You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@orc.apache.org by Dongjoon Hyun <do...@gmail.com> on 2021/10/17 03:56:44 UTC

[DISCUSS] Apache ORC Release Cadence

Hi, All.

There were some questions and discussion about Apache ORC release cadence
at our GitHub PR. I want to share it with you and get your opinion on it.

https://github.com/apache/orc/pull/936#issuecomment-944017204

According to our release page, Apache ORC release cadence requires
one or two years for the next minor version.

https://orc.apache.org/docs/releases.html
    - 1.4.0: 2017-05-08
    - 1.5.0: 2018-05-14 (About 1 year and 6 days)
    - 1.6.0: 2019-09-03 (About 1 year and 4 month)
    - 1.7.0: 2021-09-15 (About 2 years and 12 days)
    - 1.8.0: 2023-09-15 (About 2 years?)

Apache ORC release cadence is also related to the EOL of the old releases.
For example,
    - 1.5.x was maintained for 40 months and was archived on 2021-09-15.
    - 1.6.x EOL is expected on 2022-09-03 if we maintain it for 3 years.
    - 1.7.x EOL is expected on 2024-09-15 if we maintain it for 3 years.

What is your expectation on Apache ORC release cadence?

Dongjoon

Re: [DISCUSS] Apache ORC Release Cadence

Posted by Dongjoon Hyun <do...@apache.org>.
According to the discussion, we seem to achieve two things.

1. Keep long term support of released versions as a library
2. Deliver new features more swiftly (if possible)

Although we are going to follow semantic versioning, 
we may want to consider backporting differently per modules
to achieve both goals in some way. 

For example, `tool` and `benchmark` module are relatively safer
in terms of backporting. And, they are used as a final CLI instead of library.

The following is one of the examples.

ORC-1012: Support specifying columns in orc-scan

Dongjoon.

On 2021/10/29 14:26:53, Dongjoon Hyun <do...@gmail.com> wrote: 
> Thank you, William. Yes, it does indeed.
> 
> BTW, I'd like to cut 1.7.1 and 1.6.12 next week. I'll start another thread
> for that.
> 
> Dongjoon.
> 
> 
> On Thu, Oct 28, 2021 at 11:09 PM William Hyun <wi...@apache.org> wrote:
> 
> > Hello All!
> >
> > I believe that the feature release for ORC is slowing down because of the
> > lack of new features.
> > If we have meaningful new features, we can strive for annual releases.
> > However, Apache Hive and Flink still use 1.5.x (EOL) versions.
> > We had better help them migrate to 1.7.x first before 1.8.0.
> >
> > Bests,
> > William
> >
> > On 2021/10/26 05:36:41, Dongjoon Hyun <do...@gmail.com> wrote:
> > > Thank you, Yiqun.
> > >
> > > Yes, it's `alpha` or `preview` releases are also possible for the
> > > standalone customers.
> > >
> > > One problem might be that Apache ORC is a library whose majority of
> > > customers consume it
> > > usually via the Apache downstream
> > > project(Spark/Hive/Flink/Iceberg/Arrow/Druid).
> > > In this case, it's hard to reach out to the customer because
> > > `alpha` version is not acceptable for downstream PRs.
> > >
> > > Thank you again for your opinions, Yiqun and Pavan.
> > >
> > >
> > > On Thu, Oct 21, 2021 at 8:20 PM Yiqun Zhang <gu...@gmail.com>
> > wrote:
> > >
> > > > Thanks Dongjoon for initiating this discussion.
> > > >
> > > > I think the current release cadence is great, 1-2 years for a major
> > release
> > > > and we maintain each major release for about 3 years so we can make
> > sure we
> > > > maintain a consistent number of releases.
> > > >
> > > > https://github.com/apache/orc/pull/936#issuecomment-944017204
> > > > My opinion in response to this discussion is that we could add preview
> > > > releases, such as 1.8.0-alpha, which could be released at a cadence
> > that
> > > > could be adjusted as incompatible new features accumulate. To avoid new
> > > > interfaces taking too long to come out, it is also convenient for
> > users to
> > > > understand and test in advance, and we can give lower priority to
> > fixing
> > > > its issues than the major version.
> > > >
> > > > Regards,
> > > > Yiqun Zhang
> > > >
> > > > Dongjoon Hyun <do...@gmail.com> 于2021年10月17日周日 上午11:57写道:
> > > >
> > > > > Hi, All.
> > > > >
> > > > > There were some questions and discussion about Apache ORC release
> > cadence
> > > > > at our GitHub PR. I want to share it with you and get your opinion
> > on it.
> > > > >
> > > > > https://github.com/apache/orc/pull/936#issuecomment-944017204
> > > > >
> > > > > According to our release page, Apache ORC release cadence requires
> > > > > one or two years for the next minor version.
> > > > >
> > > > > https://orc.apache.org/docs/releases.html
> > > > >     - 1.4.0: 2017-05-08
> > > > >     - 1.5.0: 2018-05-14 (About 1 year and 6 days)
> > > > >     - 1.6.0: 2019-09-03 (About 1 year and 4 month)
> > > > >     - 1.7.0: 2021-09-15 (About 2 years and 12 days)
> > > > >     - 1.8.0: 2023-09-15 (About 2 years?)
> > > > >
> > > > > Apache ORC release cadence is also related to the EOL of the old
> > > > releases.
> > > > > For example,
> > > > >     - 1.5.x was maintained for 40 months and was archived on
> > 2021-09-15.
> > > > >     - 1.6.x EOL is expected on 2022-09-03 if we maintain it for 3
> > years.
> > > > >     - 1.7.x EOL is expected on 2024-09-15 if we maintain it for 3
> > years.
> > > > >
> > > > > What is your expectation on Apache ORC release cadence?
> > > > >
> > > > > Dongjoon
> > > > >
> > > >
> > >
> >
> 

