Re: monitrc erroneous line numbering off by one

[Christian Hopp]

On Thu, 12 Feb 2004, Christian Hopp wrote:

> On Thu, 12 Feb 2004, Jan-Henrik Haukeland wrote:
>
> > Martin Pala <address@hidden> writes:
> >
> > >> 2.) parser increments the line position (lineno) by all empty and
> > >> comment lines following the invalid line, which is reported. Can
> > >> some flex guru look on it? Both attached configurations still
> > >> produces this kind of error.
> > >>
> > >
> > > This problem remains. Is here some way how to ignore "\n" for empty
> > > and comment lines following invalid statement line?
> >
> > It's tricky because when the statement is treated in the action block
> > of the parser, flex has already scanned to the next token, and if
> > there was white space or comments "#.." the lineno variable will be
> > increased (at least that is the hypothesis). In other words when we
> > report an error in lineno, the actual line number can/will be off by
> > the number of whitespace lines to the next valid token. Hmm, food for
> > thoughts.. how to fix this whiteout hacking to much. Christian, any
> > ideas?
>
> This problem happens as soon as we deal with strings (same for paths,...)
> and statements with variable arguments... like allow.

Okay its anykind of "argument" we are referring to... also numbers!

> My idea is to save the position when ever a string type is parsed.  And
> whenever an error is referring to such an error we use the saved pos
> instead of the lineno.  It seems to work but it needs further testing.

I also found a possible segfault (and a fix) in the parser in
p.y:addhtpasswdentry.

CHopp

-- 
Christian Hopp                                email: address@hidden
Institut für Elektrische Informationstechnik             fon: +49-5323-72-2113
TU Clausthal, Leibnizstr. 28, 38678 Clausthal-Zellerf.   fax: +49-5323-72-3197
                             pgpkey: https://www.iei.tu-clausthal.de/pgp-keys/