* src/vm/fm.c
+ Overriding some additional C++ memory allocation operators
to cover the full set required by QT, according to this document:
http://doc.trolltech.com/4.5/qt-performance.html#alternative-memory-allocation
; Please report if it causes problem in other areas, also
retest HBQT memory allocation after change.
* src/vm/fm.c
! Fixed to not display "Warning, memory allocated but not released"
message for non-fmstat builds.
! Fixed to display "Memory allocated but not released: none" message
(without 'warning' sign) when built with fmstat and there was no
unrelease blocks.
* /src/vm/fm.c
* Not adequate defitions of the subsequent CRT functions for the MinGW implemetations:
#ifndef USE_DL_PREFIX
#define dlcalloc calloc
#define dlfree free
#define dlmalloc malloc
...
#endif /* USE_DL_PREFIX */
The malloc/free CRT functions should'nt be redefined in this way.
* The mesage: "Warning, memory allocated but not released..." is sent on request (//INFO cmdarg).
* contrib/hbqt/hbqt.h
+ Will now fail with forced compiler error if used with
QT libs older than 4.5.0.
* package/winuni/mpkg_win_uni_extra_copy.bat
* Formatting.
* config/global.mk
% Applied optimization sent by Tamas Tevesz.
% Applied little further optimization.
* One variable renamed.
- Deleted checking for MAKE_381 when using 3.81 features.
Now 3.81 is the minimum required so there is no point.
* contrib/hbide/hbide.prg
! Don't add binary extensions to target names.
It will be added by hbmk2 automatically and properly
according to target platform/compiler.
! Don't add .exe extension to hbmk2 executable name.
* harbour.spec
! Modified this line:
%define hb_lflag export HB_USER_LDFLAGS="${CC_HB_USER_LDFLAGS} %{?_with_static:-static}"
to this:
%define hb_lflag export HB_USER_LDFLAGS=
_with_static is later converted to HB_BUILD_SHARED setting which is right.
* Formatting.
* config/global.mk
% Applied little optimization from Tamas Tevesz.
* contrib/hbide/idemisc.prg
+ hbide_PathProc()
This function can combine relative paths together so it's
the key to avoid macros dealing with placing paths to their
intended location. It can replace current method of
<IdeSrc> = hb_dirBase(). F.e.:
hbide_PathProc( "src/mysource_in_project.prg", "projects_dir/" ) ->
"projects_dir/src/mysource_in_project.prg"
hbide_PathProc( "projects/myproject.hbi", hb_dirBase() ) ->
"C:/harbour/contrib/hbide/projects/myproject.hbi"
Please use it.
; TOFIX: Since hbide seems to intend to support a hbrun-like
command prompt, maybe it's a good idea to prefix all
public functions with hbide_, otherwise there will
be problems when name collision occurs with user code.
If this is not the goal, we can leave it.
* INSTALL
* Clean Linux .rpm section, added full install commands just
like for Ubuntu, added qt45-devel, added more description.
(not tested, I don't have .rpm based distro)
* config/global.mk
! Fixed debian package manager detection, as suggest by
Tamas Tevesz.
* config/global.mk
+ Added rudamentary package manager detection (so far for
darwin and linux). Please extend. This will allow to make
proper dependency checking in detect.mk.
* mpkg_rpm.sh
* Sorted '-with' checks.
+ Added freeimage and cairo enabler logic (none tested)
* INSTALL
- Deleted gtk2-devel from instructions.
+ Added freeimage-devel and cairo-devel to rpm pkg list.
(none tested)
* contrib/hbide/projects/hbide.hbi
! Deleted .exe extension from output name for portability.
! Fixed -workdir option to also have ${hb_plat} macro for portability.
; Pls remember that hbide is meant to be a portable tool.
; TOFIX: Delete -workdir option altogether. Not needed.
; TOFIX: Hard-wired dir C:\qt\2009.01\qt\lib should be stored in
SVN.
* contrib/hbide/hbide.ch
* contrib/hbide/hbide.hbp
* contrib/hbide/hbide.prg
* contrib/hbide/ideactions.prg
* contrib/hbide/idemisc.prg
* contrib/hbide/idesaveload.prg
+ contrib/hbide/idethemes.prg
* contrib/hbide/resources/themes.ui
+ Added class to save/load themes, a work-in-progress.
You can just play with user interface. More to come soon.
