Re: Derby version used by Hive

[Richard Hillegas]

Thanks! Some responses inline...

"kulkarni.swar...@gmail.com" <kulkarni.swar...@gmail.com> wrote on
09/28/2015 10:08:08 AM:

> From: "kulkarni.swar...@gmail.com" <kulkarni.swar...@gmail.com>
> To: "dev@hive.apache.org" <dev@hive.apache.org>
> Date: 09/28/2015 10:08 AM
> Subject: Re: Derby version used by Hive
>
> Richard,
>
> A quick eye-balling of the code doesn't show anything that could
> potentially be a blocker for this upgrade. Also +1 on staying on the
latest
> and greatest. Please feel free to open up a JIRA and submit the patch.

Great! I'll try my hand at this after Derby 10.12.1.1 is published to the
maven repositories next week.

>
> Also just out of curiosity, what are you really using a derby backed
store
> for?

Right now just for testing. Its standards compliance makes it a good
starting point for implementing a portable SQL layer.

Thanks,
-Rick

>
> On Mon, Sep 28, 2015 at 11:02 AM, Richard Hillegas <rhil...@us.ibm.com>
> wrote:
>
> >
> >
> > I haven't received a response to the following message, which I posted
last
> > week. Maybe my message rambled too much. Here is an attempt to pose my
> > question more succinctly:
> >
> > Q: Does anyone know of any reason why we can't upgrade Hive's Derby
version
> > to 10.12.1.1, the new version being vetted by the Derby community right
> > now?
> >
> > Thanks,
> > -Rick
> >
> > > I am following the Hive build instructions here:
> > >
> >
> > https://cwiki.apache.org/confluence/display/Hive/
> GettingStarted#GettingStarted-InstallationandConfiguration
> > > .
> > >
> > > I noticed that Hive development seems to be using an old version of
> > Derby:
> > > 10.10.2.0. Is there some defect in the most recent Derby version
> > > (10.11.1.1) which prevents Hive from upgrading to 10.11.1.1? The only
> > > Hive-tagged Derby bug which I can find is
> > > https://issues.apache.org/jira/browse/DERBY-6358. That issue doesn't
> > seem
> > > to be version-specific and it mentions a resolved Hive issue:
> > > https://issues.apache.org/jira/browse/HIVE-8739.
> > >
> > > Staying with 10.10.2.0 makes sense if you need to run on some ancient
> > JVMs:
> > > Java SE 5 or Java ME CDC/Foundation Profile 1.1. Hadoop, however,
> > requires
> > > at least Java 6 according to
> > > https://wiki.apache.org/hadoop/HadoopJavaVersions.
> > >
> > > Note that the Derby community expects to release version 10.12.1.1
soon:
> > > https://wiki.apache.org/db-derby/DerbyTenTwelveOneRelease. This might
be
> > a
> > > good opportunity for Hive to upgrade to a more capable version of
Derby.
> > >
> > > I mention this because the Derby version used by Hive ends up on the
> > > classpath used by downstream projects (like Spark). That makes it
awkward
> > > for downstream projects to use more current Derby versions. Do you
know
> > of
> > > any reason that downstream projects shouldn't override the Derby
version
> > > currently preferred by Hive?
> > >
> > > Thanks,
> > > -Rick
> >
>
>
>
> --
> Swarnim

Re: Derby version used by Hive

[kulkarni.swar...@gmail.com]

Richard,

A quick eye-balling of the code doesn't show anything that could
potentially be a blocker for this upgrade. Also +1 on staying on the latest
and greatest. Please feel free to open up a JIRA and submit the patch.

Also just out of curiosity, what are you really using a derby backed store
for?

On Mon, Sep 28, 2015 at 11:02 AM, Richard Hillegas 
wrote:

>
>
> I haven't received a response to the following message, which I posted last
> week. Maybe my message rambled too much. Here is an attempt to pose my
> question more succinctly:
>
> Q: Does anyone know of any reason why we can't upgrade Hive's Derby version
> to 10.12.1.1, the new version being vetted by the Derby community right
> now?
>
> Thanks,
> -Rick
>
> > I am following the Hive build instructions here:
> >
>
> https://cwiki.apache.org/confluence/display/Hive/GettingStarted#GettingStarted-InstallationandConfiguration
> > .
> >
> > I noticed that Hive development seems to be using an old version of
> Derby:
> > 10.10.2.0. Is there some defect in the most recent Derby version
> > (10.11.1.1) which prevents Hive from upgrading to 10.11.1.1? The only
> > Hive-tagged Derby bug which I can find is
> > https://issues.apache.org/jira/browse/DERBY-6358. That issue doesn't
> seem
> > to be version-specific and it mentions a resolved Hive issue:
> > https://issues.apache.org/jira/browse/HIVE-8739.
> >
> > Staying with 10.10.2.0 makes sense if you need to run on some ancient
> JVMs:
> > Java SE 5 or Java ME CDC/Foundation Profile 1.1. Hadoop, however,
> requires
> > at least Java 6 according to
> > https://wiki.apache.org/hadoop/HadoopJavaVersions.
> >
> > Note that the Derby community expects to release version 10.12.1.1 soon:
> > https://wiki.apache.org/db-derby/DerbyTenTwelveOneRelease. This might be
> a
> > good opportunity for Hive to upgrade to a more capable version of Derby.
> >
> > I mention this because the Derby version used by Hive ends up on the
> > classpath used by downstream projects (like Spark). That makes it awkward
> > for downstream projects to use more current Derby versions. Do you know
> of
> > any reason that downstream projects shouldn't override the Derby version
> > currently preferred by Hive?
> >
> > Thanks,
> > -Rick
>



-- 
Swarnim

re: Derby version used by Hive

[Richard Hillegas]

I haven't received a response to the following message, which I posted last
week. Maybe my message rambled too much. Here is an attempt to pose my
question more succinctly:

Q: Does anyone know of any reason why we can't upgrade Hive's Derby version
to 10.12.1.1, the new version being vetted by the Derby community right
now?

Thanks,
-Rick

> I am following the Hive build instructions here:
>
https://cwiki.apache.org/confluence/display/Hive/GettingStarted#GettingStarted-InstallationandConfiguration
> .
>
> I noticed that Hive development seems to be using an old version of
Derby:
> 10.10.2.0. Is there some defect in the most recent Derby version
> (10.11.1.1) which prevents Hive from upgrading to 10.11.1.1? The only
> Hive-tagged Derby bug which I can find is
> https://issues.apache.org/jira/browse/DERBY-6358. That issue doesn't seem
> to be version-specific and it mentions a resolved Hive issue:
> https://issues.apache.org/jira/browse/HIVE-8739.
>
> Staying with 10.10.2.0 makes sense if you need to run on some ancient
JVMs:
> Java SE 5 or Java ME CDC/Foundation Profile 1.1. Hadoop, however,
requires
> at least Java 6 according to
> https://wiki.apache.org/hadoop/HadoopJavaVersions.
>
> Note that the Derby community expects to release version 10.12.1.1 soon:
> https://wiki.apache.org/db-derby/DerbyTenTwelveOneRelease. This might be
a
> good opportunity for Hive to upgrade to a more capable version of Derby.
>
> I mention this because the Derby version used by Hive ends up on the
> classpath used by downstream projects (like Spark). That makes it awkward
> for downstream projects to use more current Derby versions. Do you know
of
> any reason that downstream projects shouldn't override the Derby version
> currently preferred by Hive?
>
> Thanks,
> -Rick