Roadmap 0.3

[huachaohuang]

Planned release date: Jan 28, 2022.

I don't have a clear plan for 0.3 yet. One possible goal is to provide a basic implementation of Luna engine. Open for discussions.

• Roadmap 0.3 - Refactor Kernel interfaces #235
• Roadmap 0.3 - Refactor Journal interfaces #229
• Roadmap 0.3 - Refactor Storage interfaces #228

[w41ter]

It seems that these APIs in Luna engine are sufficient for 0.3 yet. In addition to providing a basic implementation, it also makes sense to make this implementation more solid and practical.

[huachaohuang]

@zojw seems working on the k8s and docker stuff, any thoughts?

[zojw]

yup, I'm trying to do some investigation(learning k8s- -, write rough code, and prepare design discussion) in a local branch, the mainly direction just follow last online-meeting's idea(boostrap kernel, then kernel boostrap & monitor & scale other components...)

current progress is...

• package docker image (single binary single image, run different component by different argument)
• manager components as CustomResource(kernel, journal, storage, engine, [background])
• discovery component's address:port via API(rough HTTP now)
• deploy or scale component via API(rough HTTP now)
• another demo gRPC kernel deploy & discovery via API

[ ] still under investigation, but it seems almost workable, and I will give a detailed discussion thread when the last two items are done(maybe this weekend or early next week), we need more k8s experts to help review

although there are many decisions & designs we need to take care & discuss, IMO we can finish those discussions and ship a k8s rough runnable version in v0.3, this also can help us to verify Luna engine's solid distribute running ability

[tisonkun]

@zojw could you start a discussion thread so that we can focus on implementing deployment? You can later create an issue with concrete items by "Create issue from discussion". (Here is an example)

[tisonkun]

One possible goal is to provide a basic implementation of Luna engine.

As mentioned in #266 (comment), the journal API is still unstable as described in #229 and #260. Also, the binary is broken now.

Do you still plan to deliver a basic implementation of Luna engine at Jan 28, 2022? Will it be a memory implementation, and used as a library only?

[huachaohuang]

I think regular releases are not suitable at the current stage for two reasons:

• It forces us to make a lot of temporary design and implementation to deliver things on time.
• Too many low-quality releases will be harmful to the image of the project and people will lose interest in us.

So I suggest that we think about that bigger picture described in #245 first. We can come back to planned releases when we have a clearer path towards that goal. #252 #263 #260 #251 are some attempts to clear the path. I think I should start a new tracking issue for Engula 2022.

[tisonkun]

@huachaohuang totally agree! Before we have a clear picture, we can release when features are ready, i.e., a basically usable luna engine with deployment supports :)

Note: please update the PR description. We can still set up targets for a 0.3 release without an explicit deadline.

[huachaohuang]

Close this in favor of #270 for now. We can open a new issue for v0.3 when we have a clear estimation of the schedule.