.h
+ external/minizip/unzip.h
* contrib/hbmzip/Makefile
* contrib/hbmzip/hbmzip.hbc
* contrib/hbmzip/readme.txt
- contrib/hbmzip/minizip.dif
- contrib/hbmzip/ioapi.c
- contrib/hbmzip/zip.c
- contrib/hbmzip/unzip.c
- contrib/hbmzip/crypt.h
- contrib/hbmzip/ioapi.h
- contrib/hbmzip/zip.h
- contrib/hbmzip/unzip.h
* Moved embedded external code from hbmzip library to
separate, locally hosted library inside the 'external'
directory. This way we keep all external code under this
directory. This change also makes it possible to use
custom version of this external library, via the
HB_WITH_MINIZIP envvar.
NOTE: Everyone using hbmzip, must now add 'minizip'
to their lib list. (INCOMPATIBLE)
* contrib/hbide/idesaveload.prg
+ Implemented: hbIDE data files to be saved/retrieved in
system folders if hbide.ini with path is not supplied
on the command-line. The following algorythm is used:
FUNCTION hbide_getIniPath( cHbideIni )
LOCAL cPath, cIni
IF empty( cHbideIni )
IF ! hb_FileExists( cIni := hb_dirBase() + "hbide.ini" )
#if defined( __PLATFORM__WINDOWS )
cPath := hbide_DirAddPathSep( GetEnv( "APPDATA" ) ) + "hbide\"
#elif defined( __PLATFORM_UNIX )
cPath := hbide_DirAddPathSep( GetEnv( "HOME" ) ) + ".hbide/"
#endif
IF ! hb_dirExists( cPath )
MakeDir( cPath )
ENDIF
cIni := cPath + "hbide.ini"
ENDIF
ELSE
cIni := cHbideIni
ENDIF
RETURN cIni
This change makes hbIDE to be able to run on any system
installed via an installer where a writable folder is expected
to write to application data.
Thanks Viktor for the actual code and guidance.
* contrib/hbide/hbide.prg
+ Implemented: last "View" to be presented at startup.
Before it was always "Stats" view.
* contrib/hbide/idetools.prg
! Fixed ::execTool() public method to behave properly throgh
Keyboard Mappings protocol.
* contrib/hbxbp/xbpfiledialog.prg
* contrib/hbxbp/xbpfontdialog.prg
% An important fix for ancilliary dialogs, i.e., file and font,
to not be added as childrent to parent on :create().
This facilitate to destroy these dialogs after done with.
* contrib/hbxbp/tests/demoxbp.prg
! Destroying XbpFileDialog() explicitly. Now memory is released proper.
* src/rtl/hbini.prg
! Added proper type checks for all accepted parameters to avoid
(sometimes obscure) RTEs when passing wrong parameters.
* src/rtl/profiler.prg
* src/rtl/hbi18n2.prg
+ Added proper type checks instead of "DEFAULT .. TO' method.
* contrib/hbqt/THbQtUI.prg
* contrib/hbide/resources/themesex.ui
* contrib/hbide/resources/themesex.uic
* contrib/hbqt/qth/QBoxLayout.qth
* contrib/hbqt/qth/QGridLayout.qth
* contrib/hbqt/qth/QLayout.qth
* contrib/hbqt/qtgui/QBoxLayout.cpp
* contrib/hbqt/qtgui/QGridLayout.cpp
* contrib/hbqt/qtgui/QLayout.cpp
* contrib/hbide/idethemes.prg
+ Mode methods populated with custome code to accomodate detatched
parents.
Now there is no GPF absolutely neither any RTE at exit (WinXP).
It is 17Apr2010 today, exactly one year and one month since
I took to this project, and we are through.
Thanks to all who participated in this project in any way.
* contrib/hbqt/qtgui/QMainWindow.cpp
* contrib/hbqt/qtgui/QMenu.cpp
* contrib/hbqt/qtgui/QToolBar.cpp
* contrib/hbqt/qtgui/QTreeWidgetItem.cpp
* contrib/hbqt/qtgui/QWidget.cpp
* contrib/hbqt/qth/QMainWindow.qth
* contrib/hbqt/qth/QMenu.qth
* contrib/hbqt/qth/QToolBar.qth
* contrib/hbqt/qth/QTreeWidgetItem.qth
* contrib/hbqt/qth/QWidget.qth
* contrib/hbqt/generator/hbqtgen.prg
+ Applied patches provided by Francesco Perillo.
Now no GPF even if oObj:destroy() is not issued.
I confirm on WinXP and awaiting responses from all of you
on other platforms.
If all goes OK, then credit goes to Francesco and Istvan.
Guys, back them up.
* contrib/hbqt/qtgui/QLayout.cpp
* contrib/hbqt/qth/QLayout.qth
* contrib/hbqt/generator/hbqtgen.prg
+ Implemented to pick function body from .qth headers and
posted exactly the same within .cpp. Like:
QLayout.qth
bool activate () // some method before
virtual void addItem ( QLayoutItem * item ){
// Code publishes by Istavin.
