array_agg and array_accum (patch)

From: Jeff Davis <pgsql(at)j-davis(dot)com>
To: pgsql-hackers(at)postgresql(dot)org
Subject: array_agg and array_accum (patch)
Date: 2008-10-26 18:32:17
Message-ID: 1225045937.4434.21.camel@localhost.localdomain
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Here is a patch to support two aggregate functions:

1) ARRAY_AGG() -- SQL 2008 standard behavior, returns NULL on no input,
and skips NULL inputs.

2) ARRAY_ACCUM() -- Returns empty array on no input, and includes NULL
inputs.

These accumulate the result in a memory context that lives across calls
to the state function, so it's reasonably efficient. On my old laptop it
takes about 5s to generate an array of 1M elements -- not great, but at
least it's linear.

Although array_agg is the standard behavior, array_accum is important
because otherwise you always lose the NULLs, and that's difficult to
work around even with COALESCE.

I added them as new native functions because ARRAY_AGG is in the
standard, but if others think they should live elsewhere that's fine. I
think that they are generally pretty useful functions for people using
arrays.

This patch is contributed by Truviso.

Regards,
Jeff Davis

Attachment Content-Type Size
array_agg.diff.gz application/x-gzip 1.8 KB

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2008-10-26 19:24:57 Re: BufferAccessStrategy for bulk insert
Previous Message Tom Lane 2008-10-26 16:44:06 Re: new correlation metric