Agreed, I wouldn't be opposed to looking into approaches to make release
times more predictable. At the same time, I'd advocate that in the
community, that anyone can propose a release at any point in time. Of
course, we can discuss as a community and make sure there's a reasonable
changeset, as well as focus review time on PRs which are close to being
ready for that release.
To some degree this contradicts having a predictable release schedule, but
I feel like we can really just have a hybrid "Periodic release + arbitrary
off-cycle release" approach and things won't get too crazy. It's a way to
get the best of both frequency of release and user expectations on release
times.

An update on 1.8 to the community, we're working on updating LICENSE/NOTICE
files in the AWS/GCP/Azure bundles
<https://github.com/apache/iceberg/pull/12095/>, thank you JB for driving
that. It's something we need to get in for the release. Once that's in, I
will cut the RC.

Thanks,

Amogh Jahagirdar

On Sun, Jan 26, 2025 at 1:16 PM Jean-Baptiste Onofré <j...@nanthrax.net>
wrote:

> Hi Amogh,
>
> Thanks !
>
> I agree we should have more frequent releases, but also more
> "predictable" release time and give visibility to the community
> (especially users).
> Some ASF projects are providing "tables" with release plans:
> - https://camel.apache.org/download/
> - https://karaf.apache.org/download.html
> - https://activemq.apache.org/components/classic/download/
> - ...
>
> Maybe we can provide something similar ?
>
> Thanks !
> Regards
> JB
>
> On Sat, Jan 25, 2025 at 1:07 AM Amogh Jahagirdar <2am...@gmail.com> wrote:
> >
> > Hey all,
> >
> > Just following up here with a bit of a status update, so in the past
> week or so, items in the 1.8 release milestone have been closing out.
> > I'm aiming to cut a release next Tuesday, Jan 28.
> >
> > I'd like to reiterate that for any changes that don't make the 1.8
> release, we can do a fast follow 1.9 release, and from the last community
> sync that seems to be the direction.
> > In this particular case, the 1.8 release is a bit earlier than our
> typical release cadence and with the 1.9 being a fast follow on, I think
> we're well on track.
> > Please add the proposed changes to the 1.9 milestone so folks can review
> ahead of time!
> >
> > In general, I'd encourage more frequent releases, changes which are
> ready can just go out and with the smaller diff it reduces the risks that
> exist with larger updates.
> >
> > Thanks,
> > Amogh Jahagirdar
> >
> > On Thu, Jan 16, 2025 at 10:05 AM Daniel Weeks <dwe...@apache.org> wrote:
> >>
> >> Robert,
> >>
> >> I hear your frustration with the progress on the Auth Manager work, but
> I believe everyone recognizes that this was a large refactor further
> complicated by the need to preserve backward compatibility and handling
> deprecations appropriately.  This work has gone through many iterations as
> we explored how to make the changes cleanly.  Eventually the scale of the
> change led to breaking up the PR for closer review, which I believe was the
> right decision because we identified multiple issues after taking that
> step.  That may have slowed down progress, but a lot of hours have gone
> into discussing, reviewing, and validating the work in this area.
> >>
> >> As a project, we have leaned away from gating releases on specific
> features because it leads to slower release cycles and prevents other
> features that are ready from going out.  We also don't want to rush
> features just to hit a release target, but rather release more frequently
> to make changes available as they are ready.
> >>
> >> At this point, I believe the plan is to follow up soon with a 1.9
> release.
> >>
> >> -Dan
> >>
> >> On Thu, Jan 16, 2025 at 2:36 AM Robert Stupp <sn...@snazy.de> wrote:
> >>>
