* harbour/src/rtl/valtype.c
* harbour/include/harbour.hbx
+ added new PRG function HB_ISEVALITEM()
It returns true if item can be evaluated by EVAL() function,
i.e. for codeblocks and function symbols.
In the future it can be extended to accept objects with :EVAL()
methods.
* harbour/src/common/hbffind.c
* ulAttr -> nAttr
+ added support for softlinks in * nix systems.
Now they are reported with "L" attribute by directory() function.
This modification also fixes directory scanning interrupted by
dummy softlinks.
* harbour/contrib/xhb/dirrec.prg
! fixed double added extension to file mask
* harbour/src/3rd/pcre/Makefile
* disabled JIT support in default builds.
It fixes build process for platforms which do not support MT mode
like some old Linux distors. I also made runtime tests with some
simple expressions and on x86 based CPUs the difference is really
minor. PCRE compiled with JIT is a little bit faster in regex
compilation (~5%) but slower in regex evaluation (10%).
AFAIR completely different results then suggested by documentation.
For tests I used rather simple expression on small data set:
cData := hb_TSToStr( hb_datetime() )
cExp := "[:]([a-zA-Z]|[0-9])[0-9][:]..[.]"
But such short expressions and data are quite common in real
programs.
If someone wants to enable JIT in his build then he can make it
using SUPPORT_JIT macro, i.e.:
export HB_USER_CFLAGS=-DSUPPORT_JIT
* contrib/rddads/ads1.c
* contrib/rddads/adsfunc.c
* contrib/rddads/adsx.c
* src/common/hbfsapi.c
! deleted explicit newlines at EOL in HB_TRACE() calls
* INSTALL
+ added new '0. GUARANTEES AND LIABILITY' section
! updated instructions to create debug build
* utils/hbmk2/hbmk2.prg
* minor cleanups
* utils/hbmk2/Makefile
+ enabled dynamic version of hbmk2 only for mingw builds.
Rest of the targets may or may not work due to CRTL
usage (fprintf()) from HB_COMPILE*() calls which causes
GPF in currently generated shared builds, most probably
due to double copy of statically linked CRTL code.
mingw uses MSVCRT.DLL, so it's not prone to this.
Solution is either to avoid file handling CRTL calls
in compiler code, or to tweak -shared build setting to
death to make it work for all non-mingw compilers.
Both of these is very difficult.
This means that dynamic module loading from scripts
will only work in hbmk2 built with mingw.
; TOFIX: 'make clean' on contribs doesn't work with mingw,
because harbour dll is deleted before hbmk2 is
called, so it cannot be started for cleanup.
* contrib/hbcurl/hbcurl.hbp
* contrib/hbssl/hbssl.hbp
+ added experimental support to install 3rd party .dll
dependencies for these libs into the Harbour bin dir.
Enabled with new build option, not yet documented,
see it in the patch.
* src/rtl/cdpdet.prg
+ request linkage of all codepages. I chose this to ensure
the detection code can use all CPs automatically and it
won't be unusable if users forget to request CPs
themselves. This has an overhead if someone uses CP
autodetection, but it makes it foolproof.
* utils/hbmk2/hbmk2.prg
* minor formatting
* contrib/hbamf/hbamfobj.prg
* made it more friendly with grep when finding undocumented functions
by removing #translate.
* src/vm/hvm.c
! added HB_STACK_TLS_PRELOAD
* include/harbour.hbx
! added __ITEMSETREF(), __VMITEMID()
* src/rtl/valtoexp.prg
* hb_setItemRef() -> __itemSetRef(). Please tell if it's meant to
be a user-accessible, public core API
* utils/hbmk2/hbmk2.prg
! fixed to ignore HB_INSTALL_PREFIX variable in runner mode.
This points to invalid places while running as part of GNU
Make and prevents autodetection.
TODO: Recheck if it now works without embedded headers.
TODO: Obsolete HB_INSTALL_PREFIX also in hbmk2 mode.