Re: [DISCUSS] Apache ORC Release Cadence

Posted by Dongjoon Hyun <do...@gmail.com>.
Thank you, William. Yes, it does indeed.

BTW, I'd like to cut 1.7.1 and 1.6.12 next week. I'll start another thread
for that.

Dongjoon.


On Thu, Oct 28, 2021 at 11:09 PM William Hyun <wi...@apache.org> wrote:

> Hello All!
>
> I believe that the feature release for ORC is slowing down because of the
> lack of new features.
> If we have meaningful new features, we can strive for annual releases.
> However, Apache Hive and Flink still use 1.5.x (EOL) versions.
> We had better help them migrate to 1.7.x first before 1.8.0.
>
> Bests,
> William
>
> On 2021/10/26 05:36:41, Dongjoon Hyun <do...@gmail.com> wrote:
> > Thank you, Yiqun.
> >
> > Yes, it's `alpha` or `preview` releases are also possible for the
> > standalone customers.
> >
> > One problem might be that Apache ORC is a library whose majority of
> > customers consume it
> > usually via the Apache downstream
> > project(Spark/Hive/Flink/Iceberg/Arrow/Druid).
> > In this case, it's hard to reach out to the customer because
> > `alpha` version is not acceptable for downstream PRs.
> >
> > Thank you again for your opinions, Yiqun and Pavan.
> >
> >
> > On Thu, Oct 21, 2021 at 8:20 PM Yiqun Zhang <gu...@gmail.com>
> wrote:
> >
> > > Thanks Dongjoon for initiating this discussion.
> > >
> > > I think the current release cadence is great, 1-2 years for a major
> release
> > > and we maintain each major release for about 3 years so we can make
> sure we
> > > maintain a consistent number of releases.
> > >
> > > https://github.com/apache/orc/pull/936#issuecomment-944017204
> > > My opinion in response to this discussion is that we could add preview
> > > releases, such as 1.8.0-alpha, which could be released at a cadence
> that
> > > could be adjusted as incompatible new features accumulate. To avoid new
> > > interfaces taking too long to come out, it is also convenient for
> users to
> > > understand and test in advance, and we can give lower priority to
> fixing
> > > its issues than the major version.
> > >
> > > Regards,
> > > Yiqun Zhang
> > >
> > > Dongjoon Hyun <do...@gmail.com> 于2021年10月17日周日 上午11:57写道:
> > >
> > > > Hi, All.
> > > >
> > > > There were some questions and discussion about Apache ORC release
> cadence
> > > > at our GitHub PR. I want to share it with you and get your opinion
> on it.
> > > >
> > > > https://github.com/apache/orc/pull/936#issuecomment-944017204
> > > >
> > > > According to our release page, Apache ORC release cadence requires
> > > > one or two years for the next minor version.
> > > >
> > > > https://orc.apache.org/docs/releases.html
> > > >     - 1.4.0: 2017-05-08
> > > >     - 1.5.0: 2018-05-14 (About 1 year and 6 days)
> > > >     - 1.6.0: 2019-09-03 (About 1 year and 4 month)
> > > >     - 1.7.0: 2021-09-15 (About 2 years and 12 days)
> > > >     - 1.8.0: 2023-09-15 (About 2 years?)
> > > >
> > > > Apache ORC release cadence is also related to the EOL of the old
> > > releases.
> > > > For example,
> > > >     - 1.5.x was maintained for 40 months and was archived on
> 2021-09-15.
> > > >     - 1.6.x EOL is expected on 2022-09-03 if we maintain it for 3
> years.
> > > >     - 1.7.x EOL is expected on 2024-09-15 if we maintain it for 3
> years.
> > > >
> > > > What is your expectation on Apache ORC release cadence?
> > > >
> > > > Dongjoon
> > > >
> > >
> >
>

