Re: parsing-error-strangeness

[Jan-Henrik Haukeland]

Jan-Henrik Haukeland <address@hidden> writes:

>> It should complain over line 5 and not with ''.
>
> Hit yourself on the head :-) I had to fix the line number counting in
> l.l for SERVICETOKENS you added and the reporting is better but it is
> still not perfect because of this, it may be off by one or more lines:
>
>
>   [ \n\t]+[^,]    {
>                     lineno++;
>                     unput(yytext[strlen(yytext)-1]);
>                     BEGIN(INITIAL);
>                   }
>
> lineno is incremented once but since [ \n\t]+[^,] may contain more
> than one \n line number is off by the number of \n + 1. I have been up
> all night and is to tired to find a good solution for this, but maybe
> you can?

Just to let you know if you plan to give it a stab. If you use this
version below, line number is correct when reporting the error. I
thought I got it right because I tested with the version below, but
with your example it turned out to be wrong after all (as explained
above)

check process crond with pidfile /var/run/crond.pid
      depends ssh
      alert address@hidden

check process ssh with pidfile /var/run/sshd.pid
      port 22 protocol ssh


But if you do this, line number is off by n-1 (is the right offset)

check process crond with pidfile /var/run/crond.pid
      depends ssh
^
 Number of line numbers off
v
      alert hauk


check process ssh with pidfile /var/run/sshd.pid
      port 22 protocol ssh

or 

check process crond with pidfile /var/run/crond.pid
      depends ssh
^

 Number of line numbers off

v
check process ssh with pidfile /var/run/sshd.pid
      port 22 protocol ssh



-- 
Jan-Henrik Haukeland