Re: [HACKERS] varchar() vs char16 performance - Mailing list pgsql-hackers

From darrenk@insightdist.com (Darren King)
Subject Re: [HACKERS] varchar() vs char16 performance
Date
Msg-id 9803111637.AA55078@ceodev
Whole thread Raw
List pgsql-hackers
>
> > Will there be a warning about using a "depreciated type" in 6.4 or are
> > we going to have this gunking up the grammer forever? :)
>
> Good idea. Then we can pull it out of the grammar sometime later. Now,
> if these types are in a loadable module, then we can't actually do
> anything in the parser anyway, since the loadable module would never
> work. Are these character types worth keeping at all? Less support and
> no performance benefit leaves me thinking not...

IMHO, not worth keeping if the performance benefit is gone and the only
real benefit though was the few bytes of header space they saved per field.

They should work as loadable if removed from the parser since they would
be created thru a series of CREATE TYPE, FUNCTION and OPERATOR statements.
Should go thru the parser as just some user-created t[r]ype.

But any benefit from including them in the distribution as loadable would
be negligible.  Could be, but no real point, so I vote to depreciate them.

darrenk

pgsql-hackers by date:

Previous
From: "Thomas G. Lockhart"
Date:
Subject: Re: [HACKERS] varchar() vs char16 performance
Next
From: Hal Snyder
Date:
Subject: Re: [HACKERS] varchar() vs char16 performance