Re: TRUNCATE on foreign table - Mailing list pgsql-hackers

On Thu, Apr 22, 2021 at 2:26 PM Justin Pryzby <pryzby@telsasoft.com> wrote:
>
> On Thu, Apr 22, 2021 at 03:36:25PM +0900, Fujii Masao wrote:
> > diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
> > index 553524553b..69aa66e73e 100644
> > --- a/doc/src/sgml/fdwhandler.sgml
> > +++ b/doc/src/sgml/fdwhandler.sgml
> > @@ -1076,27 +1076,25 @@ ExecForeignTruncate(List *rels,
> >                      bool restart_seqs);
> >      <para>
> > -     <literal>behavior</literal> defines how foreign tables should
> > -     be truncated, using as possible values <literal>DROP_RESTRICT</literal>,
> > -     which means that <literal>RESTRICT</literal> option is specified,
> > -     and <literal>DROP_CASCADE</literal>, which means that
> > -     <literal>CASCADE</literal> option is specified, in
> > -     <command>TRUNCATE</command> command.
> > +     <literal>behavior</literal> is either <literal>DROP_RESTRICT</literal>
> > +     or <literal>DROP_CASCADE</literal>, which indicates that the
> > +     <literal>RESTRICT</literal> or <literal>CASCADE</literal> option was
> > +     requested in the original <command>TRUNCATE</command> command,
> > +     respectively.
>
> Now that I reread this, I would change "which indicates" to "indicating".

+1.

> > -     <literal>restart_seqs</literal> is set to <literal>true</literal>
> > -     if <literal>RESTART IDENTITY</literal> option is specified in
> > -     <command>TRUNCATE</command> command.  It is <literal>false</literal>
> > -     if <literal>CONTINUE IDENTITY</literal> option is specified.
> > +     If <literal>restart_seqs</literal> is <literal>true</literal>,
> > +     the original <command>TRUNCATE</command> command requested the
> > +     <literal>RESTART IDENTITY</literal> option, otherwise
> > +     <literal>CONTINUE IDENTITY</literal> option.
>
> should it say "specified" instead of requested ?
> Or should it say "requested the RESTART IDENTITY behavior" ?
>
> Also, I think it should say "..otherwise, the CONTINUE IDENTITY behavior was
> requested".

The original TRUNCATE document uses this - "When RESTART IDENTITY is specified"

IMO the following looks better: "If restart_seqs is true, RESTART
IDENTITY was specified in the original TRUNCATE command, otherwise
CONTINUE IDENTITY was specified."

> > +++ b/doc/src/sgml/ref/truncate.sgml
> > @@ -173,7 +173,7 @@ TRUNCATE [ TABLE ] [ ONLY ] <replaceable class="parameter">name</replaceable> [
> >
> >    <para>
> >     <command>TRUNCATE</command> can be used for foreign tables if
> > -   the foreign data wrapper supports, for instance,
> > +   supported by the foreign data wrapper, for instance,
> >     see <xref linkend="postgres-fdw"/>.
>
> what does "for instance" mean here?  I think it should be removed.

+1.

> > +++ b/doc/src/sgml/fdwhandler.sgml
> > @@ -1111,6 +1099,15 @@ ExecForeignTruncate(List *rels, List *rels_extra,
> >       if <literal>CONTINUE IDENTITY</literal> option is specified.
> >      </para>
> >
> > +    <para>
> > +     Note that information about <literal>ONLY</literal> options specified
> > +     in the original <command>TRUNCATE</command> command is not passed to
> > +     <function>ExecForeignTruncate</function>.  This is the same behavior as
> > +     for the callback functions for <command>SELECT</command>,
> > +     <command>UPDATE</command> and  <command>DELETE</command> on
>
> There's an extra space before DELETE

Good catch! Extra space after "and" and before "<command>".

> > diff --git a/doc/src/sgml/postgres-fdw.sgml b/doc/src/sgml/postgres-fdw.sgml
> > index 5320accf6f..d03731b7d4 100644
> > --- a/doc/src/sgml/postgres-fdw.sgml
> > +++ b/doc/src/sgml/postgres-fdw.sgml
> > @@ -69,6 +69,13 @@
> >    have privileges to do these things.)
> >   </para>
> >
> > + <para>
> > +  Note that <literal>ONLY</literal> option specified in
>
> add "the" to say: "the ONLY"

+1.

> > +  <command>SELECT</command>, <command>UPDATE</command>,
> > +  <command>DELETE</command> or <command>TRUNCATE</command>
> > +  has no effect when accessing or modifyung the remote table.
>
> modifying

Good catch!

With Regards,
Bharath Rupireddy.
EnterpriseDB: http://www.enterprisedb.com