}
void addWidget ( QWidget * w ) // some method after or none
Note for the opening and closing braces. These must be
last character of the method header and first character
of the methods last and separate line.
This feature facilitates to put the code dierect in headers
where we cannot generate it in automated way. This feature is
strongly discouraged to be used heavily.
* contrib/hbqt/qth/QApplication.qth
* contrib/hbqt/qth/QLayout.qth
* contrib/hbqt/generator/hbqtgen.prg
* contrib/hbqt/hbqt_hbslots.cpp
* contrib/hbxbp/hbpprocess.prg
* contrib/hbxbp/xbpgeneric.prg
* contrib/hbxbp/xbprtf.prg
* contrib/hbxbp/xbpstatic.prg
* contrib/hbide/hbide.prg
* contrib/hbide/idedocks.prg
* contrib/hbide/ideeditor.prg
* contrib/hbide/idefindreplace.prg
* contrib/hbide/idefunctions.prg
* contrib/hbide/idemisc.prg
* contrib/hbide/ideprojmanager.prg
* contrib/hbide/ideshortcuts.prg
* contrib/hbide/idetools.prg
+ This commit includes Francesco Perillo's patch, after applying some
formatting and leaving what did not worked.
+ Includes Istvan's concept of child detachment, but not made active still.
- Deleted many objects which were created in early days of hbIDE development.
+ Changed the way QApplication's init/exit procedures were implemented.
Now the behavior is accurate.
+ Applied some more control over object destruction placement.
+ Plus a lot of optimizations I cannot jot down exactly.
Basically, this commit aims at locating the GPF at exit. Now instead of
GPF, we receive two RTE's at the end, if exiting hbIDE at the quick
succession. But if you have worked into it for few seconds, it exits
normally. Please test.
David, the QPixmap bug in hbXBP on OS2 was a bug in hbXBP and is fixed.
You can test it again to verify if I am correct.
* config/win/msvc.mk
* utils/hbmk2/hbmk2.prg
* Replaced -Ot2b1 win/msvc* option with -O2 as recommended by Microsoft,
and in sync with GNU Make. This will result in somewhat larger
executable, but this gives the best overall performance. AFAICS
this change will practically allow automatic inlining to take place.
This puts msvc in the lead again compared to mingw regarding speed.
* config/wce/msvcarm.mk
* Deleted bogus option -Od turning off all optimization for
wce/msvcarm targets for new compiler versions.
* config/wce/msvcarm.mk
* config/win/msvc.mk
* Deleted Exception Handling options from all msvc cmdlines
where it was supplied. It has a heritage from the past, and
I could depict no sense in the setup:
- old wce/msvcarm: -EHsc
- new wce/msvcarm: -EHsc-
- old win/msvc*: none
- new win/msvc*: -EHs-c-
The only place this could be needed ATM is hbqt, so pls report
results after this change, or unleash any information about
proper usage of these options.
NOTE: These options were already left out from hbmk2.
* utils/hbmk2/hbmk2.prg
* Deleted -YX option for old msvc versions. (to be in sync with
this recent change in GNU Make)
* Synced msvcarm options for new compiler versions with GNU Make.
(deleted bogus -Od and some other options)
; TODO: msvc optimization options for old compiler version could
still use a revamp.
* contrib/hbqt/generator2/hbqtgen2.prg
+ Will now generate .qth files. Plus a lot of fixes.
Filtering is still needed, plus I couldnt' find out the
meaning of "New=" and "Type=" values in <CLASS> section.
* contrib/hbqt/generator2/hbqtgen2.prg
+ Refinement to QT header parser.
It should be usable now. Filters may need to be added to only
include methods/stuff relevant for HBQT.
* config/wce/poccarm.mk
* config/dos/watcom.mk
* config/win/xcc.mk
* config/win/pocc.mk
* config/win/bcc.mk
* config/win/watcom.mk
* config/linux/watcom.mk
* config/os2/watcom.mk
* Changed to not forcefully turn off warnings in some targets.
(so HB_BUILD_WARN=no is now equivalent to hbmk2 -warn=def)
; NOTE: This will result in some new warnings in /external
dir since for above compilers I've now upped the
warning level from nothing to the default level.
* external/zlib/Makefile
* Changed to not set Harbour level warnings for this
external component (in sync with all the others).
* contrib/hbide/ideprojmanager.prg
! Properly fixed old problem with executable name extraction,
leaving the ending delimiter in the extracted string.
! Added -width= hbmk2 option to cmdline to avoid splitting
the output to multiple lines (thus breaking executable name
extraction).
* contrib/hbqt/generator2/hbqtgen2.prg
+ Added more, very rudamentary code to parse QT headers.
* contrib/hbqt/hbqt_hbqplaintextedit.cpp
! Little formatting.
* contrib/hbxbp/xbpdialog.prg
+ Added :removeEventFilter() on destruction.
