David Pursehouse, Ben Rohlfs, Alice Kober-Sotzek, Patrick Hiesel, Luca Milanesio
Online, May 5, 12:30 - 13:30 CEST
The ESC meetings will take place once a month from now on.
The next meeting will be held on June 2, 09:30 CEST.
With regards to the support of the latest version of error prone, David Ostrovsky made some work and progressed the activity. Currently waiting for the Bazel Team to review and accept his GitHub PR #11271.
Sven Selberg is taking some of the work to adress the issues raised on the reviewers plugin to make it suitable for promotion. From a popularity perspective, the reviewers plugin is already the most used plugin of its type, amongst the general shortlist of the top 100 plugins for Gerrit.
There is a continuous progress on the features in development for v3.2.
The multi-master scalability of the replication plugin is close to completion. There is a concern on the test coverage and real-life production use of the functionality. It will be flagged as experimental because of lack of production use and feedback.
The support for global ref-db in the high-availability plugin has been agreed with Marco Miller: the implementation will be done on the high-availability plugin master branch and, if finished on time, released at the same time as Gerrit v3.2 but flagged as experimental.
Ben Rohlfs agreed to start validating PolyGerrit in v3.2 and master using Chrome 80. Existing stable releases will be continued to be validated against the older version of Chrome, because of the issues found with WCT tests on Chrome 80. A specific note will be included in the release notes.
Marco Miller has agreed to organize a remote joint meeting between the ESC and CM, because of the cancellation of the Spring 2020 hackathon due to COVID-19.
Luca Milanesio raised the issue that some changes merged in the past have created issues in released version of Gerrit. We do not currently document or enforce a quality gate for reviewing and approving changes.
Ben Rohlfs pointed out that we have already a quality criteria to nominate maintainers and we assume they are good enough for the job of reviewing and approving changes. A further complication in the process is not needed.
There is consensus on Ben's position: the current contribution process will remain as it is today.
The design has been analyzed by Patrick Hiesel and provided a positive feedback to move forward, as long as the instance-id is optional and not required / populated for single-master setups. There are no objections to the decision: the design is approved.
A few months ago, gerrit-review introduced a group for adding accounts that have restricted permissions. David Pursehouse raised the issue that having restricted permissions is not enough as the spammers would still be able to login and operate on gerrit-review.
Patrick Hiesel proposed to disable the spammers' accounts with a REST-API authorized to the Gerrit maintainers and Ben Rohlfs proposed to add button on the Gerrit UI should a suitable API be available in the backend.
It was supposed to happen during the next forthcoming Gerrit Hackathon in April/May 2020, now cancelled because of Covid-19. A follow-up discussion needs to happen with the CMs to understand how this can be coordinated remotely with all the different time-zones to cover.
There were no other issues that have updates or require attention.