Re: [bug-gawk] gawk erroneous comparison with large negative exponents

[Andrew J. Schorr]

Hi,

On Thu, Apr 16, 2015 at 01:23:30PM +0300, Aharon Robbins wrote:
> It looks good to me. Please pull & merge and then push to gawk-4.1-stable
> and from thence merge to master.

Done.

> > But I must confess that I'm a bit confused about the flag usage.
> > The problem with the previous patch was that values where the leading
> > part is numeric should ignore trailing alpha junk.  So "3D" should be
> > converted to a numeric value of 3.  When that conversion occurs, why do
> > we not set the NUMCUR flag?
> 
> We do. I don't understand your question.

When (ptr != cpend), i.e. there is trailing unconverted junk, the flags
are not updated.  So NUMCUR is not set.  As a result, it will be converted
each time a numeric value is needed.

> 
> > Also, the comment next to the NUMBER flag sets "assigned as number",
> > so we does r_force_number enable this flag?
> 
> It doesn't, not anywhere that I can use. It uses NUMCUR everywhere.

It may set the NUMBER flag in a couple of places:

                } else if (n->stlen == 4 && is_ieee_magic_val(n->stptr)) {
                        if ((n->flags & MAYBE_NUM) != 0)
                                n->flags &= ~MAYBE_NUM;
                        n->flags |= NUMBER|NUMCUR;
                        n->numbr = get_ieee_magic_val(n->stptr);
...
        if ((n->flags & MAYBE_NUM) != 0) {
                newflags = NUMBER;
                n->flags &= ~MAYBE_NUM;

> > It seems to me that r_force_number
> > can be called when it was not actually assigned as a number.  I'm
> > interpreting "assigned as number" to mean something like
> >    x = 3
> > Is that not what it means?
> 
> Yes, but r_force_number uses NUMCUR, not NUMBER.

It sets NUMBER in some cases, but mostly when MAYBE_NUM was set.
That might be the right thing to do.  I haven't dug into it.

> Can you show a real test case that shows a problem? In particular where
> gawk is different from BWK awk and/or mawk?
> 
>       $ git branch
>         ....
>       * gawk-4.1-stable
>         ...
>       $ ./gawk 'BEGIN { x = "3 and a half" ; print x + 2 }'
>       5

At the moment, I'm concerned that we don't set NUMCUR in this case.
With the attached debug patch, I see this:

bash-4.2$ ./gawk 'BEGIN { x = "3 and a half" ; print x + 2; print x+3 }'
debug: r_force_number(3 and a half) started with flags 0x7
debug: r_force_number(3 and a half) finished with flags 0x7
5
debug: r_force_number(3 and a half) started with flags 0x7
debug: r_force_number(3 and a half) finished with flags 0x7
6

Contrast to:

bash-4.2$ ./gawk 'BEGIN { x = "23" ; print x + 2; print x+3 }'
debug: r_force_number(23) started with flags 0x7
debug: r_force_number(23) finished with flags 0xf
25
26

Note that my (committed) patch did not change this aspect of the logic.

Regards,
Andy

debug.patch
Description: Text document