* contrib/hbide/resources/environments.ui
* contrib/hbide/resources/environments.uic
* contrib/hbide/resources/projectpropertiesex.ui
* contrib/hbide/resources/projectpropertiesex.uic
* contrib/hbide/resources/shortcuts.ui
* contrib/hbide/resources/shortcuts.uic
* contrib/hbide/hbide.ch
* contrib/hbide/hbide.prg
* contrib/hbide/ideactions.prg
* contrib/hbide/ideeditor.prg
* contrib/hbide/idefindreplace.prg
* contrib/hbide/idehome.prg
* contrib/hbide/ideobject.prg
* contrib/hbide/ideprojmanager.prg
* contrib/hbide/idesaveload.prg
* contrib/hbide/ideshortcuts.prg
! Removed keyboard shortcuts from menu items.
! Attempted fix on nixes when project folder is not created before
definition is saved.
! Dialog control labels: Exit => Close.
! More fixes to detect the executable from linker output ( still not perfect ).
! Changed the way project is launched. hope it will be useful for *nixes.
+ Implemented user-defined keyboard macros with following functionality:
New : whatever definition is typed-in the relevant fields is validated
with existing ones and if unique is appended at the end of list.
The macro is also validated for compilation.
Set : current hilighted macro is redefined from values from fields
after duplicate validation. The macro is also validated for compilation.
Test : Attempts to compile the macro as a block and reports back if
successful or not. It just compiles but not evaluates.
delete : deletes the highlighted macro after confirmation.
Load : loads macros from a disk-file and merges them with existing ones.
In case of duplicacy, last loaded macro is retained.
Save : Saves the macros in hbide.scu disk-file in hb_serialize()d form.
Save As: Saves the macros in a disk-file with any name but with .scu extension.
Any change is visible instantly,
Unless <Save> is clicked, current loaded macros are not saved on disk.
This allows to use temporary created macros for current session.
More info at http://hbide.vouch.info/ ( Topic: Keyboard Mappings )
* external/zlib/zlib.dif
* contrib/hbmzip/minizip.dif
* Updated .difs.
+ external/libhpdf/libhpdf.dif
+ Added new .dif.
* contrib/hbide/ideprojmanager.prg
% Deleted -hbcmp when creating .ppo. It was unnecessary since
later -hbraw mode is selected.
* harbour/external/zlib/deflate.c
* harbour/external/zlib/gzread.c
* pacified warnings
* harbour/external/zlib/zutil.h
! fixed to compile with XCC
* harbour/external/zlib/zconf.h
! removed wrongly added #if 0 / #endif
* harbour/external/zlib/gzguts.h
! added missing header files to fix _ALL_ builds
* harbour/external/zlib/Makefile
* reenabled warnings - it's very danger to pacify warnings in such way
what recent ZLIB update clearly shows - it was seriously broken in all
builds due to missing header files with valid function declarations
but the problem was fully hidden because C++ mode and warnings were
disabled so no problem was reported at compile time.
Please also remember that many of Windows compilers does not fully
support pure ANSI C function declaration so this new library may
not work as expected.
* harbour/external/libhpdf/hpdfcfg.h
! removed setting for all platforms HAVE_UNISTD_H
this file was generated by autoconf on platform which has <unistd.h>
but we cannot leave it because it will break all code which uses
standard autoconf settings like HAVE_UNISTD_H and is compiled
on platform where <unistd.h> is not available
* harbour/contrib/hbmzip/ioapi.c
! fixed to include ioapi.h before any other header files - it uses
some feature macros which have to be set before including standard
C files
* harbour/contrib/hbmzip/ioapi.h
! removed from feature set macros setting of internal __USE_* ones
! do not set _LARGEFILE64_SOURCE in DOS and OS2 - macro with the same
name is used by new ZLIB 1.2.4 for explicit 64 bit gzip API.
* harbour/contrib/hbmzip/hbmzip.c
! fixed old typos in type of return value in hb_zipfileParam() and
hb_unzipfileParam()
NOTE: I haven't tested if this new mzip/zlib 64 bit file IO support
works. It's even possible that previously working code in 64 bit
platforms (except Win64 on all other 64 bit platforms 'long' is
64bit integer) stopped to work so please make real tests.
* utils/hbmk2/hbmk2.prg
+ -info parameter will now enable output line with target
filename. This is done by default in -inc mode, so to get
it also in non -inc mode, just add -info to cmdline.
This fixed target detection in HBIDE.
* contrib/hbide/ideprojmanager.prg
! Fixed to not ignore hbmk2 path as set in options or via envvar.
! Fixed to not use -c when calling bash with .sh script as parameter.
! Deleted attempt to detect output filename by parsing link command for -o
option. This is not portable method.
! Fixed target filename detection to look for Chr( 10 ) instead of .exe.
Detection is now fine, but it still cannot launch it for other reasons.
Plus, detection only works when -inc mode is used.
! Deleted hard-coded help text of external tools.
* contrib/hbide/idemisc.prg
% Changed to not include '#!/bin/sh' in generated .sh file.
+ DirAddPathSep() renamed to hbide_DirAddPathSep() and made public.
* contrib/hbide/hbide.prg
- Deleted test code.
* contrib/hbblink/blinker.prg
! Reverted previous wrong fix. (-C doesn't work as cmd.exe option)
