Write after approval additions

[DJ Delorie]

fche@redhat.com (Frank Ch. Eigler) writes:
> It may be instructive to contrast this policy with that of binutils:
> 
> #                --------- Write After Approval ---------
> # 
> # Individuals with "write after approval" have the ability to check in
> # changes, but they must get approval for each change from someone in
> # one of the above lists (blanket write or maintainers).
> # 
> # [It's a huge list, folks.  You know who you are.  If you have the
> #  *ability* to do binutils checkins, you're in this group.  Just remember
> #  to get approval before checking anything in.]
> 
> 
> I've always wondered what this specific colour along the power
> spectrum was supposed to accomplish.

As the author of those paragraphs, I can say with authority that the
reason for the wording is simple:  laziness.

There was *no* list of write-after-approval people for binutils.  All
I had to go on was the cvs permissions list and a long history of
random people checking things in all over the place.

I wouldn't recommend using the binutils wording as an example of the
right way to do it.  I prefer Andrew's policy better, but I wouldn't
be able to implement it for binutils.  Perhaps Nick could/would.

> In what way is it a *useful* middle point between maintainers and
> ordinary contributors?

The maintainers just have to reply with the word "approved" and don't
have to apply the patch, build, test, and fight with cvs to get it in
the repository.  Also helps when a patch covers multiple maintainers'
territories; once both approve it the submitter can check it in to
both places.