git.net

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: CalcitePrepare inconsistantly applies parser config


I saw CALCITE-1525 <https://issues.apache.org/jira/browse/CALCITE-1525>,
which I think is the issue you are referring to. Apache Beam is currently
using Frameworks, but we've reimplimented much of CalcitePrepare
(parse/convert/executeDdl) on top of it. I'm in the process of adding JDBC
support to Beam and I don't want to duplicate code between JDBC and not
JDBC paths. It seems I can eliminate a bunch of code by just using
CalcitePrepare. I think a few small changes make this possible, I'll send
them your way.

On Mon, May 14, 2018 at 12:35 PM Julian Hyde <jhyde@xxxxxxxxxx> wrote:

> You are correct; we should be more consistent.
>
> I think there’s a JIRA case that we should use the same mechanism for
> preparing queries inside the JDBC driver (i.e. driven by CalcitePrepare) as
> outside a JDBC driver (i.e. using Frameworks). But that would be a lot of
> work. If you can find some small ways to make them more consistent (e.g.
> using the same config) that would be helpful.
>
> Julian
>
>
> > On May 14, 2018, at 11:41 AM, Andrew Pilloud <apilloud@xxxxxxxxxx.INVALID>
> wrote:
> >
> > I've noticed that the CalcitePrepare interface creates a parser in both
> the
> > parse and prepare paths, but only the prepare path applies the custom
> > parser config from the context. Is this intentional? Shouldn't both paths
> > use the config from the context?
> >
> > Andrew
>
>