[RFA:] Make 23_containers/set/modifiers/16728.cc not time out nonnatively

[Hans-Peter Nilsson]

> Date: Tue, 6 Feb 2007 01:08:49 +0000 (UTC)
> From: "Joseph S. Myers" <joseph@codesourcery.com>

> On Mon, 5 Feb 2007, Joe Buck wrote:
> 
> > But in this case, does anyone actually benefit from the extra iterations?
> > Are they testing anything that is not tested with fewer iterations?  If
> > not, why not use fewer iterations everywhere?
> 
> That would be reasonable.  Just not making the difference depend on 
> whether the toolchain is cross or not; if, say, I test compilers for 
> powerpc-linux, the test should be the same whether I run a compiler on the 
> powerpc-linux system, or run it on an i686-linux system with the tests run 
> on the powerpc-linux system via ssh.

We don't have a "simulator" property handy, and IMHO it's not
worthwhile to introduce it; the distinction would likely be
missed anyway in other new tests.

Anyway, I'd like to hear from a libstdc++ maintainer,
particularly about reducing this test to 10 iterations all over.

brgds, H-P