Re: Does this string processing work for you?

"David Frank" <dave_frank@xxxxxxxxxxx> writes: >
> "*** Hendrickson" <***.hendrickson@xxxxxxx> wrote in message
> news:EbV_e.91774$qY1.24574@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
> >
> >
> >
> > If Fortran did adopt some rule about returning an index,
> > what would you like to happen for
> > read( s, *, err=1 ) ( v(i) ,i=1,5 ), (x(j), j=1,5)
> > what values would you like for i and j if the error
> > happens in the first 5 values? How do you determine
> > whether it did or not?
> >
> > *** Hendrickson
> >
>
> Actually its NOT a case of asking for additional processing to return a
> index,
> its just asking compiler writers to keep their COTTON-PICKING HANDS OFF the
> existing index when a error is detected.

It isn't a question of "keeping ... hands off".
When code sich as this is optimised, the control variable
may be kept in a register, rather than in storage.
When the loop terminates unexpectedly, the contents
of the register are lost, and the value of the
variable in storage is not changed.
.