Re: binary representation of datatypes

From: "Merlin Moncure" <mmoncure(at)gmail(dot)com>
To: "Andrew Chernow" <ac(at)esilo(dot)com>
Cc: "Matthieu Imbert" <matthieu(dot)imbert(at)ens-lyon(dot)fr>, "Jeroen Vermeulen" <jtv(at)xs4all(dot)nl>, pgsql-hackers(at)postgresql(dot)org
Subject: Re: binary representation of datatypes
Date: 2008-10-22 12:18:17
Message-ID: b42b73150810220518vb1f02eeh1f696bf60e35e935@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Wed, Oct 22, 2008 at 8:07 AM, Andrew Chernow <ac(at)esilo(dot)com> wrote:
>>> You mean that when results are asked in textual representation (the
>>> default), data is sent on network directly as text?
>>
>> You should know that text/binary conversions rarely play a significant
>> role in terms of performance. There are exceptions...large bytea
>> columns, or enormous sets of integers. This is coming from a guy that
>> co-wrote a library that allows you to pull data directly in binary.
>>
>> merlin
>>
>
> If I remember correctly, composites and composite arrays also show worth
> while performance gains. libpq array and composite handling is what
> initially spawned the libpqtypes project (which required providing type
> handling for every basic type like int and text). So, different types were
> implemented for different reasons, it was not all performance. The ultimate
> functionality we were looking for was multiple result sets, which composite
> arrays solve nicely.

sure. That isn't, strictly speaking, a performance argument...it's
also a convenience thing.
You won't see a difference either way unless the arrays are large, or
a lot of them (big result sets). For smaller result sets, the
overhead of executing the query is where all the time is spent.

merlin

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2008-10-22 12:31:43 Re: Bitmap Indexes: request for feedback
Previous Message Andrew Chernow 2008-10-22 12:07:36 Re: binary representation of datatypes