Re: [DISCUSS] Apache ORC Release Cadence

Posted by William Hyun <wi...@apache.org>.
Hello All!

I believe that the feature release for ORC is slowing down because of the lack of new features. 
If we have meaningful new features, we can strive for annual releases. 
However, Apache Hive and Flink still use 1.5.x (EOL) versions. 
We had better help them migrate to 1.7.x first before 1.8.0.

Bests,
William

On 2021/10/26 05:36:41, Dongjoon Hyun <do...@gmail.com> wrote: 
> Thank you, Yiqun.
> 
> Yes, it's `alpha` or `preview` releases are also possible for the
> standalone customers.
> 
> One problem might be that Apache ORC is a library whose majority of
> customers consume it
> usually via the Apache downstream
> project(Spark/Hive/Flink/Iceberg/Arrow/Druid).
> In this case, it's hard to reach out to the customer because
> `alpha` version is not acceptable for downstream PRs.
> 
> Thank you again for your opinions, Yiqun and Pavan.
> 
> 
> On Thu, Oct 21, 2021 at 8:20 PM Yiqun Zhang <gu...@gmail.com> wrote:
> 
> > Thanks Dongjoon for initiating this discussion.
> >
> > I think the current release cadence is great, 1-2 years for a major release
> > and we maintain each major release for about 3 years so we can make sure we
> > maintain a consistent number of releases.
> >
> > https://github.com/apache/orc/pull/936#issuecomment-944017204
> > My opinion in response to this discussion is that we could add preview
> > releases, such as 1.8.0-alpha, which could be released at a cadence that
> > could be adjusted as incompatible new features accumulate. To avoid new
> > interfaces taking too long to come out, it is also convenient for users to
> > understand and test in advance, and we can give lower priority to fixing
> > its issues than the major version.
> >
> > Regards,
> > Yiqun Zhang
> >
> > Dongjoon Hyun <do...@gmail.com> 于2021年10月17日周日 上午11:57写道:
> >
> > > Hi, All.
> > >
> > > There were some questions and discussion about Apache ORC release cadence
> > > at our GitHub PR. I want to share it with you and get your opinion on it.
> > >
> > > https://github.com/apache/orc/pull/936#issuecomment-944017204
> > >
> > > According to our release page, Apache ORC release cadence requires
> > > one or two years for the next minor version.
> > >
> > > https://orc.apache.org/docs/releases.html
> > >     - 1.4.0: 2017-05-08
> > >     - 1.5.0: 2018-05-14 (About 1 year and 6 days)
> > >     - 1.6.0: 2019-09-03 (About 1 year and 4 month)
> > >     - 1.7.0: 2021-09-15 (About 2 years and 12 days)
> > >     - 1.8.0: 2023-09-15 (About 2 years?)
> > >
> > > Apache ORC release cadence is also related to the EOL of the old
> > releases.
> > > For example,
> > >     - 1.5.x was maintained for 40 months and was archived on 2021-09-15.
> > >     - 1.6.x EOL is expected on 2022-09-03 if we maintain it for 3 years.
> > >     - 1.7.x EOL is expected on 2024-09-15 if we maintain it for 3 years.
> > >
> > > What is your expectation on Apache ORC release cadence?
> > >
> > > Dongjoon
> > >
> >
> 

Re: [DISCUSS] Apache ORC Release Cadence

Posted by Dongjoon Hyun <do...@gmail.com>.
Thank you, Yiqun.

Yes, it's `alpha` or `preview` releases are also possible for the
standalone customers.