> >>> Hey,
> >>>
> >>> IMHO 1.8 should definitely include the Auth-Manager work, which tackles
> >>> actual bugs in the Iceberg code base wrt OAuth implementation. That
> work
> >>> was originally intended to go into 1.7 and now it shall be delayed
> again
> >>> for 1.9. The PR was originally opened in July 2024, about half a year
> >>> ago and is still getting reviewed. In the meantime there were more than
> >>> 600 other PRs that got reviewed and merged.
> >>>
> >>> The overall agreement around spring 2024, please correct me if I am
> >>> wrong, was the whole REST/OAuth area needs to be improved, and the
> oauth
> >>> endpoint removed entirely.
> >>>
> >>> Generally speaking, and I know I'm not alone, getting reviews from
> >>> Iceberg committers is extremely hard. A lot of issues and PRs just get
> >>> closed (by that stale bot) without having gotten _any_ attention from
> an
> >>> Iceberg committer. This is not a new situation but going on for a long
> >>> time. I have been talking to two Iceberg PMC members in person many
> >>> months ago that this is a very disappointing situation that needs to be
> >>> fixed. The reply was always "we are already working on it" - but at
> >>> least from my personal POV the situation did not improve.
> >>>
> >>> Robert
> >>>
> >>> On 16.01.25 10:56, Jean-Baptiste Onofré wrote:
> >>> > Hi folks,
> >>> >
> >>> > Following the Community Meeting yesterday, I would like to propose
> the
> >>> > following plan regarding releases:
> >>> >
> >>> > 0. As a prerequisite to any release (1.7.2, 1.8.0, 1.9.0), as said by
> >>> > Ryan, we have to double check the NOTICE/LICENSE. Interestingly, I
> >>> > discussed this point with Fokko at the beginning of this week,
> because
> >>> > I have some doubts about LICENSE/NOTICE content in the "uber" jar
> >>> > artifacts where we shade dependencies. I'm doing a complete pass on
> >>> > all artifacts in 1.7.2-SNAPSHOT and 1.8.0-SNAPSHOT. I should have a
> >>> > complete analysis by tomorrow. This is potentially a blocker for
> >>> > release votes.
> >>> >
> >>> > 1. As soon as (0) is done, 1.7.2 can be submitted to vote. I will
> work
> >>> > with Fokko on this one.
> >>> >
> >>> > 2. We plan to do 1.8.0 in a couple of weeks (Amogh is the release
> >>> > manager). Due to still some WIP, we "revisited" the 1.8.0 release
> >>> > content: for instance, as best effort, we wanted to include REST Auth
> >>> > Manager improvement (OAuth2 Manager) but we preferred to postpone to
> >>> > 1.9.0. That's totally fine to me, however, I would propose to
> strongly
> >>> > focus on pending PRs for 1.9.0. Imho, we should "target" (again as
> >>> > clear best effort) on variant, partition stats and Auth Manager.
> >>> >
> >>> > 3. Assuming 1.8.0 will be released at the end of Jan/beginning of
> Feb,
> >>> > according to our "release cadence", what do you think about planning
> >>> > 1.9.0 in April ? Again with the main targets listed in (2).
> >>> >
> >>> > I tried to sum up what we discussed yesterday :)
> >>> > Thoughts ?
> >>> >
> >>> > Regards
> >>> > JB
> >>> >
> >>> > On Thu, Jan 9, 2025 at 7:51 AM Jean-Baptiste Onofré <j...@nanthrax.net>
> wrote:
> >>> >> Hi folks,
> >>> >>
> >>> >> We did Apache Iceberg 1.7.0 release on Nov 8, 2024. If we want to
> keep
> >>> >> our release "pace", 1.8.0 should be released around mid February.
> >>> >>
> >>> >> I think we already have a good "train" of merged PRs (or should be
> >>> >> merged soon): default values, REST auth improvements, dependencies
> >>> >> updates, etc.
> >>> >>
> >>> >> WDYT about 1.8.0 mid Feb ? If so, I propose we update GitHub Issues
> >>> >> and PRs we would like to "target" to 1.8.0.
> >>> >>
> >>> >> Thoughts ?
> >>> >>
> >>> >> Regards
> >>> >> JB
> >>>
> >>> --
> >>> Robert Stupp
> >>> @snazy
> >>>
>

Reply via email to