Hi Duo
For complex feature, in Hadoop and HBase and related big data projects I’ve
seen in ASF, we will create a feature branch, and then open PRs against
this branch. And since it is not master, you are free to apply even half
done PR if the full one is too large. And finally when you are done, start
a vote to merge it back.
Usually we will open an issue on jira, and create a sub task for each
commit of the branch, and the branch also follow the PR-Review-Merge
workflow.
What David said many times, open a PR and then everyone push to update the
PR, I will not define it as our official workflow. You are free to use what
you like to develop a feature in your own repo, but when you want to merge
it back, you open a PR and do what every contributor should do.
As I mentioned in a previous email, I proposed some English for the
feature branch requirements at the bottom of
https://cwiki.apache.org/confluence/display/NUTTXTEST/Code+Contribution+Workflow+--+Brennan+Ashton
What I wrote there is essentially the same that you are saying here.
You are welcome to review and comment at that URL (which will be moving
when the website is moved to its final home).
Greg