* harbour/contrib/rddads/ads1.c
! fixed softseek seeklast behavior in case seek positions at EOF
* added additional ADS function return value check in seek operation
* harbour/include/hbextcdp.ch
* harbour/include/hbextlng.ch
* harbour/src/lang/Makefile
- harbour/src/lang/msghr437.c
+ harbour/src/lang/msghr646.c
- harbour/src/codepage/cphr437.c
+ harbour/src/codepage/cphr646.c
* renamed HR437 CP and LANG modules to HR646 (CROSCII / ISO-646-YU)
* harbour/include/hbextcdp.ch
* harbour/include/hbextlng.ch
* harbour/src/codepage/Makefile
- harbour/src/lang/msgsl437.c
+ harbour/src/lang/msgsl646.c
- harbour/src/codepage/cpsl437.c
+ harbour/src/codepage/cpsl646.c
* renamed CL437 CP and LANG modules to CL646 (SLOSCII / ISO-646-YU)
* harbour/src/lang/msghriso.c
! fixed language ID
+ harbour/src/codepage/cphr646.c
* harbour/src/codepage/cphr852.c
* harbour/src/codepage/cphrwin.c
+ harbour/src/codepage/cpsl646.c
* harbour/src/codepage/cpsliso.c
* harbour/src/codepage/cpslwin.c
* harbour/src/codepage/cpsl852.c
! changed the collation order so X and Y characters are sorted before Z
It's CL5.2 ntxcro.obj compatible behavior. Now HR646 and SL646 Harbour
CPs can be used to share data with Clipper application linked with
ntxcro.obj.
* harbour/include/hbextcdp.ch
* harbour/src/codepage/Makefile
+ harbour/src/codepage/cphriso.c
+ added HRISO (ISO-8859-2) Harbour CP module
Please test all these modifications.
[TOMERGE 2.0]
* harbour/src/nortl/nortl.c
! added hb_xtraced()
* harbour/src/common/hbstr.c
* do not fill unused part of buffer with '\0' bytes in
hb_strncpy(), hb_strncpyLower(), hb_strncpyUpper(),
hb_strncpyUpperTrim(), hb_strncpyTrim()
* harbour/src/common/hbtrace.c
% small optimization in building Unicode OutputDebugString() message
* harbour/src/codepage/Makefile
* harbour/include/hbapicdp.h
+ harbour/src/codepage/ucyuscii.c
+ added YUSCII unicode table
* harbour/src/codepage/cphr437.c
* harbour/src/codepage/cpsl437.c
! fixed unicode table used in HR437 and SL437 Harbour CPs.
It's not CP-437 but YUSCII
- removed my old note "TOFIX: is it really CP-437?"
TOFIX: change Harbour names of above CPs, i.e. to HRYU and SLYU
* harbour/src/codepage/uckam.c
* removed TOFIX note I forgot to remove when committed correct
Kamenicky Unicode table.
* harbour/src/codepage/uckoi8.c
* harbour/src/codepage/uc857.c
* harbour/src/codepage/uc1250.c
* harbour/src/codepage/uc1254.c
* harbour/src/codepage/uc866.c
* harbour/src/codepage/uc850.c
* harbour/src/codepage/uc1251.c
* harbour/src/codepage/ucmaz.c
* harbour/src/codepage/uc1252.c
* harbour/src/codepage/uc8859_1.c
* harbour/src/codepage/uc88591b.c
* harbour/src/codepage/uc852.c
* harbour/src/codepage/uc737.c
* harbour/src/codepage/uc1253.c
* harbour/src/codepage/uc1257.c
* harbour/src/codepage/uc8859_2.c
* updated some of my old copyright messages
* contrib/hbide/hbide.prg
* contrib/hbide/ideactions.prg
* contrib/hbide/idedocks.prg
* contrib/hbide/idemisc.prg
* contrib/hbide/idestylesheets.prg
+ Applied the patch submitted by Vailton Renato with some omissions, thanks.
1. I do like xMate's build* icons and "gotoline" icon
which depics the meaning of action properly, just update them to
have more colors.
2. Toolbar must not be clutted with so many options.
For this purpose we have menus.
* contrib/hbide/resources/*.png
! Some new and some refined imaged from Vailton.
* contrib/hbxbp/xbpmenubar.prg
! Removed default style-sheet application which has to be
done at .prg level with oXbp:setStyleSheet().
* contrib/hbxbp/xbptoolbar.prg
! Fix to adjust toolbar button size with :imageHeight and :imageWidth.
* harbour/include/hbapi.h
* harbour/src/common/hbprintf.c
+ added new function 'int hb_printf_params( const char * format )' which
returns number of parameters necessary for format string
* harbour/src/common/hbtrace.c
* added protection against recursive calls when number of parameters
used by HB_TRACE message us bigger then 16 and fm.c module was compiled
with HB_TR_DEBUG macro
% use the same number of characters for unicode and byte string buffers
% use common memory area (union) for unicode and byte messages to reduce
C stack usage
Please test these modifications in real MS-Windows systems.
