Re: [patch] Zap remaining calls to error_begin()

[Daniel Jacobowitz <drow at mvista dot com>]

On Mon, Feb 04, 2002 at 11:21:54AM +0000, Richard Earnshaw wrote:
> > > 2002-01-31  Andrew Cagney  <ac131313@redhat.com>
> > > 
> > > * utils.c (error_begin): Make static.
> > >       * defs.h (error_begin): Delete declaration.
> > > 
> > >       * linespec.c (cplusplus_error): Replace cplusplus_hint.
> > >       (decode_line_1): Use cplusplus_error instead of error_begin,
> > >       cplusplus_hint and return_to_top_level.
> > >       * coffread.c (coff_symfile_read): Use error instead of error_begin
> > >       and return_to_top_level.
> > >       * infrun.c (default_skip_permanent_breakpoint): Ditto.
> > > 
> > 
> > I've checked this in.
> 
> I'm not sure if this is related, but I'm now seeing a regression in the 
> testsuite for an arm-netbsd target:
> 
> FAIL: gdb.c++/annota2.exp: annotate-quit
> 
> Since the missing text from the log is:
> 
> -   error-begin 
> - Quit 
> -  
> -   quit 
>  
> I think the smoking gun must be pointing somewhere in this direction.

Try running the test a couple dozen times, and see what happens,
please?

This test fails spuriously, or perhaps passes spuriously.  I think
there's something actually wrong with GDB's signal handling in relation
to the way expect invokes us.

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer