* harbour/source/compiler/genc.c
* allow to use #pragma begindump / enddump in .prg files without
any .prg code
* harbour/source/vm/hvm.c
! fixed assign in overloaded []
* harbour/include/hbapicls.h
* changed extern hb_bProfiler to local HVM static variable
* harbour/source/vm/hvm.c
+ added two macros HB_NO_TRACE and HB_NO_DEBUG for these people
who needs/wants to test HVM speed without debugger support
and PRG call tracing - the difference is rather small
* harbour/contrib/hbzlib/include/stdafx.h
* harbour/contrib/hbzlib/include/ziparchive.h
* cleaned some warnings
* harbour/include/hbexprb.c
* harbour/include/hbexprop.h
* harbour/source/common/expropt2.c
* restored Mindaugas code for UPPER optimization and added logic to
optimize UPPER for strings with only Latin letters, digit and spaces
just like in Clipper. Now code like:
? Upper( chr(65)+chr(66)+chr(67)+"0123 QWE asd " ) + "ZXC"
is fully optimized at compile time to single string expression.
% optimized hb_compExprReduceCHR() by eliminating memory allocation
for new string
! fixed typo in my previous commit
* harbour/include/hbexprop.h
* harbour/include/hbexprb.c
* harbour/source/common/expropt2.c
* added UPPER("") optimisation. The side effect of this modification
is that Harbour becomes Clipper bug compatible in calculation of
AT(UPPER(""), "test"). Though this optimisation does not introduce
any buggy behaviour itself.
* harbour/harbour.spec
* set attributes explicitly to clean some wrong attributes committed
to CVS and minor cleanup
* harbour/include/hbapierr.h
* harbour/source/rtl/errorapi.c
+ added hb_errorBlock()
* harbour/source/macro/macro.yyc
* harbour/source/compiler/harbour.yyc
* manual hack to pacify BCC warning messages
* harbour/include/hbpcode.h
* harbour/include/hbxvm.h
* harbour/source/compiler/genc.c
* harbour/source/compiler/gencc.c
* harbour/source/compiler/harbour.y
* harbour/source/compiler/harbour.yyc
* harbour/source/compiler/hbdead.c
* harbour/source/compiler/hbfix.c
* harbour/source/compiler/hblbl.c
* harbour/source/compiler/hbmain.c
* harbour/source/compiler/hbopt.c
* harbour/source/compiler/hbpcode.c
* harbour/source/compiler/hbstripl.c
* harbour/source/vm/hvm.c
+ added support for:
BEGIN SEQUENCE WITH <errorBlockExp>
[<statements;...>]
[ RECOVER [ USING oErr ] ]
...
[ ALWAYS ]
...
END
It works like normal BEGIN SEQUENCE but sets <errorBlockExp> as
ErrorBlock() before executing <statements;...> and restore
previous ErrorBlock() at the end or when exception will appear.
I hope it will make some xHarbour user happy ;-)
* harbour/include/hbcompat.ch
+ added preprocessor directive for TRY / CATCH / FINALLY / END
working like in xHarbour - translated to:
BEGIN SEQUENCE WITH { |oErr| Break( oErr ) }
...
[ RECOVER [ USING oErr ] ]
...
[ ALWAYS ]
...
END
* harbour/source/compiler/harbour.yyc
* synced with recent modifications in harbour.y - I missed it.
