hb_par[v]dsbuff() return same buffer pased, szDate, but hb_par[v]ds() I think
that logic is similar to hb_parc[x] although both
return a writable area (asString.value, hb_stack.szDate), IMHO.
../../dlmalloc.c:4065: warning: dereferencing pointer 'b' does break
strict-aliasing rules
Another thing, I do not know how to clean warnings in dlmalloc.c when compiling
with g++ and strict-aliasing.
I think the code is correct and it's a problem with the name of tags and/or the
interpretation of g++ rules in macros with
(char*) but g++ option -fno-strict-aliasing or #pragma GCC diagnostic warning
"-w", that's only work, affects the rest of code.
Any idea?
--
Xavi
Przemyslaw Czerpak escribió:
On Tue, 30 Jun 2009, Xavi wrote:
Some questions reviewing hbapi.h
Should be hb_pards and hb_parvds const?
buffer returned by hb_par[v]ds() is writable so we can leave it
as is. Anyhow if you think that it will be cleaner to define
this buffer as readonly then we can change it.
Do you need to change hbtypes.h?
Yes, they should be updated though I do not know if it's worth to keep
all these function wrappers. For .prg code compiled with -gc2 we need
only hb_vmExecute() and hb_vmProcessSymbols().
For -gc3 we need also hb_xvm*() functions which are still missing.
All other function wrappers cover only small part of public
Harbour API so they are usable only in some limited way for 3-rd
party C code which does not use any other functions.
best regards,
Przemek
_______________________________________________
Harbour mailing list
[email protected]
http://lists.harbour-project.org/mailman/listinfo/harbour
_______________________________________________
Harbour mailing list
[email protected]
http://lists.harbour-project.org/mailman/listinfo/harbour
