Re: Fix preprocessor/11839

[Neil Booth <neil at daikokuya dot co dot uk>]

Geoff Keating wrote:-

> Neil Booth <neil@daikokuya.co.uk> writes:
> 
> > My cppfiles.c rewrite fixed the case where a file was unreadable
> > (we wouldn't complain, and continue searching) but missed the
> > case where a path component is a file.  I tried to create a
> > testcase for this, and failed.
> 
> It's unrelated to this patch, but have you considered instead having
> an immediate error for the case where a file is unreadable?  Normally
> that is due to a broken build environment.  You usually don't want the
> compiler going on and perhaps finding another file of that name
> somewhere else.

That's what my paragraph above attempted to explain.  We *wouldn't*
complain and continue searching.  Now we stop and complain (try it!).
It was fixing this that introduced this regression.

Neil.