On Sat, Feb 01, 2014 at 11:55:22AM +0000, Richard Sandiford wrote:
> Uros Bizjak <ubiz...@gmail.com> writes:
> >> The scan-assembler test in vector26.C was showing up unresolved for c++98,
> >> since in that case we report an expected error and don't produce any asm.
> >>
> >> Tested on mips64-linux-gnu and applied as obvious.
> >
> > -// { dg-final { scan-assembler-not "globl\[ \t]*_Z3bar" } }
> > +// { dg-final { scan-assembler-not "globl\[ \t]*_Z3bar" { target { !
> > c++98 } } } }
> >
> >
> > Shouldn't "]" be escaped as well?
>
> Not my call, but maybe for consistency. It doesn't matter in practice
> since "]" is only treated as part of a Tcl command if it closes a "[".
We have over 800 cases where the ] is not escaped and over 6500 where it is
apparently, so I think it doesn't really matter, we aren't consistent
already in many places.
Jakub