* harbour/include/hbexprb.c
! fixed hb_arrayToParams() used in array index context
( var[ hb_arrayToParams() ] ) unintentionally disabled by
2010-11-08 23:37 UTC+0100 Przemyslaw Czerpak (druzus/at/priv.onet.pl)
* harbour/src/vm/hvm.c
+ added __vmItemId() function.
* harbour/src/rtl/valtoexp.prg
+ added support for array, object and hash references to hb_valToExp().
Now this function works correctly also for items with cyclic
references.
* bin/3rdpatch.hb
* contrib/make.hb
* changed to be executed by hbmk2
- contrib/hbrun/extdyn.prg
- contrib/hbrun/hbrun.prg
- contrib/hbrun/hbrun.rc
- contrib/hbrun/headers.prg
- contrib/hbrun/p_ext.hb
- contrib/hbrun/plugins.prg
* contrib/hbrun/hbrun.hbp
% changed to use hbmk2 sources and make file to build hbrun,
implementing the build method described in previous ChangeLog
entry. The lost feature is netio management plugin, which
is now not loaded by default. Plus, since this is a static
build, dynamic extensions don't work.
* contrib/make.hb
* src/pp/ppcore.c
* deleted reference to hbrun in comment
* contrib/hbnetio/utils/hbnetio/netiocon.prg
* contrib/hbnetio/utils/hbnetio/netiomgm.hb
* __hbrun_plugin() -> __hbshell_plugin()
* INSTALL
* minor in previous change
* harbour/include/hbset.h
* harbour/src/vm/set.c
% removed HB_SET_OSCODEPAGE
% replaced HB_SET_DBCODEPAGE with hb_set_dbcp
* generate RTE when wrong parameter type (not string or NIL) is
passed to HB_SET_LANGUAGE, HB_SET_CODEPAGE, HB_SET_OSCODEPAGE,
HB_SET_DBCODEPAGE
* harbour/src/rtl/langapi.c
* generate RTE 1303 when not existing lang module name is passed to
hb_langSelectID() C function. It affects on all upper level
code using this function. NULL is accepted as valid parameter.
* harbour/src/rtl/cdpapi.c
* generate RTE 1302 when not existing codepage name is passed to
hb_cdpFindExt() C function. It affects on all upper level
code using this function. NULL is accepted as valid parameter.
If programmer does not want to generate RTE when wrong codepage
name is passed then he should use hb_cdpFind() instead.
* harbour/src/rtl/cdpapihb.c
* use hb_cdpFindExt() instead if hb_cdpFind() in HB_CDPUNIID()
function.
* harbour/src/rtl/cdpapihb.c
* harbour/contrib/hbexpat/internal.c
! use hb_parc() instead of hb_parcx() to get codepage name
from optional parameter.
* harbour/utils/hbmk2/Makefile
! fixed build process, now hbmk2 need HB_LIBS_MT_RDD
* harbour/src/rtl/tpersist.prg
% optimized decoding code which could be two slow with UTF8EX or
similar CPs.
* contrib/hbhttpd/log.prg
* contrib/hbtip/log.prg
* contrib/hbziparc/ziparc.prg
* src/rtl/hbdoc.prg
* src/rtl/hbi18n2.prg
* src/rtl/memvarhb.prg
* src/rtl/tlabel.prg
* src/rtl/treport.prg
% use HB_FNAMEEXTSETDEF() instead of manual logic.
it also fixes RTEs in hbziparc when passed non-string
filename to nearly any of its APIs.
* utils/hbi18n/hbi18n.hbp
* missed -shared enabler in .hbp
* utils/hbmk2/Makefile
* utils/hbmk2/hbmk2.hbp
+ enabled -shared build for hbmk2
* utils/hbmk2/hbmk2.prg
% consolidated .hbc finder logic
% moved 'hbmk' structure initializations to subfunctions
+ added Harbour installation autodetection for hbmk2's
runner mode. It's copy-paste code yet.
+ added automatic include path configuration in hbmk2's
runner mode. It means that now #require-d extensions
will have their include paths setup, so their header
will be found, so they can be used now.
