Re: [HACKERS] Custom compression methods

On Tue, Mar 16, 2021 at 7:57 PM Robert Haas <robertmhaas(at)gmail(dot)com> wrote:
>

> That behavior feels unacceptable to me from a user expectations point
> of view. I think there's an argument that if I update a tuple that
> contains a compressed datum, and I don't update that particular
> column, I think it would be OK to not recompress the column. But, if I
> insert data into a table, I as a user would expect that the
> compression settings for that column are going to be respected.
> Deciding that's optional because we don't have a good way of making it
> fast seems like a major cop-out, at least to me. I think from a user
> perspective you don't expect INSERT INTO .. SELECT FROM to create a
> different final state than a dump and reload, and that if we deviate
> from that people are gonna be unhappy. I could be wrong; maybe it's
> only me who would be unhappy.

If that is only the argument then it's possible today as well. I mean
you can INSERT INTO .. SELECT FROM where source attribute as
compressed data but the target attribute as external storage then also
we will move the compressed data as it is to the target table.

--
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com