Kudos Will!
I had received many private notes wondering if Accelerite will continue to play
a strong role in contributions to the community; here is your proof!
We wanted to take on the biggest pain points in the community, and see how we
can make positive contributions. Koushik Das will work alongside Will and
Patrick to address both of these problem areas. I believe that this will put
the community on a path to more manageable releases going forward.
Best
Samir
-----Original Message-----
From: Will Stevens [mailto:[email protected]]
Sent: Wednesday, March 02, 2016 9:15 AM
To: [email protected]
Subject: 4.9 Release Management
Hello Everyone,
I have mentioned this in other related threads, but I wanted to make an
official thread on the topic.
I am nominating myself as the release manager for 4.9. Please feel free to
discuss if you have comments or concerns.
I will not be working alone, I will be assisted by Koushik Das and Patrick
Dube. I will be running point, but all three of us will be working together as
a unit for this release.
Our main focus for this release is the integration of hardware Continuous
Integration (CI) into the PR flow. Koushik and his team will be setting up a
CI environment which will be used for testing PRs and I will also be setting up
a CI environment for testing PRs.
The details of the CI integration will be handled publicly, but we will likely
have to work with a minimum viable implementation initially and move forward
from there. Here are some of the key aspects of the CI which are top of mind
for me.
- Standardize a feedback mechanism to post the result of CI runs back to the
relevant PR. I believe the best way to do this would be to post a summary of
the CI run in the PR thread on Github. With the existing integration, this
will then get pushed to the mailing list (since all comments on a PR are pushed
to the mailing list).
- Ideally, we will also make the CI logs available for the run. We are still
working out the details of how we do this, but we will likely be pushing the
logs to an object store with a cleanup window to remove the logs after a set
period of time (probably a week). This should give people the opportunity to
pull the logs if they are interested in the test results, but will reduce the
need for ever growing storage.
- In order to parallelize the CI operations, we will not be automatically
kicking off a CI run for every PR for now. Instead, we will communicate
between us and each run distinct PRs so we can maximize the utilization of our
hardware.
Some longer term goals of the CI in my mind are as follows:
- I would like the core CI framework to be easily distributed and accessible to
anyone who has hardware available. This would enable anyone to setup a CI on
their hardware and it would automatically be hooked up to feedback the results
to the Github PRs. I feel this is very important long term because every
individual or organization depends on a different configuration and hardware
setup, so it empowers them to validate their own use case while adding value
back to the community.
Additional details will follow, namely the release schedule etc.
Please contribute your ideas and feedback.
Cheers,
Will
DISCLAIMER
==========
This e-mail may contain privileged and confidential information which is the
property of Accelerite, a Persistent Systems business. It is intended only for
the use of the individual or entity to which it is addressed. If you are not
the intended recipient, you are not authorized to read, retain, copy, print,
distribute or use this message. If you have received this communication in
error, please notify the sender and delete all copies of this message.
Accelerite, a Persistent Systems business does not accept any liability for
virus infected mails.
