maria-developers team mailing list archive
Mailing list archive
Re: new timestamp fields
On Jan 15, Zardosht Kasheff wrote:
> Unfortunately, we need to do comparisons during phases when the Table
> object, and therefore the Field object, may not be around, such as
> during recovery. So for fixed length fields, we privately encode their
> length and metadata that explains how to compare them.
Yes, of course. I've seen this in other engines too.
I mean, that to decide on how to compare, it's better to look at
Field::key_type(), not at Field::type(). You can still privately encode
this decision and store it in the table or elsewhere, field's key_type()
will never change without an ALTER TABLE.
> > But really the API-correct way is to use Field::key_type() method
> > and compare the values using the specified method. It's more
> > future-proof than to hard-code comparison method mapping based on
> > the field types.