Am 2022-01-26 um 22:00 schrieb Slawomir Jaranowski:
Hi,
What do you think ...
Maven Core and important components (list needed)
- RTC obligatory, approve from X committers, no request changes
Good.
Common rules:
- build and tests always pass or are under control
- after each commit project is ready to next release
- as maintainer add page https://github.com/pulls/review-requested to your
favorite
Insane, I did not even know about this overview.
Recommended changes without PR:
- fix links in docs
- fix misspells
- code comments
- improve for configuration, like .asf.yaml, .gitignore, GitHub workflows,
etc
- another ....
Agreed.
Rest - CTR with recommended PR with rules
1. No review
- PR - create
- request review by GitHub from at least 2 commiers
- wait 3 days (the same as for votes or another value)
- at least one approve or no response
- process
At least three business days, not calendar days. But also consider that
even non-core can have a lot implication if it is a important component.
I don't know whether 3 days is enough. I need often much more than three
days. Consider that almost everyone here is a volunteer.
2. Review with comments
- PR - commented without request changes
- consider and make suggested fix, response for question
- rerequest review on GitHub
- wait 3 days
- process
As above.
3. Review with request changes
- PR commented with request changes
- make suggested fix or explanation, response for question
- re request review on GitHub
- at least one approve from someone else, or remove request changes by
author
- process
As above.
The expected behavior at the invitation for review
- do review in 3 days
- leave comments that you want to review later, point when
- drop yourself from review request if don't want to do review
- add someone else for review request
Makes sense.
Of course each change is unique and the author knows the best if should
wait for review and the author is aware and responsible for self change.
+2 no this one. Think twice whether you are 100% certain that your
change pushed unreviewed.
śr., 26 sty 2022 o 10:34 Olivier Lamy <ol...@apache.org> napisał(a):
makes sense for core and it looks to be a de facto standard.
but plugins and components maintained by only 1 (maybe 2) persons.
we already have plenty of PRs from external contributors never reviewed...
And I do not mention some PRs which take weeks to review..
I feel like adding this will:
- ADDING EVEN MORE EMAILS TO THE ALREADY HUGE AMOUNT WE RECEIVE!!! (frankly
it's already hard to follow anything and I already miss some ping)
- delay even more moving forward (such process works for companies with
people dedicated to it but we are volunteers)
core or important components that's fine but I definitely vote against
moving plugins/components to RTC
.
On Wed, 26 Jan 2022 at 19:19, Enrico Olivelli <eolive...@gmail.com> wrote:
Il giorno mer 26 gen 2022 alle ore 07:44 Benjamin Marwell
<bmarw...@apache.org> ha scritto:
Hi!
Aside from typo fixes, PRs might slow down the dev speed, but greatly
improve code quality.
Even small PRs might have side effects you might not notice at first
glance.
But then, even with typo fixes I have seen people introducing more
fixes or unintentionally
thought they would fix something when in fact they introduced a type.
In my humble opinion, reviews BEFORE commit should be the default case.
- review from at least 2 committers (so only one if the patch comes
from a committer)
- CI passing
+1
Enrico
Am Di., 25. Jan. 2022 um 20:47 Uhr schrieb Slawomir Jaranowski
<s.jaranow...@gmail.com>:
Hi,
On the page "Apache Maven Project Roles" [1] we have paragraph
about Committers with:
The Apache Maven project uses a Commit then Review policy and has a
number
of conventions which should be followed.
Looks like Review then Commit policy is from svn time, so should be
refreshed or confirmed that is actual.
When we want Review than Commit policy, we need some rules which
allow
us
to effectively work if nobody is interested for feedback.
We also need rules / examples for direct commits, when they are
acceptable.
Do we need different rules for Maven core, plugins, shared ... etc
Example of rule:
PR -> no feedback for X days -> send a mail on dev@, if still not
feedback
after X days after mail -> proceed alone
[1] https://maven.apache.org/project-roles.html#committers
--
Sławomir Jaranowski
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@maven.apache.org
For additional commands, e-mail: dev-h...@maven.apache.org