; I more and more see it a reality to integrate hbrun
functionality into hbmk2. #require logic needs
much of hbmk2's facilities, and hbmk2 already has
basic runner capabilities. Contrib libs (and plugins)
will all have to be loaded dynamically in such case,
but since it works well, it should not be a problem.
Finally hbmk2 can be the utility that runs scripts
dynamically and also able to build an exe from them,
using the exact same source code, without any external
configuration, if the source code provides "#require"
clues. All it needs is both dynamic and static versions
of extensions (=contribs or addons).
* harbour/contrib/hbct/misc1.c
% optimized XTOC()
* harbour/doc/xhb-diff.txt
* added some additional description to the section
OOP AND CLASS OBJECT/CLASS MESSAGES
* harbour/doc/Makefile
* added xhb-diff.txt
* harbour/include/hbapi.h
* harbour/src/vm/arrays.c
+ added new C function hb_arrayFromId()
* harbour/contrib/hbqt/qtcore/hbqt_bind.cpp
* use hb_arrayFromId() instead of local hb_arrayCreateClone()
with hb_arrayPushBase() inside
* do not use hbvmint.h
! fixed few bugs in this code. I haven't analyzed this code deeply
and how it's used so I cannot say it's correct. I only fixed few
completly wrong places which caused HVM stack corruption, internal
GC item list corruption or were completly dummy calls.
* include/harbour.hbx
* src/vm/dynsym.c
+ added HB_ISFUNCTION( <cName> ) -> <lExists>
uses Przemek's code from hbfship/ISFUNCTION() with minor
adaptation for VM.
Use it instead of __DYNSISFUN() and TYPE( <cName> ) == "UI".
* include/harbour.hbx
* formatted HB_IS*() functions used for type checking
* contrib/hbfship/isfunc.c
% converted ISFUNCTION() to wrapper for HB_ISFUNCTION()
* contrib/hbide/ideplugins.prg
* contrib/hbrun/hbrun.prg
* contrib/hbxpp/xppop.prg
* contrib/xhb/xhbcomp.prg
* contrib/xhb/xhberr.prg
* examples/hbdoc/hbdoc.prg
% Using HB_ISFUNCTION() instead of __DYNSISFUN() and TYPE( <cName> ) == "UI"
* src/rtl/cdpdet.prg
! *nix specific CP autodetection code missed to call
a local conversion subfunction.
* src/rtl/gtfunc.c
* HB_SETDISPCP(), HB_SETTERMCP(), HB_SETKEYCP() will
now not RTE when passed NIL as 1st paramater. This
makes it smoother to fall-back on purposedly invalid
values, f.e. return values of new HB_CDPOS() and
HB_CDPTERM() in case of failure.
* contrib/hblzf/tests/test2.prg
+ replaced in-source hbmk2 cmdline instructions with
#require directive
* contrib/gtwvg/tests/wvgactivex.prg
! fixed reference to non-existing batch file
* doc/en/file.txt
* doc/gmake.txt
* doc/windll.txt
* examples/gtwvw/docs/gtwvw.txt
* examples/gtwvw/docs/whatsnew.txt
* examples/gtwvw/tests/drawimg.prg
* examples/guestbk/guestbk.txt
* examples/hscript/readme.txt
- deleted references to non-existing/obsolete batch files
* include/harbour.hbx
* src/rtl/Makefile
+ src/rtl/cdpdet.prg
+ src/rtl/cdpdetc.c
+ moved terminal and OS detection logic from hbmk2 to RTL
with changes necessary to fit into core. New functions are:
hb_cdpOS() -> <cCP | NIL>
hb_cdpTerm() -> <cCP | NIL>
it means these functions can be used in any apps now.