* harbour/source/hbpcre/pcrecomp.c
* cleaned some waning reported by Lorenzo
* harbour/common.mak
* harbour/makefile.bc
* harbour/makefile.vc
* harbour/bin/hb-func.sh
* harbour/config/bsd/gcc.cf
* harbour/config/darwin/gcc.cf
* harbour/config/hpux/gcc.cf
* harbour/config/linux/gcc.cf
* harbour/config/sunos/gcc.cf
* harbour/source/Makefile
+ harbour/source/hbpcre/ChangeLog
+ harbour/source/hbpcre/Makefile
+ harbour/source/hbpcre/chartabs.c
+ harbour/source/hbpcre/config.h
+ harbour/source/hbpcre/dftables.c
+ harbour/source/hbpcre/pcre.h
+ harbour/source/hbpcre/pcrecomp.c
+ harbour/source/hbpcre/pcreconf.c
+ harbour/source/hbpcre/pcredfa.c
+ harbour/source/hbpcre/pcreexec.c
+ harbour/source/hbpcre/pcrefind.c
+ harbour/source/hbpcre/pcrefinf.c
+ harbour/source/hbpcre/pcreget.c
+ harbour/source/hbpcre/pcreglob.c
+ harbour/source/hbpcre/pcreinal.h
+ harbour/source/hbpcre/pcreinfo.c
+ harbour/source/hbpcre/pcremktb.c
+ harbour/source/hbpcre/pcreoutf.c
+ harbour/source/hbpcre/pcreprni.c
+ harbour/source/hbpcre/pcrerefc.c
+ harbour/source/hbpcre/pcrestud.c
+ harbour/source/hbpcre/pcretabs.c
+ harbour/source/hbpcre/pcretryf.c
+ harbour/source/hbpcre/pcrever.c
+ harbour/source/hbpcre/pcrevutf.c
+ harbour/source/hbpcre/pcrexcls.c
+ harbour/source/hbpcre/ucp.h
+ harbour/source/hbpcre/ucpinter.h
+ harbour/source/hbpcre/ucptable.c
+ added HBPCRE library - based on older xHarbour 6.3 version
filenames changed to 8.3 DOS format
* harbour/include/hbregex.h
* harbour/source/rtl/Makefile
* harbour/source/rtl/hbregex.c
+ harbour/source/rtl/hbregexc.c
* divided harbour regular expression functions into two files
Now regular expression low level library is not linked with
application until user will not use or REQUEST for one of HB_REGEX*
functions. It also means that also DBOI_SKIPREGEX will not work when
regex module is not linked.
+ added support for build-in regular expression library
+ added ulLen parameter to hb_regexMatch() to support strings with
embedded 0
* harbour/contrib/bmdbfcdx/bmdbfcdx1.c
* harbour/source/rdd/dbfcdx/dbfcdx1.c
* harbour/source/rdd/dbfntx/dbfntx1.c
* harbour/source/rtl/strmatch.c
* use new hb_regexMatch() format
* harbour/utils/hbrun/Makefile
+ added hbpcre to linked library list
* harbour/include/hbcompdf.h
* harbour/include/hbexprb.c
* harbour/source/common/expropt1.c
* harbour/source/compiler/harbour.y
* harbour/source/compiler/harbour.yyc
* harbour/source/macro/macro.y
* harbour/source/macro/macro.yyc
! fixed bug in operator precedence - seems that it existed from
beginning and partially were fixed in xHarbour. This code exploits
the problem:
#define EOL chr(13)+chr(10)
#command ? [<x,...>] => [outstd( <x> );]outstd(EOL)
#command TEST <exp> => BEGIN SEQUENCE ;
; ? <exp> ;
; RECOVER USING oErr ;
; ? "Error:", oErr:subCode, ;
oErr:description, oErr:operation ;
; END
proc main()
local oErr, s1:="X", s2:="Y", t:=.T.
