This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: New AIX libstdc++ testsuite failures
- From: Simon Martin <simartin at users dot sourceforge dot net>
- To: David Edelsohn <dje at watson dot ibm dot com>
- Cc: libstdc++ at gcc dot gnu dot org
- Date: Tue, 20 Feb 2007 08:02:46 +0100
- Subject: Re: New AIX libstdc++ testsuite failures
- References: <200702192241.l1JMfZ835226@makai.watson.ibm.com>
Hello David.
> The number of libstdc++ testsuite failures doubled a few days
> ago. The failures all are of the same form. Any idea what caused this
> and how to fix it?
The cause is the fix that I've committed for PR c++/14622 (see
http://gcc.gnu.org/ml/gcc-patches/2007-02/msg01082.html), that detects type
mismatches between explicit instantiations and their declaration.
It's exactly what happens here: basic_string<...>::_Rep::_S_max_size is
declared as a const size_type, and it is explicitly instantiated as a
size_type in the test case, hence the error message.
Since I've tested on i686-pc-linux-gnu, I did not see those occurences of the
error in the testsuite (there were some others that I've fixed), sorry...
I can look into this at the end of the day and send you a patch (I don't have
access to AIX).
Sorry for the inconvenience.
Best regards,
Simon