Example:
hb_SetTermCP( hb_cdpTerm() )
Set( _SET_OSCODEPAGE, hb_cdpOS() )
* contrib/hbrun/hbrun.prg
* utils/hbmk2/Makefile
- utils/hbmk2/hbmk2c.c
* utils/hbmk2/hbmk2.hbp
* utils/hbmk2/hbmk2.prg
* using hb_cdpOS() and hb_cdpTerm() to automatically
configure OS and terminal CP
* include/Makefile
* include/common.ch
+ include/hbhash.ch
+ moved hash-related constants from common.ch to new hbhash.ch
; common.ch include hbhash.ch for compatibility, protected by
HB_LEGACY_LEVEL4 so it will stop being included automatically.
If you use HB_HAUTOADD_* or HB_HMERGE_* constants, make sure
to #include "hbhash.ch".
* contrib/hbrun/headers.prg
* utils/hbmk2/hbmk2.prg
+ include hbhash.ch
* src/rdd/hbsix/sxini.prg
* use hbhash.ch instead of common.ch
* include/hbapi.h
- deleted unused old HB_BASEARRAY_PTR type. INCOMPATIBLE.
Same as PHB_BASEARRAY, use this instead.
* include/hbtypes.h
* src/vm/maindllp/dllext.c
+ added hb_macroTextValue() to hbmaindllp.
; TODO: rename all HB_ITEM_PTR to PHB_ITEM and delete former.
pls shout if there is any difference in usage between
these two.
* src/pp/ppcore.c
+ accept and ignore '#require' PP directive
* contrib/hbrun/hbrun.prg
* utils/hbmk2/hbmk2.prg
+ use '#require "name"' to request modules.
(This replaces former '//#require', '//#pragma module' and
'*#pragma module' directives.)
* utils/hbmk2/hbmk2.prg
+ add .hbc automatically for '#require' directives (experimental)
; TODO: to work also in non-incremental mode and to work
reliably in -inc mode. To not impact performance, the
compiler could do a callback when #require is found,
hbmk2 could find the .hbc and extend compiler options
dynamically.
! fixed finding .hbc files that were detected automatically
* contrib/hbrun/hbrun.1
+ added my name as author
* contrib/hbrun/hbrun.prg
* contrib/hbrun/pullext.prg
+ changed to use '//#require' syntax.
! filter empty dynamic modules
+ display error when trying to load dynamic modules in static
hbrun build
* include/hbver.ch
* src/rtl/version.c
+ added hb_Version( HB_VERSION_SHARED ) to query whether
the app was built in shared mode or static.
* src/debug/debugger.prg
* src/rdd/dbstrux.prg
* src/rdd/dbtotal.prg
* src/rdd/dbupdat.prg
* src/rdd/rddord.prg
* src/rdd/usrrdd/rdds/logrdd.prg
* src/rtl/adir.prg
* src/rtl/alert.prg
* src/rtl/checkbox.prg
* src/rtl/dbedit.prg
* src/rtl/getsys.prg
* src/rtl/getsys53.prg
* src/rtl/hbi18n2.prg
* src/rtl/hbini.prg
* src/rtl/listbox.prg
* src/rtl/menuto.prg
* src/rtl/objfunc.prg
* src/rtl/pushbtn.prg
* src/rtl/radiobtn.prg
* src/rtl/radiogrp.prg
* src/rtl/scrollbr.prg
* src/rtl/tbrowse.prg
* src/rtl/tget.prg
* src/rtl/tgetlist.prg
* src/rtl/tlabel.prg
* src/rtl/tmenuitm.prg
* src/rtl/tmenusys.prg
* src/rtl/tpopup.prg
* src/rtl/treport.prg
* src/rtl/ttopbar.prg
* converted 'DEFAULT ... TO ...' to
a) __defaultNIL() where bug compatibility is desired
b) IF ... == NIL ... where default value had execution cost
c) hb_default() where above side-effect were not in play
% deleted '#include "common.ch"'
* minor formatting
; "common.ch" is now eliminated from all core core and all
contribs except gtwvg/hbide/hbxbp/xhb/examples where it's
used excessively and very few more places. Please
don't readd it, but use HB_DEFAULT() and HB_IS*() functions
instead. Only use "common.ch" if you need HB_HAUTOADD_* or
HB_HMERGE_* constants.
