Re: WIP patch for basic window frame support - Mailing list pgsql-hackers

From Hitoshi Harada
Subject Re: WIP patch for basic window frame support
Date
Msg-id e08cc0400901041616t279627c4u85ded78d4b6d8efc@mail.gmail.com
Whole thread Raw
In response to WIP patch for basic window frame support  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: WIP patch for basic window frame support  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
2008/12/31 Tom Lane <tgl@sss.pgh.pa.us>:
> No docs or regression tests yet, but it seems to work ... please check
> against Oracle and DB2 behavior.  Supported cases are
>
>        RANGE UNBOUNDED PRECEDING       -- same as AND CURRENT ROW
>        ROWS UNBOUNDED PRECEDING        -- same as AND CURRENT ROW

Is this true?
I guess that the 7.11 rule 5.b in the spec says as far as the bound is
not specified in the window frame clause, all rows of the partition
are contained in the frame. The rule then removes rows from the
initial frame as the frame bound indicates.

So as the result,
RANGE UNBOUNDED PRECEDING
ROWS UNBOUNDED PRECEDING
both mean
(RANGE / ROWS) BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING
if I don't miss something.

Regards,


-- 
Hitoshi Harada


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: generic reloptions improvement
Next
From: "Hitoshi Harada"
Date:
Subject: Re: WIP patch for basic window frame support