This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
RE: Something needs to change but I'm not certain what.
- From: "Stephen M. Webb" <stephen at bregmasoft dot com>
- To: "Phil Edwards" <pedwards at disaster dot jaj dot com>,<libstdc++ at gcc dot gnu dot org>
- Date: Mon, 26 Nov 2001 19:20:19 -0500
- Subject: RE: Something needs to change but I'm not certain what.
> Phil Edwards wrote:
>
> My autobuilder pukes on 23_containers/list_modifiers.cc with an error
> generated by the concept checkers. The class 'T' (which by the way is
> really useful, thanks Stephen!) has:
>
> private:
> const int itsId;
> const bool willThrow;
>
> Trying to instantiate list<T> trips over the Assignable requirement of
> [23.1]/3 and Table 64:
>
> Personally, I feel that we should be able to violate certain requirements
> in our own testsuite in order to test other requirements, as long as the
> two sets of requirements are orthogonal. So the easiest fix is to force
> the concept checks off for this file. Shall I add the appropriate #undef?
Gack. I wanted a minimal class for doing the testing, and I didn't want any
assignments (only copy constructors).
I added an assignment operator with no implementation and it compiles for
me.
But then again, so did the original, since I built with checks turned off.
I would imagine we want to leave concept checks on for test runs, so I'll
look at what I was trying to do and redo it if necessary. I believe I
originally
wrote that file in May but since I want to copy the idea elsewhere, I think
it
should work within the standard.
__
Stephen M. Webb