* src/rtl/radiogrp.prg
% deleted '#include "common.ch"'
* contrib/hbtip/sessid.prg
! playing it safe: using simple IF/ENDIF instead of HB_DEFAULT()
on a multi-type variable
; NOTE: One more implication of HB_DEFAULT vs. DEFAULT ... TO ...
is that the default expression is always evaluated with
the former.
* include/harbour.hbx
* src/rtl/Makefile
+ src/rtl/hbdef.c
+ added HB_DEFAULT( @<xVariable>, <xDefaultValue> )
it can replace DEFAULT ... TO ... command. In addition
it will set the value to the default one even when the
variable has any other type, not only NIL.
* src/3rd/zlib/Makefile
* src/3rd/zlib/*
* updated to 1.2.7 from 1.2.6. Using 3rdpatch.hb and .dif reapplied manually.
* src/rtl/hbzlibgz.c
+ HB_GZOPEN() updated to use new in zlib 1.2.7 wide gzopen()
on Windows. It means now it supports unicode filenames.
* contrib/hbmzip/3rd/minizip/zip.c
* updated manually from zlib 1.2.7 (two warnings fixed)
* harbour/src/vm/arrays.c
* allow to call hb_arrayId() with NULL parameter
* harbour/include/hbapiitm.h
* harbour/src/vm/itemapi.c
+ added internal HVM function hb_itemPutPtrRawGC() - it's
necessary for blocks allocated with hb_gcAllocRaw().
* harbour/src/vm/thread.c
! use hb_itemPutPtrRawGC() instead of hb_itemPutPtrGC()
It fixes internal errors (i.e. reported in build process)
after my previous commit.
* harbour/include/hbgtinfo.ch
+ added new HB_GTI_FONTATTRIBUTE value: HB_GTI_FONTA_CTRLCHARS
When used it forces using CTRL characters (from chr(0) to chr(31))
directly without translation to CP437 unicode values.
In windows some fonts need it to correctly display them.
* harbour/src/rtl/gtwvt/gtwvt.h
* harbour/src/rtl/gtwvt/gtwvt.c
+ added support for
hb_gtInfo( HB_GTI_FONTATTRIBUTE, HB_GTI_FONTA_CTRLCHARS )
* harbour/src/rtl/hbgtcore.c
! do not replace CTRL chars with '?' in HB_GTSELF_GETUC() method
* harbour/src/rtl/cdpapi.c
% use HB_CDP_ISUTF8( cdp ) macro instead of cdp == &s_utf8_codepage
It should eliminate unnecessary translations and speedup some
operations.
* harbour/include/hbstack.h
* harbour/src/vm/codebloc.c
* harbour/src/vm/estack.c
* harbour/src/vm/hvm.c
% split hb_stackLocalVariable( int * ) macro and function into
two different actions:
hb_stackLocalVariable( int ) // access/assign local var
hb_stackLocalVariableAt( int * ) // create reference to local var
* harbour/src/vm/classes.c
% small optimization in __CLSGetProperties() function
* harbour/src/rtl/tpersist.prg
! force linkin ARRAY() function when HBPersistent class is used
! use hb_ValToExp() instead of local ValToText() function which
didn't work correctly for memos and some other types
! fixed infinite loop in HBPersistent:LoadFromText() used with
empty text. The problem was also exploited by :LoadFromFile()
method if file does not exists or it's empty()
! fixed deserialization of strings containing "=" char inside
! fixed deserialization of strings containing "::"
! added support for decoding nested objects - the code was not
finished and this part was not implemented at all
+ added support for serialized text using ":=" assign operator
instead of "="
* use ":=" instead of "=" as assign operator in new serialized
text
+ ignore lines starting with "//" and added support for files
which do not start with OBJECT directive
* ignore pointer and codeblock items stored in instance variables
during serialization
; now HBPersistent class in Harbour should read serialized files
created in xHarbour with few exceptions:
- Harbour does not allow to serialize codeblocks
See "CODEBLOCK SERIALIZATION / DESERIALIZATION" in
doc/xhb-diff.txt for the reasons.
