← Back to team overview

maria-developers team mailing list archive

Re: 0608c273598: MDEV-15966: Behavior for TRUNCATE versioned table is not documented and not covered by tests


Hi, Nikita!

On Feb 28, Nikita Malyavin wrote:
> On Fri, Dec 28, 2018 at 9:18 AM Sergei Golubchik <serg@xxxxxxxxxxx> wrote:
> > On Dec 28, Nikita Malyavin wrote:
> > > Hi, Sergei!
> > >
> > > In short, I am on the side of standard conformance.
> >
> > Our TRUNCATE is not exactly a standard TRUNCATE. it's defined as
> > DROP+REPLACE. But generally I agree, standard conformance.
> The fun thing is that even handler::truncate() is defined as
> delete_all_rows() 🙂
> Actually even ha_innodb does something same, and frm file is never
> deleted/touched.
> So physically it's always deletion of rows data, not DROP + CREATE
> Maybe I'm missing something, am I?

you are :)

If the handlerton has a flag HTON_CAN_RECREATE, then
Sql_cmd_truncate_table::lock_table() will set hton_can_recreate
and Sql_cmd_truncate_table::truncate_table() will use
dd_recreate_table(), not handler::delete_all_rows()

> > I'd try to remove HTON_CAN_RECREATE flag completely and implement
> > recreate logic inside handler::delete_all_rows()
> >
> > Yes, good idea, I'm thinking the same.
> So am I supposed to file MDEV for it then?

Yes, please

Chief Architect MariaDB
and security@xxxxxxxxxxx

Follow ups