Aluminium MRI project status


Robert Varga
 

Hello everyone,

I just checked the release schedule
(https://docs.opendaylight.org/en/latest/release-process/release-schedule.html)
and it seems we are something like two week behind the schedule:

Release Integrated Deadline was supposed to happen on 4/13, yet it is
not quite there, yet. I am sorry for the delay, there is a number of
contributing factors, some of which are outside of my control.

TL;DR: downstream projects need to pick up
https://git.opendaylight.org/gerrit/q/topic:aluminium-mri and start
creating their integration patches.

At any rate, downstreams are still not ready to receive the version bump
patches, as the migration to MD-SAL APIs is still ongoing -- and we
cannot contemplate creating version bump patches until that activity is
completed, those patches are merged and whatever CSIT fallout is taken
care of.

We also have a major infrastructure problem with releases not mirrored
out to Central, which is tracked here:

https://jira.linuxfoundation.org/servicedesk/customer/portal/2/IT-19554

Until that ticket is resolved (and Central is resynced), we cannot
really make complete releases. Our inter-MRI-project Javadoc generation
relies on javadoc.io, which operates solely on Maven Central.

What that means in concrete terms is Javadocs for yangtools-5.0.1 are
not available on javadoc.io, which in turn means mdsal-6.0.0 has broken
javadoc links to yangtools.

Rather than holding off until LF IT finally gets its act together w.r.t.
actually monitoring the services
(https://jira.linuxfoundation.org/servicedesk/customer/portal/2/IT-19525)
they are supposed to provide, I have opted to release such a broken release.

With that out of the way, though, there is a pretty clear picture as to
what the MRI versions are going to be:

1) odlparent-7.0.1 is ready

While there is a testing problem around Geronimo and refreshes, this
affects only speed of SFT -- it is a Karaf/Pax-Exam interplay only and
has no bearing on how our code works in production. This is tracked
here: https://jira.opendaylight.org/browse/ODLPARENT-232 and will
eventually be rectified (contributions are more than welcome, I will not
take a further look at it for at least next two months).

2) yangtools-5.0.1 is ready

While yangtools-5.0.0 was missing a critical (API-breaking) piece to
make ends meet between yangtools/mdsal/netconf, version 5.0.1 has those
rectified and does not seem to contain any regressions.

3) mdsal-6.0.0 is ready

Just hot off the press, this looks to be a solid release with a good
balance between imposed breakage and feature completeness. I ended up
reverting MDSAL-490 and punting it until 7.0.0 just because a number of
downstream projects are nowhere near to being ready to receive this cleanup.

4) controller-2.0.0 needs a few more days

While all the functionality (removals) have landed quite some time ago,
which are the reason for the 2.0.0 moniker, there remains some work in
the land of feature packaging to reflect the ~34K lines removed in this
release.


The migration guide update lives here:
https://git.opendaylight.org/gerrit/c/docs/+/89407 and is subject to
finalization over the course of next week or so.

Regards,
Robert

Join TSC@lists.opendaylight.org to automatically receive all group messages.