* harbour/include/hbclass.ch
+ added validation for class data names. As additional preprocessor
rule to not use <!marker!>
* harbour/include/hboo.ch
* harbour/source/rtl/tclass.prg
* harbour/source/vm/classes.c
+ added delegate messages
* harbour/include/hbstack.h
* harbour/source/vm/arrays.c
* harbour/source/vm/estack.c
* harbour/source/vm/hvm.c
* harbour/source/vm/proc.c
+ added startup symbol to hb_stack. It gives very good stop
condition for all procedures which trace stack calls and
resolves the problem with hb_stack.pBase which was never
pointing to valid function/procedure symbol when no symbol
was put on HVM stack. Now after hb_stackInit() the first
item is allocated for "hb_stackInit()" symbol so hb_stack.pBase
is always a pointer to valid function/procedure symbol.
* changed the guard condition for buggy code in hb_stackPop()
and similar code from:
hb_stack.pPos < hb_stack.pItems
to:
hb_stack.pPos <= hb_stack.pBase
The old condition was generating usable error message only in the
startup function. In deeply called functions it was only waste of
CPU time on one of the most often call functions. Before it was
activated internal stack structures were corrupted.
If we were leaving for many years without really working stack
underflow protection then maybe we should think about disabling
it at when HB_STACK_MACROS are used and keep it only in real
stack functions for HVM developers. Single protection in
hb_stackOldFrame() I've just added is enough for basic C code
validation.
I created the second version of stack macros without it in
hbstack.h but I would like to hear other developers opinion.
* harbour/include/hbexprc.c
* Temporary disabled optimization with references to object variables
until we will not have extended reference items in our HVM.
It can be enabled by compiling with HB_USE_OBJMSG_REF
* harbour/include/hbclass.h
* restored previous CLASSDATA behavior which ignores SHARED clause
and always create shared class variables
* harbour/include/hbclass.ch
* most of the rules rewritten
! fixed some wrong rules and general cleanup
+ added additional code validation
! fixed compilation of more then one class in single file.
Now it's even possible to declare all classes at beginning of
single file and then implementing their methods in any order
! fixed using static classes and classes
! fixed compilation without <ClassName>_ prefix in method names
+ added support for HB_CLS_NO_DECORATION macro which disable
adding <ClassName>_ prefix to method names - this macro is
set by default when HB_SHORTNAMES is set.
+ added support for declared parameters validation - it can be
disabled with HB_CLS_NO_PARAMS_ERR and I had to disable it
by default due to problems with our preprocessor.
Ryszard seems that our PP has serious problems with decoding
directives when there is no space between symbol and some other
non symbol character. I had to add some workarounds and even
introduce buggy rules to make it working. Please look at it.
You can remove #define HB_CLS_NO_PARAMS_ERR from hbclass.ch
and try to rebuild Harbour core code to see the problem.
* harbour/include/hboo.ch
* harbour/source/vm/classes.c
+ added support for new primitive message: HB_OO_MSG_PERFORM
* harbour/source/rtl/tclass.prg
- removed <lPersistent> parameter from HBClass messages and
internals data. Persistent is supported as scope bit and
separate variable was redundant.
- removed stripping of () from message names. Here is not a place
to fix wrong preprocessor rules.
* harbour/utils/hbtest/rt_class.prg
* use: METHOD PROCEDURE ... CALSS ...
instead of: PROCEDURE ... CALSS ...
The first version is preferable syntax.
* harbour/source/debug/dbgtmenu.prg
* harbour/source/rtl/checkbox.prg
! fixed some parameters in method declaration - global cleanup
will have to wait for preprocessor fixes
Hi all,
Please make test with current hbclass.ch code.
I hope that I haven't broken too much things ;-) but I rewrite
from scratch most rules and it's possible that I missed sth or
made some stupid typos. Current version is much shorter and should
be easier to updated. For sure I've intentionally changed one thing.
CLASSDATA was ignoring SHARED attribute and always created shared
class variables. Seems that it was long existing typo but the fix
may interact with already existing code which needs SHARED class
variables but does not use SHARED clause in CLASSDATA declaration.
In such case please update it and add missing SHARED.
Also in the end of CLASS declaration we have:
[ ; #translate Super( <SuperClassN> ): => ::<SuperClassN>: ] ;
[ ; #translate Super( <SuperClass1> ): => ::<SuperClass1>: ] ;
[ ; #translate Super(): => ::<SuperClass1>: ] ;
[ ; #translate Super: => ::<SuperClass1>: ] ;
[ ; #translate ::Super : => ::<SuperClass1>: ]
These rules introduce very serious bug - they are breaking supercasting
in code which makes sth like:
::super:super:super:msg
or in any other code which sends SUPER message to some other class
objects. I will have to remove them. At least the last three ones.
