Re: JSONB marshalling

Hi All -

  I am not sure that the psycopg2 project is understanding the
strategic and tactical significance of JSONb.
I will briefly use Lucene / SOLR are a (distant) example story, per my
limited perspective:

  Lucene was invented in the 1990s by Doug Cutting, and has proven to
be deeply strong in its text search abilities.
Lucene is designed to run in a single Java Virtual Machine. Compute
power was added as machines and budgets 
grew, and it became clear that Lucene could handle millions of
documents, sometimes large and dense documents.
Among certain users, Lucene use grew, but on the whole, the world
continued on, with many many users looking for other 
solutions.

  Sometime in the 2000's, a user in San Francisco decided that an
internet-ready wrapper would enhance Lucene,
and so the SOLR project was begun. SOLR basically provides GET/PUT and
admin verbs using http.

  This long story made short, SOLR is now arguably the "center of
gravity" for Lucene. Lucene is a remarkable 
engine, and now stable, but new features, new users, new advances are
almost entirely in terms of SOLR, 
not Lucene.  What does this mean to psycopg2 and PostgreSQL ?

  PostgreSQL is a core engine like Lucene, and shares many properties
described above. However, the internet
world of information exchange looked past Postgres to invent REST/http
based stores because the way the 
data moves is as important as the engine, in human terms. When Heroku
intervened and guided the JSONB
implemenetation in Postgres, they knew that JSONB in core Postgres
enables interoperability .. in this story,
like SOLR.

  I submit that JSONB, and following, Psycopg2 with JSONB support,
will grow and grow. I say to the psycopg2
project, this is not a point-point release update with a sidenote, but
rather a very strategic feature that ought to 
be fully support, documented and presented to all psycopg2 developers.
 I will leave the current status of
psycopg2 in the python stack to be described by others that are more
capable of that than I.

  agree, disagree or indifferent, psycopg2 would be missing a major
opportunity if the project does not 
fully embrace and present JSONB in PostgreSQL.

  best regards from Berkeley, California
  Brian M Hamlin
  OSGeo California Chapter

On Fri, 25 Jul 2014 18:01:15 +0200, Guido Notari <gnotari(at)linkgroup(dot)it> wrote:
Il giorno 24/lug/2014, alle ore 23:05, Daniele Varrazzo
<daniele(dot)varrazzo(at)gmail(dot)com> ha scritto:
>
> > When would this happen? I don't think we can say it's a bugfix and
> > it's a change in the adapter behaviour so I think it should be
> > released in a future 2.6, which is still unplanned and I don't see
> > many new features to release. OTOH PG 9.4 has not been released yet
> > and the jsonb oids have never been used in the wild (user-defined
> > types get bigger oids, right?) so if there was interest in it we could
> > consider its release in 2.5.4, which would be released in a much
> > shorter time.
>
> I second the proposal about a 2.5.4 mini-release, with this feature
> as it’s possibly sole content.
> This will permit easy experimentation of the "new" types pretty soon,
> and avoid burdening the maintainer(s) with long feature lists in the
> future.
> Anyway, not sooner than full 9.4 release, as you said — no reason
> to rush, really.
>
> Thanks as always
> Guido
>
> -- Sent via psycopg mailing list (psycopg(at)postgresql(dot)org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/psycopg
>
>