errorBlock({|oErr|break(oErr)})
TEST t != s1 $ s2
TEST t == s1 $ s2
TEST t = s1 $ s2
TEST t < s1 $ s2
TEST t <= s1 $ s2
TEST t > s1 $ s2
TEST t >= s1 $ s2
TEST s1 $ s2 != t
TEST s1 $ s2 == t
TEST s1 $ s2 = t
TEST s1 $ s2 > t
TEST s1 $ s2 < t
TEST s1 $ s2 >= t
TEST s1 $ s2 <= t
return
* harbour/makefile.vc
* synced with BETA branch
* harbour/contrib/Makefile
* disable BMDBFCDX for DOS builds due to long names which do not
pass 8.3 convention
* use OLE2 instead of OLE
* harbour/contrib/ole2/w32ole.c
* warning fixes
* harbour/include/hbregex.h
+ added REGEX support for DJGPP build
* harbour/source/rtl/net.c
* warning cleanup
* harbour/make_xmingw.sh
* harbour/source/pp/Makefile
* recent xHarbour fixes for MinGW Linux->W32 cross build by Phil Krylow
with small modifications
* harbour/config/rules.cf
+ added rules for C++ files
* harbour/include/hbapi.h
* harbour/source/vm/arrays.c
+ added hb_arraySetForward()
* harbour/include/hbapi.h
* harbour/source/rtl/strmatch.c
+ added hb_strMatchFile() - compare two strings using platform dependent
rules for file matching, respects platform dependent wildcards
+ added hb_strMatchCaseWildExact() - compare two strings using pattern
with wildcard (?*) ignoring the case of the characters - patern have
to cover whole string
* changed hb_strMatchRegExp() to use real reguar expresions, if build
does not support regex then redirected to hb_strMatchWildExact()
- removed not longer used hb_strMatchDOS() - in fact it was neither DOS
nor classic wildcards (?*) compatible
* harbour/source/common/hbstr.c
! casting to avoid possible problem on platforms where toupper does not
accept negative char value
* harbour/include/hbcomp.h
* harbour/source/compiler/hbcomp.c
* harbour/source/compiler/hbmain.c
* minor modifications in cleanup functions
* harbour/source/rtl/hbffind.c
* use hb_strMatchFile()
* harbour/source/vm/memvars.c
* use hb_strMatchCaseWildExact() instead of hb_strMatchRegExp()
I still keep the hack which translates any mask starting with '*'
to "*" but if you think that we should fully respect wildcards
patterns then we should remove it from function hb_memvarGetMask()
Opinions?
* harbour/source/pp/ppcore.c
! fixed bug reported by Lorenzo
* harbour/contrib/rdd_ads/ads1.c
* changed static variable name from iSetListenerHandle to
s_iSetListenerHandle to keep common naming convention
* harbour/include/hbapi.h
* harbour/source/vm/hashes.c
* harbour/source/vm/hashfunc.c
* harbour/source/vm/hvm.c
+ added new 3-rd parameter to hb_hashGetItemPtr() to control
independently hash autoadd behavior in access/assign operation
* harbour/include/hbrdddbf.h
* harbour/include/hbrddntx.h
* harbour/source/rdd/dbfntx/dbfntx1.c
* moved static variables holding RDDI_* settings to RDDNODE data so
they can be set separately for each RDD which inherits from DBFNTX
* harbour/source/vm/classes.c
+ added disabled code to access hash items using messages
* harbour/source/vm/garbage.c
* comment
* harbour/include/hbapi.h
* harbour/include/hbexpra.c
* harbour/source/macro/macro.y
* harbour/source/macro/macro.yyc
* harbour/source/macro/macrolex.c
* harbour/source/vm/macro.c
- removed hb_macroIsIdent()
+ added hb_macroTextSymbol()
! fixed old bug (it exists in all [x]Harbour versions) in macrocompiler
which caused HVM stack corruption
* generate from lexer symbols which are well formatted
(uppercase and size <= HB_SYMBOL_NAME_LEN)
% use hb_dynsymGetCase() instead of hb_dynsymGet()
% some minor speed improvement
* harbour/source/rtl/gtwin/gtwin.c
% use hb_dynsymFind() instead of hb_dynsymFindName()
* harbour/source/rdd/dbcmd.c
* harbour/source/rdd/wafunc.c
