Simulataneous changes to Avatica and Calcite

[Mihai Budiu <mb...@gmail.com>]

Hello,

I have submitted a fix to Avatica improving the output of times and timestamps with precision, for this issue:
https://issues.apache.org/jira/browse/CALCITE-6282 (Avatica ignores time precision when returning TIME results)

However, fixing this issue makes several Calcite tests fail, including some Quidem tests.
I can submit a PR on the Calcite side, but the two PRs have to be merged at the same time.
I have seen conditional code in Java based on the avatica version, but I don't know what to do about Quidem tests.
Any suggestions on what I should do?

Thank you,
Mihai

Re: Simulataneous changes to Avatica and Calcite

[Julian Hyde <jh...@apache.org>]

In https://github.com/apache/calcite/commit/614b4350f558a4fabaafc06fef9689ec4c267994
you'll see the Stamatis upgraded Avatica and removed the "!if
(fixed.calcite2776) {" wrapper in a Quidem test. That is a reasonably
clean way to comment out tests pending a bug fix or an upgrade.

Calcite release 1.37 is currently being discussed. I guess you'll want
to make an Avatica release first. If so, you'll need an Avatica RM,
and a Jira case for each release (one depending on the other).

Julian

On Mon, Feb 26, 2024 at 6:44 PM Mihai Budiu <mb...@gmail.com> wrote:
>
> You can see some failures here
>
> https://github.com/apache/calcite-avatica/actions/runs/8058235539/job/22010704583
>
> Some of them are very strange, but that's a separate point.
> ________________________________
> From: Tanner Clary <ta...@google.com.INVALID>
> Sent: Monday, February 26, 2024 6:29:38 PM
> To: dev@calcite.apache.org <de...@calcite.apache.org>
> Subject: Re: Simulataneous changes to Avatica and Calcite
>
> How do the tests fail? You could use deprecated methods? You could also
> block the quidem tests with a TODO for the JIRA (is that hacky?)
>
> On Mon, Feb 26, 2024 at 5:58 PM Mihai Budiu <mb...@gmail.com> wrote:
>
> > Hello,
> >
> > I have submitted a fix to Avatica improving the output of times and
> > timestamps with precision, for this issue:
> > https://issues.apache.org/jira/browse/CALCITE-6282 (Avatica ignores time
> > precision when returning TIME results)
> >
> > However, fixing this issue makes several Calcite tests fail, including
> > some Quidem tests.
> > I can submit a PR on the Calcite side, but the two PRs have to be merged
> > at the same time.
> > I have seen conditional code in Java based on the avatica version, but I
> > don't know what to do about Quidem tests.
> > Any suggestions on what I should do?
> >
> > Thank you,
> > Mihai
> >

Re: Simulataneous changes to Avatica and Calcite

[Mihai Budiu <mb...@gmail.com>]

You can see some failures here

https://github.com/apache/calcite-avatica/actions/runs/8058235539/job/22010704583

Some of them are very strange, but that's a separate point.
________________________________
From: Tanner Clary <ta...@google.com.INVALID>
Sent: Monday, February 26, 2024 6:29:38 PM
To: dev@calcite.apache.org <de...@calcite.apache.org>
Subject: Re: Simulataneous changes to Avatica and Calcite

How do the tests fail? You could use deprecated methods? You could also
block the quidem tests with a TODO for the JIRA (is that hacky?)

On Mon, Feb 26, 2024 at 5:58 PM Mihai Budiu <mb...@gmail.com> wrote:

> Hello,
>
> I have submitted a fix to Avatica improving the output of times and
> timestamps with precision, for this issue:
> https://issues.apache.org/jira/browse/CALCITE-6282 (Avatica ignores time
> precision when returning TIME results)
>
> However, fixing this issue makes several Calcite tests fail, including
> some Quidem tests.
> I can submit a PR on the Calcite side, but the two PRs have to be merged
> at the same time.
> I have seen conditional code in Java based on the avatica version, but I
> don't know what to do about Quidem tests.
> Any suggestions on what I should do?
>
> Thank you,
> Mihai
>

Re: Simulataneous changes to Avatica and Calcite

[Mihai Budiu <mb...@gmail.com>]

The expected test output changes.
But we don't know exactly when calcite moves to the next avatica version, so I shouldn't disable the quidem tests until the last moment...
________________________________
From: Tanner Clary <ta...@google.com.INVALID>
Sent: Monday, February 26, 2024 6:29:38 PM
To: dev@calcite.apache.org <de...@calcite.apache.org>
Subject: Re: Simulataneous changes to Avatica and Calcite

How do the tests fail? You could use deprecated methods? You could also
block the quidem tests with a TODO for the JIRA (is that hacky?)

On Mon, Feb 26, 2024 at 5:58 PM Mihai Budiu <mb...@gmail.com> wrote:

> Hello,
>
> I have submitted a fix to Avatica improving the output of times and
> timestamps with precision, for this issue:
> https://issues.apache.org/jira/browse/CALCITE-6282 (Avatica ignores time
> precision when returning TIME results)
>
> However, fixing this issue makes several Calcite tests fail, including
> some Quidem tests.
> I can submit a PR on the Calcite side, but the two PRs have to be merged
> at the same time.
> I have seen conditional code in Java based on the avatica version, but I
> don't know what to do about Quidem tests.
> Any suggestions on what I should do?
>
> Thank you,
> Mihai
>

Re: Simulataneous changes to Avatica and Calcite

[Tanner Clary <ta...@google.com.INVALID>]

How do the tests fail? You could use deprecated methods? You could also
block the quidem tests with a TODO for the JIRA (is that hacky?)

On Mon, Feb 26, 2024 at 5:58 PM Mihai Budiu <mb...@gmail.com> wrote:

> Hello,
>
> I have submitted a fix to Avatica improving the output of times and
> timestamps with precision, for this issue:
> https://issues.apache.org/jira/browse/CALCITE-6282 (Avatica ignores time
> precision when returning TIME results)
>
> However, fixing this issue makes several Calcite tests fail, including
> some Quidem tests.
> I can submit a PR on the Calcite side, but the two PRs have to be merged
> at the same time.
> I have seen conditional code in Java based on the avatica version, but I
> don't know what to do about Quidem tests.
> Any suggestions on what I should do?
>
> Thank you,
> Mihai
>