On 2/3/07, Ted Husted <[EMAIL PROTECTED]> wrote:
On 2/2/07, William A. Rowe, Jr. <[EMAIL PROTECTED]> wrote:
> If they aren't a committer yet, they post a patch (jira or list) just like
> every other wannabe future committer. When the volume and quality are
> reasonable, they are offered commit access. But the suggested policy is to
> state "no backchannel dealings with codesubmissions. bring it to the list."
Agreed, but a detailed Subversion log is also part of the list. If the
Subversion entry details the source of the commit, cites a CLA, and
includes the design justification, then that's no different than
bringing it up on the list as a matter of lazy consensus. Over the
long term, it may be even better, since the information is attached to
the commit, and not just floating around on the dev list.
+1
Any controversial commit should be discussed first, regardless of its
source. Committers need to apply the same good judgment and discretion
to our own donations as we do to the donations of others. I think a
key problem with this policy is that it implies we can apply a
different standard to our own donations.
there are two different standards which need to be applied to two
difference classes of document:
* donations (whether covered by a CLA, JIRA opt in or a software grant)
* others (should be covered by compatible licenses)
i'm not sure that the proposed policy correctly capture this difference
it is important that committers understand that they need to be
certain that if the code is not an original work covered by a CLA they
need to note that in the commit record
A key idea is that the PMC is accepting donations on behalf of the
foundation. Whether the donation is code we happen to write, or
someone else has donated, isn't important. When I commit code I wrote,
I may be sheparding my own donation, but, I'm still not committing as
the "author", but as a PMC member authorized to accept donations on
behalf of the foundation, including those I happened to write myself.
the origin of the code matters and needs to be recorded in the commit
message. conventionally, it is expected that any code that is not
originally created for apache by the contributor has appropriate
attribution in the commit notice
whether the code has been donated or just reused under a compatible
license matters. conventionally, it is expected that any code that is
not originally created for apache by the contributor has it's license
status noted in the commit notice.
this could be summerised as lazy attribution: if there is no notice in
the commit message then it is implicit that the code is an original
work created by the committer and is covered by a CLA. in all other
cases, the attribution is explicit.
this seems important and needs to be documented somewhere, i think
> Otherwise we'll fail to recognize the merits of *individual's* contributions
> and therefore won't offer commit access when it's warrented. And *that* is
> the third part of the issue.
To me, this sounds like an issue with the Subversion log entries. The
source of all donations must be cited so that there is a clear
providence for the work. If we don't cite a source, then the
assumption is that the committer is also the donor. Every commit is a
donation, and every commit has an explicit or implied donor.
+1
except that some commits introduce documents (for example, third party
jars) that are not covered by software grant, AL2.0#5 or CLA. not sure
in this case that committers have the rights required to donate under
the CLA.
- robert
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