If Harbour application restores xHarbour HBPersistent files
with serialized codeblocks then RTE
Undefined function: HB_RESTOREBLOCK
is generated.
- xHarbour serialize pointer items as numeric items
saved in hexadecimal notation. Then restore them as numbers.
- HBPersistent does not support hash arrays - they are
serialized by default serialization code used by both
compilers to generate expressions.
In Harbour it's done by hb_valToExp() function and
in xHarbour it's ValToPrg() is used.
ValToPrg() does not create valid macrocompiler expressions
for arrays and objects so HBPersistent files created
by xHarbour are broken and cannot be correctly deserialized.
It happens if objects has hash arrays in instance variables
and these hash arrays contain normal arrays or object
If Harbour application restores such xHarbour HBPersistent
file then RTE "Syntax error: &" is generated.
; xHarbour encapsulates deserialization code inside TRY/CATCH/END
statement saving errors to trace.log file. Harbour generates RTE.
If programmer needs similar behavior then he should
call :LoadFromText() and :LoadFromFile() methods inside
BEGIN SEQUENCE [ / RECOVER ... ] / END SEQUENCE
statement.
; xHarbour reinitialize all or properties instance variables to
default state inside :LoadFromText() and :LoadFromFile() methods.
The 3-rd parameter in above method <lPropertiesOnly> allows to
chose which install variables should be reinitialized:
all (default) or properties only.
Harbour does not have such functionality.
If programmer needs it then he should reinitialize them himself.
It can be easy done be simple function.
Alternatively we can implement this functionality but in such
case I'd suggest to define three actions for such switch:
none (default), all, properties only.
; Harbour does not support 2-nd parameter <lIgnoreBadIVars> which
exist in xHarbour versions of HBPersistent:LoadFromText() and
HBPersistent:LoadFromFile():
If necessary we can implement it though it's usable only if we
want to ignore some wrong lines and process others.
; Warning: Neither Harbour nor xHarbour supports arrays and objects
with cyclic references in HBPersistent code - infinite
loop appears in such case.
; I've never used HBPersistent and the state of the previous
HBPersistent code in Harbour suggests that no one used it for
some serious jobs so I'm open for any opinions and suggestions
about it from xHarbour users.
* harbour/src/compiler/harbour.y
* generate line numbers before extended codeblocks - it gives
more debugger friendly code.
* harbour/src/compiler/harbour.yyh
* harbour/src/compiler/harbour.yyc
* regenerated (with bison 2.4.1)
* harbour/src/common/expropt2.c
* small code simplification
* harbour/src/rtl/hbregex.c
% eliminated call to pcre_config() on each regex compilation
* harbour/utils/hbmk2/hbmk2.prg
! fixed -head=full for code using non UTF8 characters.
It was broken because in UNIX builds UTF8 is always enabled
as HVM CP in HBMK2 and this setting was inherited by PCRE
with PCRE_UTF8 flag. If this flag is used PCRE validates
all strings and refuse to make any operations if they are
not valid UTF8 string so it was not working at all for
source code using different encoding.
! fixed regex used in -head=full to respect shortcuts in #include
directive, i.e.:
#incl "file.ch"
! fixed regex used in -head=full to recognize #include directives
separated by ';', i.e.:
#include "file1.ch" ; #include "file2.ch" ; #include "file3.ch"
; TOFIX: UTF8 mode on output should not be enabled unconditionally in
all UNIX builds. There are many *nix installations where UTF8
is not system CP, i.e. older Linux distributions. In fact
it's user attribute so each user can use different encoding
local to his connection and/or terminal settings. Probably
it's good idea to use code like:
lUTF8 := "UTF-8" $ GetEnv( "LANG" ) .OR. ;
"UTF-8" $ GetEnv( "LC_CTYPE" )
* utils/hbmk2/hbmk2.prg
! fixed RTE when logging sign command
! fixed `` operator to strip CRs and convert LFs to space
in stdout output.
* src/rdd/dbcmd.c
! clarified DBCREATE() 4th parameter in comment
