Re: RFC: Restructuring pg_aggregate - Mailing list pgsql-hackers

From Bruce Momjian
Subject Re: RFC: Restructuring pg_aggregate
Date
Msg-id 200204110416.g3B4GNr29322@candle.pha.pa.us
Whole thread Raw
In response to Re: RFC: Restructuring pg_aggregate  (Hiroshi Inoue <Inoue@tpf.co.jp>)
List pgsql-hackers
Hiroshi Inoue wrote:
> If the client has to bear the some part, isn't the invisible
> column approach much simpler ?
> 
> I've put a pretty much time into DROP COLUMN feature but
> I am really disappointed to see the comments in this thread.
> What DROP COLUMN has brought me seems only a waste of time.
> 
> Possibly I must have introduced either implementation forcibly.

I understand.  I personally think maybe we have been a little to picky
about patches being accepted.  Sometimes when something is not 100%
perfect, we do nothing rather than accept the patch, and replace or
improve it later.  The DROP COLUMN approach you had clearly is one of
them.

Personally, now that we have relfilenode, I think we should implement
drop of columns by just recreating the table without the column.

The big problem with DROP COLUMN was that we couldn't decide on what to
do, so we did nothing, which is probably worse than just choosing one
and doing it.

Our code is littered with my 80% solutions for LIKE optimization,
optimizer statistics, BETWEEN, and lots of other solutions that have met
a need and are now being replaced with better code.  My code was not
great, but I hadn't dont them, PostgreSQL would have had even more
missing features than we do now.  DROP COLUMN is clearly one where we
missed getting something that works and would keep people happy.

As far as my proposal, my idea was not to do it in the client, but
rather to do it just before the data is sent from/to the client.  Maybe
that is a stupid idea.  I never really researched it.  My idea was more
to make the physical/logical column numbers distinct so certain tricks
could be performed.  It wasn't for DROP COLUMN specifically, and in fact
to do DROP COLUMN with my code, there would have to be more code similar
to what you had where clients would see a column and have to skip it.  I
was focusing more on physical/logical to enable other features.

--  Bruce Momjian                        |  http://candle.pha.pa.us pgman@candle.pha.pa.us               |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026
 


pgsql-hackers by date:

Previous
From: "Christopher Kings-Lynne"
Date:
Subject: Re: RFC: Restructuring pg_aggregate
Next
From: Bruce Momjian
Date:
Subject: Re: RFC: Restructuring pg_aggregate