Re: Proposal : REINDEX SCHEMA

On Mon, Oct 20, 2014 at 11:24 AM, Sawada Masahiko <sawada(dot)mshk(at)gmail(dot)com>
wrote:
>
> On Mon, Oct 20, 2014 at 9:41 AM, Fabrízio de Royes Mello
> <fabriziomello(at)gmail(dot)com> wrote:
> >
> >
> > On Sun, Oct 19, 2014 at 10:37 PM, Fabrízio de Royes Mello
> > <fabriziomello(at)gmail(dot)com> wrote:
> >>
> >>
> >> On Sun, Oct 19, 2014 at 1:02 PM, Sawada Masahiko <sawada(dot)mshk(at)gmail(dot)com
>
> >> wrote:
> >> >
> >> > On Fri, Oct 17, 2014 at 4:32 AM, Fabrízio de Royes Mello
> >> > <fabriziomello(at)gmail(dot)com> wrote:
> >> > > On Wed, Oct 15, 2014 at 11:41 AM, Sawada Masahiko
> >> > > <sawada(dot)mshk(at)gmail(dot)com>
> >> > > wrote:
> >> > >>
> >> > >> On Mon, Oct 13, 2014 at 11:16 PM, Robert Haas <
robertmhaas(at)gmail(dot)com>
> >> > >> wrote:
> >> > >> > On Sun, Oct 12, 2014 at 1:27 PM, Stephen Frost <
sfrost(at)snowman(dot)net>
> >> > >> > wrote:
> >> > >> >> * Alvaro Herrera (alvherre(at)2ndquadrant(dot)com) wrote:
> >> > >> >>> Sawada Masahiko wrote:
> >> > >> >>> > Attached WIP patch adds new syntax REINEX SCHEMA which does
> >> > >> >>> > reindexing
> >> > >> >>> > all table of specified schema.
> >> > >> >>> > There are syntax dose reindexing specified index, per table
and
> >> > >> >>> > per
> >> > >> >>> > database,
> >> > >> >>> > but we can not do reindexing per schema for now.
> >> > >> >>>
> >> > >> >>> It seems doubtful that there really is much use for this
feature,
> >> > >> >>> but
> >> > >> >>> if
> >> > >> >>> there is, I think a better syntax precedent is the new ALTER
> >> > >> >>> TABLE ALL
> >> > >> >>> IN TABLESPACE thingy, rather than your proposed REINDEX
SCHEMA.
> >> > >> >>> Something like REINDEX TABLE ALL IN SCHEMA perhaps.
> >> > >> >>
> >> > >> >> Yeah, I tend to agree that we should be looking at the 'ALL IN
> >> > >> >> TABLESPACE' and 'ALL IN SCHEMA' type of commands to keep things
> >> > >> >> consistent. This might be an alternative for the vacuum /
analyze
> >> > >> >> /
> >> > >> >> reindex database commands also..
> >> > >> >
> >> > >> > Urgh. I don't have a problem with that syntax in general, but
it
> >> > >> > clashes pretty awfully with what we're already doing for REINDEX
> >> > >> > otherwise.
> >> > >> >
> >> > >>
> >> > >> Attached patches are latest version patch.
> >> > >
> >> > > Ok.
> >> > >
> >> > >
> >> > >> I changed syntax to REINDEX ALL IN SCHEMA, but I felt a sense of
> >> > >> discomfort a little
> >> > >> as Robert mentioned.
> >> > >>
> >> > >
> >> > > I understood, but the real problem will in a near future when the
> >> > > features
> >> > > will be pushed... :-)
> >> > >
> >> > > They are separated features, but maybe we can join this features
to a
> >> > > one
> >> > > future commit... it's just an idea.
> >> > >
> >> > >
> >> > >> Anyway, you can apply these patches in numerical order,
> >> > >> can use REINDEX ALL IN SCHEMA feature and "-S/--schema" option in
> >> > >> reindexdb.
> >> > >>
> >> > >> 000_reindex_all_in_schema_v2.patch : It contains REINDEX ALL IN
> >> > >> SCHEMA
> >> > >> feature
> >> > >
> >> > > 1) Compile without warnings
> >> > >
> >> > >
> >> > > 2) IMHO you can add more test cases to better code coverage:
> >> > >
> >> > > * reindex a schema that doesn't exists
> >> > > * try to run "reindex all in schema" inside a transaction block
> >> > >
> >> > >
> >> > > 3) Isn't enough just?
> >> > >
> >> > > bool do_database = (kind == OBJECT_DATABASE);
> >> > >
> >> > > ... instead of...
> >> > >
> >> > > + bool do_database = (kind == OBJECT_DATABASE) ? true : false;
> >> > >
> >> > >
> >> > > 4) IMHO you can add other Assert to check valid relkinds, like:
> >> > >
> >> > > Assert(kind == OBJECT_DATABASE || kind == OBJECT_SCHEMA);
> >> > >
> >> > >
> >> > > 5) I think is more legible:
> >> > >
> >> > > /* Get OID of object for result */
> >> > > if (do_database)
> >> > > objectOid = MyDatabaseId
> >> > > else
> >> > > objectOid = get_namespace_oid(objectName, false);
> >> > >
> >> > > ... insead of ...
> >> > >
> >> > > + /* Get OID of object for result */
> >> > > + objectOid = (do_database) ? MyDatabaseId :
> >> > > get_namespace_oid(objectName,
> >> > > false);
> >> > >
> >> > >
> >> > >
> >> > >> 001_Add_schema_option_to_reindexdb_v1.patch : It contains
reindexdb
> >> > >> "-S/--schema" supporting
> >> > >>
> >> > >
> >> > > The code itself is good for me, but IMHO you can add test cases to
> >> > > src/bin/scripts/t/090_reindexdb.pl
> >> > >
> >> >
> >> > Thank you for reviewing.
> >>
> >> You're welcome!
> >>
> >>
> >> > I agree 2) - 5).
> >>
> >> :-)
> >>
> >>
> >> > Attached patch is latest version patch I modified above.
> >>
> >> All is fine to me now... all work as expected and no compiler warnings.
> >>
> >> There are just a little fix to do in src/bin/scripts/t/090_reindexdb.pl
> >>
> >> -use Test::More tests => 7;
> >> +use Test::More tests => 8;
> >>
> >> Because you added a new testcase to suittest, so you need to increase
the
> >> test count at beginning of the file.
> >>
> >
> > Patch attached. Now the regress run without errors.
> >
>
> Thank you for reviewing and revising!

You're welcome!

> I also did successfully.
> It looks good to me :)
>

Changed status to "Ready for commiter".

Regards,

--
Fabrízio de Royes Mello
Consultoria/Coaching PostgreSQL
>> Timbira: http://www.timbira.com.br
>> Blog: http://fabriziomello.github.io
>> Linkedin: http://br.linkedin.com/in/fabriziomello
>> Twitter: http://twitter.com/fabriziomello
>> Github: http://github.com/fabriziomello