This is the mail archive of the cygwin mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

pcre, C++


In version 6.0, the (upstream) pcre maintainer added a C++ wrapper
contributed by Google.  However, in pcre-6.7 (and going back all the way
to 6.0), I can find no case where the C++ unittests
pcre_scanner_unittest or pcrecpp_unittest actually pass; they always
segfault on cygwin (gcc/g++ 3.4.4-1, binutils 20060817-1) [and also
mingw].

A little digging shows that the problem occurs in pcrepp.cc (RE::Init):

void RE::Init(const char* pat, const RE_Options* options) {
  pattern_ = pat;  <<< HERE

where std::string pattern_, a member of class RE, is assigned the const
char* pat -- and pat is NOT null (verified in gdb) (it is a valid,
null-terminated regex string).  This is certainly a legal thing to do,
as std::string has a operator=(const char*).  But for some reason, it
fails here.  I think that somehow this code is tickling a g++/stdlibc++
bug, but my gdb-foo is not sufficient to figure it out.  I tried
recompiling with -O0, but (a) the test progs still failed, and (b) there
was no additional enlightenment under gdb.

[Note: pcrecpp works fine on linux[gcc]/solaris[Sun Workshop
8.0]/irix[MIPS-Pro], at least as tested by pcre_scanner_unittest and
pcrecpp_unittest]

Has anyone else observed this failure on cygwin with pcrecpp?  Anybody
got a clue what's really going on here?

--
Chuck




--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]