[MERGE][PQM][Bug #246846] Attribute work using the 'author' attribute

[Stephen J. Turnbull]

Daniel Watkins writes:
 > Hi John,
 > 
 > Thanks for the feedback.  Below are some thoughts following some
 > discussion with my fellow sprinters.

 > So we've got several views on this here:

Your second option basically comes down to "don't forget any
potentially relevant information".  The other opinions seem to have to
do with who is *responsible* for maintaining the code committed
vs. who *deserves credit* for the code (who is aka the expert who gets
dragged out of bed @ 3am if something blows up).

ISTM (ie, I haven't thought about it at all ;-) that in most workflows
the responsible person will typically be the committer and the expert
naturally is the author.  However, there are too many exceptions
(responsible person authorizes commit by phone, etc).  And what if the
mechanics of the commit gets screwed up?  The person at the keyboard
(or the robot) may have relevant information.  How about reviewers?
Sometimes they deserve substantial credit (ie, are consultable
experts).

So I think it's wrong to think of this as "how do we fill in the
Submitter and Author blanks?"  Those are common and useful roles, but
some workflows won't fit that model.  This is already visible in the
fact that the now-traditional Committer role has been omitted (it's
implicitly "PQM") and the responsible party has morphed into
Submitter.

This could be implemented as a "role dictionary" or as an "actor
object" with appropriate attributes.

 >   * the author of the merge commit should be explicitly specified by the
 >     submitter

This too annoying and for many workflows will have an overwhelmingly
sensible default, it is going to get scripted.  Needs that default,
probably configurable by project.