CFV: New Project: CRaC

Anton Kozlov akozlov at azul.com
Fri Aug 6 08:04:20 UTC 2021



On 8/5/21 9:33 PM, Mario Torre wrote:
> However, like others I feel that there was not enough engagement and not enough discussion to put the basis for a project just yet. It has been mentioned that there are others exploring this problem space and this highlights the need to have a common infrastructure, however some of the comments coming from developers of said projects were largely ignored or dismissed even if they did offer some good points of discussion, so that doesn’t seem consequential.

If that happened, it was not intentional. I'm asking a pardon in advance. If
you'd point to the comments I overlooked, that would be very helpful.

The Project is the first step to the common infrastructure, where at least we
start moving technical discussions and combining all different efforts.

> Also, as was again mentioned already, some more people could contribute to the discussion, but the timing has bee so short, especially considering we're at the end of July/beginning of August, and many of us had noticed the project discussion only while on vacation (including me) without much chance to contribute.

I'm sorry for this. There was enough traffic on the discuss maillist for me to
assume the timing was OK. But it is always possible to add more
committers/reviewers to the project. The proposed Project is a long-running
activity, so I hope the discussions will always keep going. As a result, the
specific goals, artifacts, contributors are going to change over time.

> I think this project has the opportunity to be extremely promising, but it needs to go beyond the small set of original committers and include more groups, and engage in a bit more discussion to better define its premises. In my opinion, but apparently also in the opinion of others here, the overlapping with other groups is significant, and it's a concern for me that this is not recognised, for this reason I suggest to take a little extra time to reflect back on the problems we're trying to solve before requesting a formal project, and maintain the goal to commit to have an inclusive and wider audience of JVM and core JDK developers, including the security group.

I acknowledge that the integration of some deliverables of the Project needs
close cooperation with the OpenJDK community.   But it will be crucial in the
later stages of the Project when pieces will become ready for integration (so
far they are not). At the current very early stage it is important to establish
a common place and bring all discussion there. It seems we don't completely
understand all requirements from different parties, so there is no point to ask
for a review of details, before they may change. And as usual, there is a
trade-off possible when some concerns can be outweighed by the usefulness of
the result. The edge between completely useless but flawless implementation and
a very useful one but with known limitations is moveable. The effects of some
decisions are best observed in practice and prototypes.

The Project can assure the contributions are done properly. So far, I'm not
sure enough people have looked at the prototype, as it is hosted in a
third-party repo, not covered by OCA for example, so I can imagine possible
concerns from the legal point of view. The Project will serve as a place where
collaboration can happen.

So I don't see how more discussion before creating the Project will help to get
the idea moving, but I see a benefit of having the Project.

> Also, I really, really encourage you to change the name :) but that’s probably just me being pedantic.

The abbreviation is short, pronounceable, and accidentally reflects what
happens with the execution (there are "cracks" in the timeline of the Java
lifecycle). The Project name has leaked into API, but this may be changed
later.

Thanks,
Anton


More information about the discuss mailing list