Обсуждение: struct RelOptInfo member relid comments

Поиск
Список
Период
Сортировка

struct RelOptInfo member relid comments

От
jian he
Дата:
hi

typedef struct RelOptInfo
{
....
/*
* information about a base rel (not set for join rels!)
*/
Index relid;
...
}

imho, the above comment is not very helpful.
we should say more about what kind of information relid says about a base rel?

I don't know much about RelOptInfo, that's why I ask.



Re: struct RelOptInfo member relid comments

От
Tender Wang
Дата:


jian he <jian.universality@gmail.com> 于2024年5月24日周五 10:58写道:
hi

typedef struct RelOptInfo
{
....
/*
* information about a base rel (not set for join rels!)
*/
Index relid;
...
}

imho, the above comment is not very helpful.
we should say more about what kind of information relid says about a base rel?

I don't know much about RelOptInfo, that's why I ask.


The fields in struct RelOptInfo between comment " information about a base rel " and
commnet "Information about foreign tables and foreign joins" are all about a base rel.

Every field has a comment. I think that's already helpful for understanding what information
we need to optimize a base rel.
--
Tender Wang
OpenPie:  https://en.openpie.com/

Re: struct RelOptInfo member relid comments

От
Tom Lane
Дата:
jian he <jian.universality@gmail.com> writes:
> imho, the above comment is not very helpful.
> we should say more about what kind of information relid says about a base rel?

"Relid" is defined at the very top of the file:

/*
 * Relids
 *        Set of relation identifiers (indexes into the rangetable).
 */
typedef Bitmapset *Relids;

Repeating that everyplace the term "relid" appears would not be
tremendously helpful.

            regards, tom lane



Re: struct RelOptInfo member relid comments

От
jian he
Дата:
On Fri, May 24, 2024 at 11:14 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> jian he <jian.universality@gmail.com> writes:
> > imho, the above comment is not very helpful.
> > we should say more about what kind of information relid says about a base rel?
>
> "Relid" is defined at the very top of the file:
>
> /*
>  * Relids
>  *              Set of relation identifiers (indexes into the rangetable).
>  */
> typedef Bitmapset *Relids;
>
> Repeating that everyplace the term "relid" appears would not be
> tremendously helpful.
>

`Index relid;`
is a relation identifier, a base rel's rangetable index.

Does the above description make sense?


BTW, I found several occurrences of "base+OJ", but I cannot find the
explanation of "OJ" or the "OJ" Acronyms.



Re: struct RelOptInfo member relid comments

От
Ashutosh Bapat
Дата:


On Fri, May 24, 2024 at 9:09 AM jian he <jian.universality@gmail.com> wrote:
On Fri, May 24, 2024 at 11:14 AM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> jian he <jian.universality@gmail.com> writes:
> > imho, the above comment is not very helpful.
> > we should say more about what kind of information relid says about a base rel?
>
> "Relid" is defined at the very top of the file:
>
> /*
>  * Relids
>  *              Set of relation identifiers (indexes into the rangetable).
>  */
> typedef Bitmapset *Relids;
>
> Repeating that everyplace the term "relid" appears would not be
> tremendously helpful.
>

`Index relid;`
is a relation identifier, a base rel's rangetable index.

Does the above description make sense?


BTW, I found several occurrences of "base+OJ", but I cannot find the
explanation of "OJ" or the "OJ" Acronyms.


OJ is an outer join, AFAIU. OJ's have their own relids. If you are wondering why not all joins - I think inner joins need not be tracked as separated relations in parse tree, but OJ's need to be.

--
Best Wishes,
Ashutosh Bapat

Re: struct RelOptInfo member relid comments

От
Tom Lane
Дата:
Ashutosh Bapat <ashutosh.bapat.oss@gmail.com> writes:
> OJ is an outer join, AFAIU. OJ's have their own relids. If you are
> wondering why not all joins - I think inner joins need not be tracked as
> separated relations in parse tree, but OJ's need to be.

An outer join is necessarily associated with explicit JOIN syntax
in the FROM clause, and each such JOIN has its own rangetable entry
and hence a relid.  Inner joins might arise from comma syntax
(that is, "SELECT ... FROM tab1, tab2").  For perhaps-historical
reasons that syntax doesn't give rise to an explicit RTE_JOIN
rangetable entry, so the implied join has no relid.

            regards, tom lane