There were some other things I wanted to write about but it's too late
and I'm to tired - sorry. If you will have any question please ask. if
you will notice some problems with current rules please inform me.
* harbour/include/hbclass.ch
* updated some not enabled by default rules to use function pointers
instead of function names which I forgot to update in previous
commit
* harbour/include/common.ch
+ added ISSYMBOL()
* harbour/include/hbextern.ch
- removed __CLS_PARAM()
* harbour/include/hbclass.ch
* harbour/source/rtl/tclass.prg
* harbour/source/vm/classes.c
* replaced __CLS_PAR00() and __CLS_PARAM() functions by preprocessor
rules. These functions are not longer necessary though I left them
in classes.c for backward binary compatibility. Probably they will
be removed in the future.
! use function pointers (@<funcname>()) instead of function names
with REQUEST for super classes ID. It fixes using STATIC class
functions.
* harbour/make_rpm.sh
* check for rpmbuild also when BUGGY_RPM is set
* harbour/make_tgz.sh
* removed rebuild hbpptest with shared libraries
* harbour/include/hbapicls.h
+ added definitions for hb_clsCreate(), hb_clsAdd(), hb_clsAssociate()
* harbour/include/hbapierr.h
* harbour/source/rtl/errorapi.c
+ added hb_errFuncName - pseudo function name which can be used as
operation description in new error object. It should help to simplify
some code and fix function names when RT error is generated from C
function which can be executed from different .prg functions.
* harbour/include/hbclass.ch
+ added some missing Classy(y) compatible object variable type
descriptions: Int, Integer, Block
* harbour/source/rdd/dbfcdx/dbfcdx1.c
! disabled early retrieving from the cache key position and key count
* harbour/include/hbvm.h
* harbour/source/vm/hvm.c
+ added: void hb_vmPushEvalSym( void ) for the code which do not
want to access any [P]HB_SYMB structures
* changed symbol name in hb_symEval from __EVAL to EVAL and modified
hb_symEval registration method to not register hb_vmDoBlock as global
function. Having PHB_DYNS address for "EVAL" in hb_symEval.pDynSym
allows to eliminate some strcmp() when we are looking for EVAL public
symbol.
! fixed symbol table updating in PCODE functions for future multi module
.hrb files
* harbour/source/vm/proc.c
* use hb_symEval.pDynSym instead of strcmp()
* harbour/source/vm/arrays.c
* harbour/source/vm/debug.c
* minor code cleanup
* harbour/source/rtl/tclass.prg
* harbour/source/vm/classes.c
% separate message hash table from methods table - it will cause farther
reduction of allocated memory - many thanks to all people who send
me statistics about number of used classes and symbols from their code.
% use hb_symEval.pDynSym
+ added support for object variable type scope checking like in Class(y).
! fixed typo in pushing object class variables by reference
% s_pClasses is now indexed directly by class handle
% some other optimiztions
! make hidden and non virtual invisible in child classes if it overloaded
some other non hidden method. In such case the overloaded method is used
and can be accessed.
! make friend functions real method class oriented not (super)object for
hidden messages
* harbour/include/hbclass.ch
* harbour/include/hbvm.h
* harbour/source/rtl/tclass.prg
* harbour/source/vm/classes.c
* harbour/source/vm/hvm.c
! fixed bug in real codeblock scoping which was exploited by
dictionary resizing
* changed ACCESS messages to ASSIGN ones when object item
reference is created for proper READONLY scope checking.
The reference to object variable can be created only when
caller has sufficient ASSIGN privileges.
+ added FRIEND CLASS and FRIEND FUNCTION support.
It's enough to add class definition:
FRIEND CLASS <classFunc1> [, <classFuncN> ]
and/or:
FRIEND FUNCTION <FuncName1> [, <FuncNameN> ]
and all methods of given class or given function will be able
to access private variables.
Warning!!! Friends cannot access overloaded non virtual methods.
This feature is available _ONLY_ for real class members.
+ added MODULE FRIENDLY to class definition. It causes that all other
functions and classes defined in the same .prg module will have
friend privileges. In such way works xHarbour and there is now way
to disable this "feature" what is IMHO bug. In Harbour programmer
has to explicitly enable it (until he will not change / add new
preprocessor rule and set it as default ;-)). Syntax:
CREATE CLASS .... MODULE FRIENDLY
...