* reenabled by default strict clipper compatible behavior for
code select(), dbSelectArea() and alias allocating
* harbour/source/rdd/dbf1.c
+ added DBI_ISENCRYPTED for tables encrypted by SIX3
* harbour/source/rtl/hbinet.c
! added missing break (Ron fix from xHarbour)
* harbour/source/compiler/hbfix.c
! fixed local IND/DED when PARAMETERS is used
! fixed local names generated for debugger when PARAMETERS is used
* harbour/source/debug/dbgentry.c
! fixed very bad typo in my previous commit
% minor optimization
* harbour/bin/hb-func.sh
+ added -lpcreposix to system libraries when C_USR has -DHB_PCRE_REGEX
* harbour/include/hbregex.h
* removed pcre/ path from PCRE header files - if you want to build
Harbour with PCRE regex and PCRE header files are not in default\
path then set valid path in C_USR, f.e.:
export C_USR="${C_USR} -DHB_PCRE_REGEX -I/usr/include/pcre"
* harbour/include/hbxvm.h
* harbour/source/compiler/gencc.c
* harbour/source/vm/hvm.c
* minor optimizatin for -gc3 output
* harbour/include/hbclass.ch
* formatting
* harbour/include/hbpp.h
* harbour/source/pp/ppcore.c
+ added support for:
#pragma escapedstrings = on
cStr := "some value with internal \" character"
#pragma escapedstrings = off
+ added support for multiline strings in escaped strings
* harbour/source/compiler/gencobj.c
* minor simplification
* harbour/source/rtl/hbffind.c
* always set hb_fsError()
* harbour/source/rtl/mlcfunc.c
! fixed minor typo
* harbour/TODO
* removed some not longer necessary TODO/TOFIX
* harbour/common.mak
* harbour/source/rtl/Makefile
- harbour/source/rtl/memoline.c
- harbour/source/rtl/mlcount.c
- harbour/source/rtl/mlctopos.c
- harbour/source/rtl/mlpos.c
- harbour/source/rtl/mpostolc.c
+ harbour/source/rtl/mlcfunc.c
* new version of memo line functions:
MEMOLINE(), MLCOUNT(), MLPOS(), MLCTOPOS(), MPOSTOLC()
* harbour/include/hbexpra.c
! fixed in HB_COMPAT_XHB build hb_enumIndex() used inside internal
FOR/NEXT loop
* harbour/include/hbpp.h
* harbour/source/pp/ppcore.c
! removed some not longer necessary hack
% added some speed optimizations - it's not done in the way I want to
have in the final version but with some very small modifications it
should make people using a lot of #[x]translate (Enrico ;-)) happy.
* harbour/source/rtl/set.c
* minor cleanup in HB_SET_EOL initialization
* harbour/source/vm/hvm.c
* clean MEMO flag in array and hash items created by { MEMOFILED, ... }
or { "KEY" => MEMOFILED, ... } and optimized results of PLUS operation
* harbour/utils/hbtest/rt_misc.prg
* SET EOL to CHR(13) + CHR( 10 ) before MLCTOPOS() tests to avoid false
alarm on platforms with differ default EOL
* harbour/source/compiler/hbmain.c
* harbour/source/compiler/harbour.yyc
! Fixed error message text when unclosed multiline codeblock
was detected (said: <||...> instead of {||...}).
(two more places)
* harbour/source/compiler/harbour.y
! Fixed error message text when unclosed multiline codeblock
was detected (said: <||...> instead of {||...}).
; TOFIX: In this same case two memory blocks are not released:
source\main\harbour.c:329: HB_TR_ERROR Block 1 0098C39C (size 100)
source\main\harbour.c:329: HB_TR_ERROR Block 2 0098C200 (size 92)
* harbour/include/hbcomp.h
* harbour/include/hbexpra.c
* harbour/include/hbexprb.c
* harbour/include/hbexprop.h
* harbour/include/hbpp.h
* harbour/source/compiler/complex.c
* harbour/source/compiler/harbour.y
* harbour/source/compiler/harbour.yyc
* harbour/source/compiler/harbour.yyh
* harbour/source/compiler/hbmain.c
* harbour/source/pp/ppcore.c
+ added support for extended multiline codeblocks:
{ |<params,...>| <EOL>
<statement1>
...
