Hi Sergei,
Sergei Golubchik wrote:
...
> Does your patch fixes their bug (makes test suite to pass) too ?
No. It doesn't. Their test doesn't negate anything, so my fix isn't even
used. But even if it would, it would just avoid an over-length of 1. In
their case (literal number casted to decimal) the field length is 83! 67
would be allowed with sign and decimal point. So it's far, far away from
what my fix could cure.
However, it might perhaps be better to fix the "cast to decimal", not to
create field lengths beyond what a decimal can handle. 83 could perhaps
come from 9 * 9 + sign + decimal point. But for a new decimal type 16
has to be subtracted. Maybe this has been forgotten somewhere? A quick
look didn't support this idea though.
Regards
Ingo
--
Ingo Strüwing, Senior Software Developer
MySQL GmbH, Radlkoferstr. 2, D-81373 München
Geschäftsführer: Kaj Arnö - HRB München 162140
