Boris Kolpackov wrote:
Vitaly Prapirny <[EMAIL PROTECTED]> writes:
I can't explain such behaviour.
I can: compiler bugs ;-). What amazes me the most is not how buggy
the Borland compilers are but that people keep using them release
after release. I think Borland will need to format their disk or
reflash the BIOS with garbage to finally get rid of them.
:)
After
#include <string.h>
or
#include <string.h>
#include <cstring>
we can use memcpy.
After
#include <cstring>
#include <string.h>
we can use std::memcpy only.
After
#include <mem.h>
(non-standard borland header file) we can use memcpy and this not
depends on header files order.
There is the bug in the borland header file _str.h indeed.
Two questions arises:
1. Should we fix this? There is opened Jira issue XERCESC-1560 on this
bug. The bug exists in bcb6 also.
2. Should we migrate to C++ header files instead of old-style (and
deprecated according to last drafts of C++ standard) C ones in some
future?
Good luck!
Vitaly
