Thank you for the comments. my comments in-line fyr.
On Fri, Oct 9, 2020 at 9:24 AM Yanhui ZHAO <[email protected]> wrote: > Hi Mnemonic folks > > As you know that we hope to reinstate mnemonic project from current stage. > To have a clear scope and goal for the project to continue move on, Gang > has drafted the following roadmap and timelines, and I have added my own > comments. > > This by no means is that we have already made decisions on the project > (that is why we hesitate to show it on the first place), but just would > like to share our thoughts on the project with the community. We sincerely > hope you can provide your inputs/insights, and proposals as well, as I can > see some folks are still interested in contributing. > > BR > > Yanhui > > Project goal: > To deliver a pre-product durable object model library working with > persistent memory (we are only focusing on persistent memory right? Do we > consider other storage options? I remember we also use HD and SSDs) and > high performance storage at the end of next year. > > Project objectives > Add one: Organizing a sustainable PMC group. > 1) A stable and efficient CI system (agreed) > 2) A improved documentation and website (The github page and web > definitely need lots of work. From a user's prospective, there is no > information from us on why our product will benefit them. We may need to > stress on easy adoptions and performance benchmark to highlight our core > values) > Gary: Completely agreed, my 2 cents, it is also a difficult work, we may need to simplify the documentation and improve the code comments for document generation as well. > 3) Integrated with Bytebuddy (used for code generation? should be helpful > but not as important as #2) > Gary: Yes, I recall one of our contributors mentioned it before, the Bytebuddy looks a very promising alternative way for code generation, but I cannot 100% sure about fitting our project yet, currently, we are using the Javapoet, probably we can keep using it if it is not very necessary so far. > 4) Supporting JDK14 and leveraging the feature of non-volatile mapped > byte buffers (incubator) (highly agreed, question on whether our JDK14 > support will conflict with other modules such as Spark/Hadoop. Mutiple > version of JDK might reduce adoption for stability concerns.) > Gary: I think the dependency of Spark/Hadoop showcases need to upgrade accordingly. the Spark/Hadoop frameworks may already have changed so the stability is a real concern as you mentioned. > 5) Adding Redis as one of backend of durable object (highly agreed ) > 6) Verified code quality and testing (agreed, test cases need to be > improved to reflect the project core values, and it is the interface > between us and our potential customers) > > 7) new language support such as python, golang, etc > > Highlight the key values of our project is (1) provide unified interface > for storage(memory) backend, (2) remove se-des process to improve the > access speed thus improving read/write performance > > Project Milestones > 1) Dec. 2020, CI documentation and website (v0.14.0) > 2) May. 2021, Bytebuddy and JDK14 (v0.15.0) > 3) Oct. 2020, Redis backend and quality (v0.16.0) > > Project Communication > 1) Weekly regular meeting 30mins - 60 mins (how to coordinate with ppls > from different time zones?) > 2) Event based release meeting > 3) Reporting meeting > 4) Maillist, Jira, PR based discussion > 5) Management meeting on demand (need to absorb more PMC members) > > >