* contrib/hbide/hbide.prg
* contrib/hbide/ideactions.prg
* contrib/hbide/idedocks.prg
* contrib/hbide/idemisc.prg
+ Applied Vailton's patch. It sets the menu's in right perspective.
Actions will follow shortly.
* contrib/hbide/hbide.hbp
* contrib/hbide/hbide.prg
+ contrib/hbide/idesaveload.prg
+ Added new file separating the code for loading configuration.
Another round of reforms, more follows.
* contrib/hbide/hbide.hbp
* contrib/hbide/hbide.prg
+ contrib/hbide/idedocks.prg
+ Added new file separating the code building docking widgets.
Another round of reforms, more follows.
* contrib/hbide/idefindreplace.prg
! Minor.
* src/vm/fm.c
* include/hbapi.h
+ Added hb_xtraced(). Please check it might be wrong.
* src/common/hbtrace.c
- Deleted hack to disable winout to avoid recursive calls.
+ Added TOFIX to avoid recursive calls with some clever
detection on error case (hb_xtracd() && no_of_formatt_strings > 16)
I didn't implement it, pls do so if interested.
! Fixed MultiByteToWideChar sizeof() to HB_SIZEOFARRAY().
* contrib/hbqt/hbqt_slots.cpp
* contrib/hbqt/hbqt_slots.h
+ Added more signal/slots.
* contrib/hbide/hbide.ch
* contrib/hbide/hbide.hbp
* contrib/hbide/hbide.prg
! First step of reforms. You can expect bumps.
Please DONOT update atlease HBIDE in any way till I declare
I am done with.
* src/common/hbtrace.c
* Deleted TOFIX for va_end(). It's not required (and wasn't
ever required there) since there is no va_start() call in
hb_tracelog_() function at all.
* src/common/hbtrace.c
! Fixed some leftover causing error on non-win.
% Replaces \r\n with \n in printf() calls.
[TOMERGE 2.0]
* config/win/bcc.mk
+ Added -CP437 bcc option as default, so that it fixes compilation
on windows systems are setup to use certain (f.e. far east)
multibytes codepages as default non-Unicode CP.
* src/common/hbtrace.c
! Readded hack to disable reentrancy for win specific trace logic
until we clear up the issue with some Harbour API calls used from
win specific trace code are calling back to the trace engine.
* src/common/hbtrace.c
! Fixed to not call hb_xgrab() via HB_TRACE_CONVTO() causing
infinite recursion into trace engine. This replaces previously
committed trick to avoid the same thing in a generic manner.
It was a bug in previous Windows-support addition, which is now fixed.
* Adding \r\n for OutputDebugString() text to make WinDbg display
look readable, this time by simply extending printf() mask.
- Deleted space trimming from OutputDebugString() passed text.
Such issues should be fixed in high level code, it's not trace
engine's job to make cosmetic corrections on trace text content.
; Tested with both WinDbg and DbgView.
[TOMERGE 2.0]
; TOFIX:
va_end() has been commented for all platforms, although
it's required by all documentation I could verify.
If this causes problems for specific platforms/compilers,
we should only add this hack for those targets, and only
when HB_TR_WINOUT is enabled.
Now every non-Windows and Windows non-HB_TR_WINOUT
setups are not complying with CRTL documentation.
Could someone help here?
* src/common/hbtrace.c
! Formatting.
! Fixed to use HB_ISSPACE().
; NOTE/TOFIX: Anyhow I think last patch should be reverted to not manipulate
passed by user string before passing it to Windows.
Plus current anti-reentrancy logic should be changed to something
robust, as current solution drops trace messages which is a very
dangerous thing to do if we want to rely on trace output for
debugging. It's not clear what "stack overflow events" are.
* doc/codestyl.txt
! Fixed link. Thanks Vailton.
[TOMERGE 2.0]
* harbour/src/common/hbtrace.c
* early va_end( ap ) elimination, to avoid access violation generation in a subsequent hb_vsnprintf
* fix nested HB_TRACE calls handlig to avoid stack overflow events
* normalized debug info sending to OutputDebugString with ending CR/LF/NUL to ensure a correct view in windbg.
* contrib/hbide/hbide.ch
* contrib/hbide/hbide.prg
* contrib/hbide/ideactions.prg
* contrib/hbide/idemisc.prg
* contrib/hbide/resources/projectproperties.ui
+ contrib/hbide/resources/selectproject.ui
+ Applied Vailton Renato's submitted patch with one or two fixes
and little formatting.
This patch provides the missing functionalities which I did not
concentrated on yet. Many thanks.
This patch also implies that now HBIDE will reach to its final
goal earlier than expected. Thanks Vailton, for your jumping
in the project.
To Vailton: I was having a different concept for .ppo display
in non-modal independant widget having only one edit
object. Can you change the current way of doing so.
