[Bug libstdc++/21772] exception safety testing

["bkoz at gcc dot gnu dot org" <gcc-bugzilla at gcc dot gnu dot org>]

------- Comment #18 from bkoz at gcc dot gnu dot org  2010-01-01 03:54 -------

> multiset error

... was bogus. I adjusted the traits to fix this.

> The std::array error seems indeed bogus: if I'm not wrong, it happens when
> swapping arrays, and there are no  guarantees that the operation doesn't throw
> for std::array, because it's requires to just swap the ranges, thus copy
> construct and copy assign each array element.

It does happen when swapping arrays. I believe that array::swap does have a
strong requirement via 23.2.1 p 10, but have xfailed this for the moment.

In addition, I have removed the throwing on user defined ctors for
throw_value_* via _GLIBCXX_IS_AGGREGATE to simulate a more aggregate-ish thing.
However, even aggregates can have assignment operators and other operator
actions that throw, and this still fails. (And note the value_type of
std::array can have user defined ctors ie 8.5.1 p 13). 

> Another thing I noticed, about basic_string, something seems inappropriate for
> it in the framework, since the value_type must be a POD!

Four specializations are defined: char, wchar_t, char16_t, char32_t. That is
the only part of basic_string that is special WRT container requirements that I
can see. 

?

I went ahead and checked this in, xfailing the following failures:

FAIL: 21_strings/basic_string/requirements/exception/propagation_consistent.cc
execution test
FAIL: 23_containers/array/requirements/exception/generation_prohibited.cc
execution test
FAIL: 23_containers/deque/requirements/exception/generation_prohibited.cc
execution test
FAIL: 23_containers/deque/requirements/exception/propagation_consistent.cc
execution test
FAIL: 23_containers/vector/requirements/exception/generation_prohibited.cc
execution test


I believe that deque/.../propagation_consistent.cc will have to be modified to
take into account 23.3.2.3, and copy/assignment should be elided in this case
(and vector via 23.3.6.4) 

Regardless, this can be tweaked in place.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21772