Re: Syntax for partitioning - Mailing list pgsql-hackers

From Itagaki Takahiro
Subject Re: Syntax for partitioning
Date
Msg-id 20091118135257.A4AD.52131E4D@oss.ntt.co.jp
Whole thread Raw
In response to Re: Syntax for partitioning  (Marko Tiikkaja <marko.tiikkaja@cs.helsinki.fi>)
Responses Re: Syntax for partitioning
List pgsql-hackers
Marko Tiikkaja <marko.tiikkaja@cs.helsinki.fi> wrote:

> this looks like a mistake:
> partinfo = (PartitionInfo *) malloc(ntups * sizeof(PartitionInfo));

Oops, it should be "p"alloc. Thanks.

> Maybe we should use something like
> PARTITION bar VALUES OPERATOR 0
> when the user specifies the operator?

I think we could have reasonable restrictions to the operator
for future optimization. Is the VALUES OPERATOR syntax too freedom?

For the same reason, USING operator also might be too freedom.
RANGE (and maybe also LIST) partition keys should be sortable,
operator class name might be better to the option instead of
any operators. i.e., PARTITION BY RANGE ( foo [ USING operator ] )
should be:  PARTITION BY RANGE ( foo [ btree_ops_name ] )

If we do so, there will be no inconsistency in LESS THAN syntax
because btree_ops always have < operator.

Regards,
---
ITAGAKI Takahiro
NTT Open Source Software Center




pgsql-hackers by date:

Previous
From: Itagaki Takahiro
Date:
Subject: Re: Syntax for partitioning
Next
From: "Joshua D. Drake"
Date:
Subject: Re: plpgsql: open for execute - add USING clause