Re: [pcre-dev] [Bug 1038] memcmp not available in pcrecpp_s…

} Yes, I'm following the thread. I am about to pull together an 8.11
} candidate release, so it would be good to decide exactly what to do
} for #1026, #1038, and #1038 (all Borland issues) soon. I'll go ahead
} with the candidate for testing anyway as soon as I have it ready,
} even without agreed patches, because we could put it in afterwards,
} before the proper release, if necessary.

OK, put in the change I described (string.h to cstring, and the two
'using' statements) for the candidate release. It *shouldn't* break
anyone, but is good to test as part of the RC.

} #1026, #1038, and #1038 (all Borland issues) soon. I'll go ahead

I think you mean 1036, 1038, and 1039. :-)

As for 1039, let's just remove the line (from pcre_stringpiece_unittest.cc)
using std::make_pair

and replace the three lines that say
map.insert(make_pair(p1, 0));
map.insert(make_pair(p2, 1));
map.insert(make_pair(p3, 2));
with
map.insert(std::make_pair(p1, 0));
map.insert(std::make_pair(p2, 1));
map.insert(std::make_pair(p3, 2));

While in the area, let's get rid of the line
using std::map;

and replace the line
typedef map<StringPiece, int> TestMap;
with
typedef std::map<StringPiece, int> TestMap;

Philip, sound ok to you? Those should fix the two borland bugs that
have to do with c++, at least.

craig