Обсуждение: new aggregate functions v3

Dear patchers,

please find attached my third patch submission for adding new aggregate
functions. This new version adds a note and some indexes in the
documentation about sql standard names for bool_and and bool_or, as
suggested by Robert Treat. It also fixes conflicts created by the removal
of shameful aclitem accessors.

The added aggregates are:

(1) boolean-and and boolean-or aggregates named bool_and and bool_or.
    they (SHOULD;-) correspond to standard sql every and some/any aggregates.
    they do not have the right name as there is a problem with
    the standard and the parser for some/any. Tom also think that
    the standard name is misleading because NULL are ignored.

(2) bitwise integer aggregates named bit_and and bit_or for
    int2, int4, int8 and bit types. They are not standard, but I find
    them useful. I needed them once.


The patches adds:

- 2 new very short strict functions for boolean aggregates in
  src/backed/utils/adt/bool.c,
  src/include/utils/builtins.h and src/include/catalog/pg_proc.h

- the new aggregates declared in src/include/catalog/pg_proc.h and
  src/include/catalog/pg_aggregate.h

- some documentation and validation about these new aggregates.


It also updates the catalog version. It validates for me.

Have a nice day,

--
Fabien Coelho - coelho@cri.ensmp.fr

[ Use newest version.]

Your patch has been added to the PostgreSQL unapplied patches list at:

    http://momjian.postgresql.org/cgi-bin/pgpatches

I will try to apply it within the next 48 hours.

---------------------------------------------------------------------------


Fabien COELHO wrote:
>
> Dear patchers,
>
> please find attached my third patch submission for adding new aggregate
> functions. This new version adds a note and some indexes in the
> documentation about sql standard names for bool_and and bool_or, as
> suggested by Robert Treat. It also fixes conflicts created by the removal
> of shameful aclitem accessors.
>
> The added aggregates are:
>
> (1) boolean-and and boolean-or aggregates named bool_and and bool_or.
>     they (SHOULD;-) correspond to standard sql every and some/any aggregates.
>     they do not have the right name as there is a problem with
>     the standard and the parser for some/any. Tom also think that
>     the standard name is misleading because NULL are ignored.
>
> (2) bitwise integer aggregates named bit_and and bit_or for
>     int2, int4, int8 and bit types. They are not standard, but I find
>     them useful. I needed them once.
>
>
> The patches adds:
>
> - 2 new very short strict functions for boolean aggregates in
>   src/backed/utils/adt/bool.c,
>   src/include/utils/builtins.h and src/include/catalog/pg_proc.h
>
> - the new aggregates declared in src/include/catalog/pg_proc.h and
>   src/include/catalog/pg_aggregate.h
>
> - some documentation and validation about these new aggregates.
>
>
> It also updates the catalog version. It validates for me.
>
> Have a nice day,
>
> --
> Fabien Coelho - coelho@cri.ensmp.fr

Content-Description:

[ Attachment, skipping... ]

>
> ---------------------------(end of broadcast)---------------------------
> TIP 8: explain analyze is your friend

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073

Fabien COELHO wrote:
> (1) boolean-and and boolean-or aggregates named bool_and and bool_or.
>     they (SHOULD;-) correspond to standard sql every and some/any aggregates.
>     they do not have the right name as there is a problem with
>     the standard and the parser for some/any. Tom also think that
>     the standard name is misleading because NULL are ignored.

As I understand it, there's an ambiguity issue with SOME/ANY, but not
with EVERY. If so, can we implement EVERY per-spec at least? It's okay
if we just add EVERY as an alias for BOOL_AND for the sake of homogeneity.

A few trivial points:

> + /* EVERY aggregate implementation conforming to SQL 2003 standard.
> +  * must be strict.
> +  */

This comment is misleading if we don't actually provide an
implementation of EVERY that conforms to spec. There's a similar comment
WRT to SOME/ANY.

> + PG_FUNCTION_INFO_V1(booland_statefunc);

Not needed for builtin functions (they are assumed to be V1).

> + /* what about every? */
> + DATA(insert OID = 2517 ( bool_and                        PGNSP PGUID 12 t f f f i 1 16 "16" _null_ aggregate_dummy
-_null_ )); 
> + DESCR("boolean-and aggregate");
> + /* what about any/some? */

Seems these questions should be removed, no?

-Neil

Dear Neil,

> As I understand it, there's an ambiguity issue with SOME/ANY, but not
> with EVERY. If so, can we implement EVERY per-spec at least? It's okay
> if we just add EVERY as an alias for BOOL_AND for the sake of homogeneity.

Ok.

> > + /* EVERY aggregate implementation conforming to SQL 2003 standard.
> > +  * must be strict.
> > +  */
>
> This comment is misleading if we don't actually provide an
> implementation of EVERY that conforms to spec. There's a similar comment
> WRT to SOME/ANY.

I agree it is somehow misleading. I'll clarify.

> > + PG_FUNCTION_INFO_V1(booland_statefunc);
> Not needed for builtin functions (they are assumed to be V1).

Ok, I'll drop that.

> > + /* what about every? */
> > + DATA(insert OID = 2517 ( bool_and                        PGNSP PGUID 12 t f f f i 1 16 "16" _null_
aggregate_dummy- _null_ )); 
> > + DESCR("boolean-and aggregate");
> > + /* what about any/some? */
>
> Seems these questions should be removed, no?

Well, the question really means "what about naming it every", that is
you're very question above!

I'll do a fix wrt to your comments, and send a 4th version.

Thanks for your comments.

--
Fabien Coelho - coelho@cri.ensmp.fr