END CLASS
* harbour/source/vm/proc.c
* harbour/source/vm/runner.c
* updated function symbols processing
* harbour/source/compiler/harbour.c
* added note in hb_compOptimizeFrames() about exceeding maximum number
of local variables (255). We should add new pcode(s) HB_P_LARGE[V]FRAME
or generate compile time error.
* harbour/source/vm/macro.c
% minor optimizations
+ harbour/doc/destruct.txt
+ added description for object destructors in Harbour
* harbour/include/error.ch
+ added new error code EG_DESTRUCTOR
* harbour/source/lang/msgpl852.c
* harbour/source/lang/msgpliso.c
* harbour/source/lang/msgplmaz.c
* harbour/source/lang/msgplwin.c
* harbour/source/rtl/langapi.c
+ added desription for new error code - other language modules
have to be updated
* harbour/include/hbapi.h
+ added hb_gcRefCheck() and cover some hb_gc* functions by
_HB_API_INTERNAL_ macro
* harbour/source/vm/itemapi.c
! fixed possible RT error generation when some exception is active
* harbour/include/hbapicls.h
* harbour/include/hbclass.ch
* harbour/include/hboo.ch
* harbour/source/rtl/tclass.prg
* harbour/source/vm/arrays.c
* harbour/source/vm/classes.c
+ added support for object destructors
* harbour/source/vm/garbage.c
+ added support for object destructors
+ added logic to detect buggu .prg code which uses destructors
see doc/destruct.txt for more info.
It's also possible that this code will exploit some bugs
in other code which uses GC allocated memory blocks.
* harbour/include/hbapi.h
* harbour/source/vm/hvm.c
* harbour/source/vm/codebloc.c
* harbour/source/vm/arrays.c
* moved static base offset from hb_struBlock structure
to HB_CODEBLOCK structure
+ added hclass member to hb_struBlock - it will be used in
the future for checking codeblock scope in classy so we will
real scope checking also for messages sent from codeblocks.
- removed supercast and superoffset members from hb_struArray
structure.
* harbour/include/hbclass.ch
* added class(y) like
@:<MessageName>([<MsgParams,...>])
send operator. It's not exactly the same as in class(y) where
this operator is hardcoded to executing function directly,
needs method name instead of message name and is linked statically.
In Harbour this operator uses message name so can be used also for
instance variables and make dynamic casting to the class from which
current method is inherited. In short words sending messages to @:
instead of :: causes that they work like non-virtual messages in
C++ mode.
If you do not use the same method body in different classes
then you can also use explicitly self casting:
::<myclass>:<msgname>[(...)]
and it will be a little bit faster
* harbour/include/hboo.ch
+ added: HB_OO_MSG_ASSIGN, HB_OO_MSG_ACCESS,
HB_OO_MSG_CLSASSIGN, HB_OO_MSG_CLSACCESS
They should be used insted of HB_OO_MSG_DATA and HB_OO_MSG_CLSDATA
This resolves problems with name conflicts when we were detecting
type of message (ACCESS/ASSIGN) by checking the first character
in message name. F.e. now it's possible to create exported instance
variable called __WithObject and it will be used in all WITH OBJECT
statement instead of the base object value. It's simple and effective
WITH OBJECT overloading.
I kept backward compatibility for HB_OO_MSG_DATA and HB_OO_MSG_CLSDATA
but I strongly suggest to update code to use new constants.
+ added HB_OO_MSG_REALCLASS
* harbour/source/rtl/objfunc.prg
* harbour/source/rtl/tobject.prg
* harbour/source/rtl/tclass.prg
* use HB_OO_MSG_[CLS]{ASSIGN,ACCESS} instead of HB_OO_MSG_[CLS]DATA
* harbour/source/rtl/tclass.prg
+ added REALCLASS message to creted classes - it's used for @: operator
* harbour/source/vm/classes.c
* updated for above modifications
* calculate instance area offset in supercasting dynamically - it
will allow to eliminate multiple instance area allocating when
in the inheritance tree the same class exists more then once and
also quite easy add support for non-virtual messages.
! fixed GPF in __CLSINSTSUPER() class function generate HVM exception
% do not inherit unaccessible inline blocks
* some other minor optimization, fixes and code cleanups
* harbour/source/rdd/usrrdd/usrrdd.c
! fixed HB_TRACE message
* harbour/source/rtl/errorapi.c
! generate internal error if ErrorNew() function does not return an object
* source/pp/ppcore.c
+ Added logic for new Idenitifier Match Marker <!AnyId!> which will only match Identifiers.
