On Wed, Sep 7, 2011 at 23:48, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Also, the way yours is set up, I'm dubious that it
> does the right thing when the last '0' specifier is to the left of the
> decimal point.
When the last '0' is left of the decimal point, Num->zero_end is set
to 0, so the branch dealing with that is never executed.
> I'm currently testing this patch:
Looks good to me. It's potentially less efficient, but I'm not worried
about that.
Regards,
Marti