Re: fix the sscanf usage in etags

[Jan Djärv]

Kevin Ryde skrev:
> Masatake YAMATO <address@hidden> writes:
> > The problem I found was that etags crashed when `#line 123 L"wide"' was given
> > as input. If the crashing can be suppressed, it is enough.
>
> That's the only bad case, you don't have some strange libc that
> incorrectly counts %n?
>
> Jan Djärv <address@hidden> writes:
> >           if (sscanf (lbp->buffer, "#line %d %n\"", &lno, &start) >= 1
> >               && lbp->buffer[start] == '"')
>
> I think that still uses start uninitialized on say "#line 123 ".

No, in that case start will be 10. If the %d matches, sscanf will return 1 and 
in that case %n must match (whitespace is optional).

        Jan D.