Re: [patch] Proposal for \crosstabview in psql

2016-03-11 14:49 GMT+01:00 Robert Haas <robertmhaas(at)gmail(dot)com>:

> On Thu, Feb 18, 2016 at 9:23 AM, Daniel Verite <daniel(at)manitou-mail(dot)org>
> wrote:
> > Dean Rasheed wrote:
> >
> >> If I want to sort the rows coming out of a query, my first thought
> >> is always going to be to add/adjust the query's ORDER BY clause, not
> >> use some weird +/- psql syntax.
> >
> > About the vertical sort, I agree on all your points.
> > It's best to rely on ORDER BY for all the reasons mentioned,
> > as opposed to a separate sort in a second step.
> >
> > But you're considering the case when a user is designing
> > or adapting a query for the purpose of crosstab
> > viewing. As mentioned in my previous reply (about the
> > methods to achieve horizontal sort), that scenario is not really
> > what motivates the feature in the first place.
> >
> > If removing that sort option is required to move forward
> > with the patch because it's controversial, so be it,
> > but overall I don't see this as a benefit for the end user,
> > it's just an option.
>
> Discussion on this patch seems to have died off. I'm probably not
> going to win any popularity contests for saying this, but I think we
> should reject this patch. I don't feel like this is really a psql
> feature: it's a powerful data visualization tool which we're proposing
> to jam into psql. I don't think that's psql's purpose. I also think
> it's quite possible that there could be an unbounded number of
> slightly different things that people want here, and if we take this
> one and a couple more, the code for these individual features could
> come to be larger than all of psql, even though probably 95% of psql
> users would never use any of those.
>

crosstabview is really visualization tool. **But now, there are not any
other tool available from terminal.** So this can be significant help to
all people who would to use this functionality.

The psql has lot of features for 5% users. Currently it is famous not as
"bloated software" but like most comfortable sql console on the world. The
implementation of crosstabview is not complex and with last Daniel's
modification the complexity is less.

The crosstabview is not 100% equal to ANSI SQL PIVOT clause. The ANSI SQL
command is much more rigid (it is one stage statement with predefined
columns), so argument of duplicate implementation one things is not valid.
Probably we would not implement non ANSI SQL feature on server.

Regards

Pavel

>
> Now, that having been said, if other people want this feature to go in
> and are willing to do the work to get it in, I've said my piece and
> won't complain further. There are a couple of committers who have
> taken positive interest in this thread, so that's good. However,
> there are also a couple of committers who have expressed doubts
> similar to mine, so that's not so good. But worse than either of
> those things, there is no real agreement on what the overall design of
> this feature should be. Everybody wants something a little different,
> for different reasons. If we can't come to an agreement, more or less
> immediately, on what to try to get into 9.6, then this can't go into
> this release. Whether it should go into a future release is a
> question we can leave for another time.
>
> --
> Robert Haas
> EnterpriseDB: http://www.enterprisedb.com
> The Enterprise PostgreSQL Company
>