This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: Remove testsuite/23_containers/map/operators/1.cc ?
- From: Jonathan Wakely <jwakely at redhat dot com>
- To: Paolo Carlini <paolo dot carlini at oracle dot com>
- Cc: François Dumont <frs dot dumont at gmail dot com>, "libstdc++ at gcc dot gnu dot org" <libstdc++ at gcc dot gnu dot org>
- Date: Wed, 4 Jun 2014 21:52:50 +0100
- Subject: Re: Remove testsuite/23_containers/map/operators/1.cc ?
- Authentication-results: sourceware.org; auth=none
- References: <538F7FE8 dot 2020302 at gmail dot com> <538F8265 dot 3040903 at oracle dot com>
On 04/06/14 22:32 +0200, Paolo Carlini wrote:
Hi,
On 06/04/2014 10:22 PM, François Dumont wrote:
Hello
Do you see any objection in removing this file ? It doesn't look
like a test at all to me, just send some datas to std::cerr. There
is also a weird comment talking about multiset while code is using
map.
The comment seems a pasto, which should be fixed. Likewise the uses of
the inserters should be either turned into VERIFYs or alternately the
testcase changed to do-compile (the original issue was a compile-time
issue, but years ago we used to mostly add to the library testsuite
do-run testcases). Indeed, we could also move it to where the other
insert tests are (I think the reason of the weird directory has to do
with the semantics of the operator[] in C++98).
It looks like it only needs to do this, as a compile-only test:
typedef std::map<int,const int> MapInt;
MapInt m;
for (MapInt::const_iterator i=m.begin();i!=m.end();++i)
{ }
for (MapInt::const_iterator i=m.begin();m.end()!=i;++i)
{ }
The bug that it tests was in iterator comparisons:
https://gcc.gnu.org/ml/libstdc++/2000-11/msg00093.html