One problem might be that Apache ORC is a library whose majority of
customers consume it
usually via the Apache downstream
project(Spark/Hive/Flink/Iceberg/Arrow/Druid).
In this case, it's hard to reach out to the customer because
`alpha` version is not acceptable for downstream PRs.

Thank you again for your opinions, Yiqun and Pavan.


On Thu, Oct 21, 2021 at 8:20 PM Yiqun Zhang <gu...@gmail.com> wrote:

> Thanks Dongjoon for initiating this discussion.
>
> I think the current release cadence is great, 1-2 years for a major release
> and we maintain each major release for about 3 years so we can make sure we
> maintain a consistent number of releases.
>
> https://github.com/apache/orc/pull/936#issuecomment-944017204
> My opinion in response to this discussion is that we could add preview
> releases, such as 1.8.0-alpha, which could be released at a cadence that
> could be adjusted as incompatible new features accumulate. To avoid new
> interfaces taking too long to come out, it is also convenient for users to
> understand and test in advance, and we can give lower priority to fixing
> its issues than the major version.
>
> Regards,
> Yiqun Zhang
>
> Dongjoon Hyun <do...@gmail.com> 于2021年10月17日周日 上午11:57写道:
>
> > Hi, All.
> >
> > There were some questions and discussion about Apache ORC release cadence
> > at our GitHub PR. I want to share it with you and get your opinion on it.
> >
> > https://github.com/apache/orc/pull/936#issuecomment-944017204
> >
> > According to our release page, Apache ORC release cadence requires
> > one or two years for the next minor version.
> >
> > https://orc.apache.org/docs/releases.html
> >     - 1.4.0: 2017-05-08
> >     - 1.5.0: 2018-05-14 (About 1 year and 6 days)
> >     - 1.6.0: 2019-09-03 (About 1 year and 4 month)
> >     - 1.7.0: 2021-09-15 (About 2 years and 12 days)
> >     - 1.8.0: 2023-09-15 (About 2 years?)
> >
> > Apache ORC release cadence is also related to the EOL of the old
> releases.
> > For example,
> >     - 1.5.x was maintained for 40 months and was archived on 2021-09-15.
> >     - 1.6.x EOL is expected on 2022-09-03 if we maintain it for 3 years.
> >     - 1.7.x EOL is expected on 2024-09-15 if we maintain it for 3 years.
> >
> > What is your expectation on Apache ORC release cadence?
> >
> > Dongjoon
> >
>

Re: [DISCUSS] Apache ORC Release Cadence

Posted by Yiqun Zhang <gu...@gmail.com>.
Thanks Dongjoon for initiating this discussion.

I think the current release cadence is great, 1-2 years for a major release
and we maintain each major release for about 3 years so we can make sure we
maintain a consistent number of releases.

https://github.com/apache/orc/pull/936#issuecomment-944017204
My opinion in response to this discussion is that we could add preview
releases, such as 1.8.0-alpha, which could be released at a cadence that
could be adjusted as incompatible new features accumulate. To avoid new
interfaces taking too long to come out, it is also convenient for users to
understand and test in advance, and we can give lower priority to fixing
its issues than the major version.

Regards,
Yiqun Zhang

Dongjoon Hyun <do...@gmail.com> 于2021年10月17日周日 上午11:57写道:

> Hi, All.
>
> There were some questions and discussion about Apache ORC release cadence
> at our GitHub PR. I want to share it with you and get your opinion on it.
>
> https://github.com/apache/orc/pull/936#issuecomment-944017204
>
> According to our release page, Apache ORC release cadence requires
> one or two years for the next minor version.
>
> https://orc.apache.org/docs/releases.html
>     - 1.4.0: 2017-05-08
>     - 1.5.0: 2018-05-14 (About 1 year and 6 days)
>     - 1.6.0: 2019-09-03 (About 1 year and 4 month)
>     - 1.7.0: 2021-09-15 (About 2 years and 12 days)
>     - 1.8.0: 2023-09-15 (About 2 years?)
>
> Apache ORC release cadence is also related to the EOL of the old releases.
> For example,
>     - 1.5.x was maintained for 40 months and was archived on 2021-09-15.
>     - 1.6.x EOL is expected on 2022-09-03 if we maintain it for 3 years.
>     - 1.7.x EOL is expected on 2024-09-15 if we maintain it for 3 years.
>
> What is your expectation on Apache ORC release cadence?
>
> Dongjoon
>

Re: [DISCUSS] Apache ORC Release Cadence

Posted by Dongjoon Hyun <do...@gmail.com>.
That's a good question. I asked the same question in April by myself. :)

https://lists.apache.org/thread.html/r84e4e575dd7414921aadee1df116e0e900fb1e000919e921a0d08993%40%3Cdev.orc.apache.org%3E

Apache ORC will follow `Semantic Versioning` like many other Apache
projects.

    https://semver.org (Semantic Versioning 2.0.0)

Given a version number MAJOR.MINOR.PATCH, increment the:

    MAJOR version when you make incompatible API changes,
    MINOR version when you add functionality in a backwards compatible
manner, and
    PATCH version when you make backwards compatible bug fixes.