+ Added static BOOL IsIdentifier( char *szProspect )
* include/hbclass.ch
! Changed <name> to <!name!> in 3 VO OO compatibility rules, where regular match marker was TOO generic.
! Reinstated default #define HB_CLS_VO
* contrib/dot/pp.prg
* Added support fort '|' as a continuation token (as in Clipper).
* include/hbclass.ch
- Commented out 3 rules used for VO OO compatibility that are TOO generic, and match unexpected input.
* source/pp/ppcore.c
- Recommented out test for continuation token at the end of an Expression in isExpres() (after correcting cause for prior reinstatment).
* include/hbclass.ch
! Corrected few minor typos.
* source/compiler/harbour.slx
* source/macro/macro.slx
- Commented few unused lines.
* contrib/dot/pp.prg
% Rewrote NextToken() and NextExp().
+ Added NextIdentifier()
! Found that Clipper PP is not processing tokens read by NextToken() and NextExp() so removed recursive support,
and changed logic in ProcessLine() to linear process:
#defines against all valid Identifiers - Expansion forces a resacn from top.
#[x]translates against all tokens - Expansion forces a resacn from top of #defines.
#[x]command agains the first token - Expansion forces a resacn from top of #defines.
! Corrected and refined many aspects of Interpreter mode.
+ Added ExecuteMethod() and support for OO syntax.
+ Added support for -I<includepaths...> command line switch.
* contrib/dot/rp_run.ch
+ Added support for OO syntax.
* contrib/dot/pp.txt
* Updated documentation.
* include/hbclass.ch
% Removed many redundant rules (recently added for OO integrity checks).
/* Found the PP problem that stopped me from implementing this optimized form before.
It seems the PP has a problem with "[" follwing a an identifier without a space prefix like:
#command TEST[([<param,...>])]
It seems to work OK like this:
#command TEST [([<param,...>])]
*/
* include/hbclass.ch
! Fixed declaration support for OnError() METHODS
! Fixed support for Methods having space prefixing the "( ... )" in the method implementation.
+ Refined #error messages to report the Class Name .
* source/pp/ppcore.c
* Removed recently added but not required support of #define expansion in #error directive.
* include/hbclass.ch
+ Added logic for NO Checking of OO Syntax (#ifdef NO_OO_ERR) - Use /dNO_OO_ERR to disable OO Syntax Checking.
* source/pp/ppcore.c
* Temporarily disabled support of #define expansion in #error directive.
* include/hbclass.ch
% Removed numerous redundant rules.
+ Added logic to detect and raise #error on implementation of NON Declared method.
* source/rtl/tclass.prg
+ Added logic to dispose of posibble "(...)" suffix to Method Name in AddVirtual(), AddMethod(), and AddClsMethod().
* include/hbclass.ch
! Added missing parameter to declaration of AddMultiData()
* source/compiler/harbour.c
! Corrected broken Linked-List links in hb_compDeclaredInit()
* source/compiler/hbpcode.c
* Minor correction declared parameters checking.
* source/vm/hvm.c
+ Added complete call stack display to hb_vmRequestCancel()
/* Cancelling will now show the complete call stack, rather than the not very helpful just current procedure. */
* contrib/dot/pp.prg
* contrib/dot/rp_run.ch
! Few corrections to Pre Processor, and Interpreter.
* include/hbclass.ch
+ Added support for METHOD implementations without requiring the CLASS Clause.
+ Added support for METHOD implementations without requiring the "()" if no parameters needed.
* contrib/dot/pp.prg
! Added logic to support multi-word stoppers needed by multi-word restricted list, like
[ <lClick: ON CLICK, ON LEFT CLICK> <uLClick> ]
/* Only 1 known error remains in processing of fwh/samples/fwbrow.prg */
* include/hbclass.ch
* Temprarily removed \ preceding < of not existing match markers, in nested # directive in result.
/* Need to be readded once PP is fixed. */
* contrib/dot/pp.prg
! More corrections and refinments in pre-processing
* Improved handling of command line parameters (no longer requires -P, -CCH doesn't require -U)
! Corrected handling of file extension.
* include/hbclass.ch
* Corrected <MethodName> to \<MethodName> when result marker not a match marker.
* Corrected <params,...> to <param> (result marker used match marker format).