<statementN>
return <val>
}
Such codeblocks allow nested definitions. They can also have their
own variable declarations (local, static, field, memvar, parameters,
private, public) and visibility of declared variables is similar to
nested function in Pascal but please note that there is one limitation
which exists also for normal codeblocks (Clipper compatible behavior):
internal codeblock local parameters and local variables cannot be
used in nested codeblocks, f.e. such code:
eval( { |p| eval( {|| qout( p + 1 ) } ) } )
is illegal in Clipper and Harbour. This limitation also can be
eliminated but but it will be necessary to make deeper modifications
in code for local detaching changing the method of generating
codeblocks local variables in nested blocks so unlike the above
also some modifications in HVM will be necessary.
Please test.
* harbour/include/hbcomp.h
* harbour/include/hbcompdf.h
* harbour/include/hbexpra.c
* harbour/include/hbexprb.c
* harbour/source/compiler/complex.c
* harbour/source/compiler/genc.c
* harbour/source/compiler/gencli.c
* harbour/source/compiler/genobj32.c
* harbour/source/compiler/harbour.y
* harbour/source/compiler/harbour.yyc
* harbour/source/compiler/harbour.yyh
* harbour/source/compiler/hbmain.c
* changed code used for variable scoping - now it's much more
simple (only one function) and allows to use declarations in
nested functions/codeblocks with visibility similar to Pascal
! fixed parsing FILED <fields,...> IN <alias>
we were accepting code like:
FIELD f1, f2 IN db1, f3 IN db2 IN db3
and for all fields the last alias (db2 in this example) was used
! fixed calculating number of static variables with -b (debugger)
compilation (number of file wide statics were doubled)
! fixed generation of static variable names for debugger when declared
with array dimensions, f.e.:
static sVar[3]
* harbour/source/rtl/hbffind.c
* minor cleanup *nix version
* harbour/contrib/bmdbfcdx/bmdbfcdx1.c
! fixed casting for C++ compilation
* harbour/source/rdd/dbfdbt/dbfdbt1.c
* updated for some old API modifications (this library is not used now
but if we keep it in CVS then I think it should be updated)
* harbour/source/rtl/filehb.c
* Changed one old TOFIX to a NOTE regarding a minor
incompatibility in FILE() where the filename is RTrim()-ed
in Clipper but not in Harbour. It became a NOTE because
making it compatible would mean losing some portability.
* harbour/source/rtl/hbffind.c
! Removed code in UNIX branch which would change an empty
file mask to "*" (thus causing FILE("") to return .T.
and DIRECTORY("") to return the whole dir content.)
! Made sure in UNIX branch that an empty file mask won't
return any files regardless of the filename matching
method used.
* harbour/source/rtl/strmatch.c
+ Added NOTE about hb_WildMatch() / sx_WildMatch()
incompatibility.
! Guarded WILDMATCH() with HB_COMPAT_XHB.
* WILDMATCH() now calls HB_WILDMATCH() internally.
* harbour/source/rdd/dbcmd.c
* initialize lpaParent in DBRELINFO structure to pArea - please
note that this initialization is ignored by default SETREL()
RDD method
* harbour/source/rdd/usrrdd/usrrdd.c
! fixed GPF when NULL lpaParent or lpaChild was used in DBRELINFO
structure
* harbour/source/rdd/wacore.c
* harbour/source/vm/classes.c
* harbour/include/hbapirdd.h
* harbour/source/compiler/hbmain.c
* updated HB_TRACE messages and some minor cleanups
* harbour/source/rdd/dbcmd.c
! fixed initialisation of DBRELINFO dbSetRelation() - thanks to Antonio
for locating the problem
* harbour/source/rtl/idle.c
* increase timeout in nanosleep() from 1 microsecond to 1 millisecond
- removed some wrong comment
* harbour/source/vm/classes.c
* changed __cls_IncData() to return index to newly allocated instance
item and __cls_DecData() to return number of local (not inherited)
instance items