For example, 1.7.1 is a patch version of 1.7.0 which can have only
backward-compatible bug fixes. 1.8.x can have new backward-compatible
features.

Dongjoon.

On Thu, Oct 21, 2021 at 9:53 AM Pavan Lanka <pl...@apple.com.invalid>
wrote:

> Thanks Dongjoon for initiating this.
>
> I wanted to include the release policy into this discussion as that will
> influence what kind of cadence we might need.
>
> I would like to understand the differentiation between a Patch Release
> X.Y.Z and a Minor Release X.Y.
> Are there any scenarios where a minor enhancement might be included into a
> patch release instead of waiting for a minor release?
>
> Regards,
> Pavan
>
> > On Oct 16, 2021, at 8:56 PM, Dongjoon Hyun <do...@gmail.com>
> wrote:
> >
> > Hi, All.
> >
> > There were some questions and discussion about Apache ORC release cadence
> > at our GitHub PR. I want to share it with you and get your opinion on it.
> >
> > https://github.com/apache/orc/pull/936#issuecomment-944017204
> >
> > According to our release page, Apache ORC release cadence requires
> > one or two years for the next minor version.
> >
> > https://orc.apache.org/docs/releases.html
> >    - 1.4.0: 2017-05-08
> >    - 1.5.0: 2018-05-14 (About 1 year and 6 days)
> >    - 1.6.0: 2019-09-03 (About 1 year and 4 month)
> >    - 1.7.0: 2021-09-15 (About 2 years and 12 days)
> >    - 1.8.0: 2023-09-15 (About 2 years?)
> >
> > Apache ORC release cadence is also related to the EOL of the old
> releases.
> > For example,
> >    - 1.5.x was maintained for 40 months and was archived on 2021-09-15.
> >    - 1.6.x EOL is expected on 2022-09-03 if we maintain it for 3 years.
> >    - 1.7.x EOL is expected on 2024-09-15 if we maintain it for 3 years.
> >
> > What is your expectation on Apache ORC release cadence?
> >
> > Dongjoon
>
>

Re: [DISCUSS] Apache ORC Release Cadence

Posted by Pavan Lanka <pl...@apple.com.INVALID>.
Thanks Dongjoon for initiating this.

I wanted to include the release policy into this discussion as that will influence what kind of cadence we might need.

I would like to understand the differentiation between a Patch Release X.Y.Z and a Minor Release X.Y.
Are there any scenarios where a minor enhancement might be included into a patch release instead of waiting for a minor release?

Regards,
Pavan

> On Oct 16, 2021, at 8:56 PM, Dongjoon Hyun <do...@gmail.com> wrote:
> 
> Hi, All.
> 
> There were some questions and discussion about Apache ORC release cadence
> at our GitHub PR. I want to share it with you and get your opinion on it.
> 
> https://github.com/apache/orc/pull/936#issuecomment-944017204
> 
> According to our release page, Apache ORC release cadence requires
> one or two years for the next minor version.
> 
> https://orc.apache.org/docs/releases.html
>    - 1.4.0: 2017-05-08
>    - 1.5.0: 2018-05-14 (About 1 year and 6 days)
>    - 1.6.0: 2019-09-03 (About 1 year and 4 month)
>    - 1.7.0: 2021-09-15 (About 2 years and 12 days)
>    - 1.8.0: 2023-09-15 (About 2 years?)
> 
> Apache ORC release cadence is also related to the EOL of the old releases.
> For example,
>    - 1.5.x was maintained for 40 months and was archived on 2021-09-15.
>    - 1.6.x EOL is expected on 2022-09-03 if we maintain it for 3 years.
>    - 1.7.x EOL is expected on 2024-09-15 if we maintain it for 3 years.
> 
> What is your expectation on Apache ORC release cadence?
> 
> Dongjoon