[DISCUSS] commit messages on backports

[Eric Covener <co...@gmail.com>]

In a branch of a private discussion, some issues with how backports
are committed was raised.

IMO, our use of STATUS kind of makes the current Reviewed By: a little
misleading in http://people.apache.org/~jorton/svn.merge

* My older copy of the script doesn't have it at all, and I rarely
edit the "clog" it generates -- meaning I almost never capture the
original reviewers in the commit log.

* Jim had modified his copy, presumably related to the same confusion
vs. what we call the reviewers in STATUS, but it introduced a
different misleading overlap when the work to port a fix was
noteworthy.


Do we want to call the list of reviewers from STATUS mandatory in the
commit to the stable branches?

I am personally -0 on _requiring_ it as STATUS and backporting can
already be a bit tedious, and ultimately most reviews seem to be
desk-checks.   I wouldn't mind if svn.merge required more input and
stopped me from breaking a rule though.


-- 
Eric Covener
covener@gmail.com

Re: [DISCUSS] commit messages on backports

[Jim Jagielski <ji...@jaguNET.com>]

For the record, this is what I use:

  http://home.apache.org/~jim/code/svn.merge

Most likely I will change it to have it accept $1 as the
names of people to mark as "Reviewed by" via a simple cut/paste
of the line from STATUS.

> On Dec 23, 2016, at 8:36 AM, Eric Covener <co...@gmail.com> wrote:
> 
> In a branch of a private discussion, some issues with how backports
> are committed was raised.
> 
> IMO, our use of STATUS kind of makes the current Reviewed By: a little
> misleading in http://people.apache.org/~jorton/svn.merge
> 
> * My older copy of the script doesn't have it at all, and I rarely
> edit the "clog" it generates -- meaning I almost never capture the
> original reviewers in the commit log.
> 
> * Jim had modified his copy, presumably related to the same confusion
> vs. what we call the reviewers in STATUS, but it introduced a
> different misleading overlap when the work to port a fix was
> noteworthy.
> 
> 
> Do we want to call the list of reviewers from STATUS mandatory in the
> commit to the stable branches?
> 
> I am personally -0 on _requiring_ it as STATUS and backporting can
> already be a bit tedious, and ultimately most reviews seem to be
> desk-checks.   I wouldn't mind if svn.merge required more input and
> stopped me from breaking a rule though.
> 
> 
> -- 
> Eric Covener
> covener@gmail.com