MIME-Version: 1.0
In-Reply-To: <CAL0fb63-zObvzirU1T6-xQnKc4=Ly2go5BBF9Q0XjqAc3o8V7A@mail.gmail.com>
References: <CAL0fb63-zObvzirU1T6-xQnKc4=Ly2go5BBF9Q0XjqAc3o8V7A@mail.gmail.com>
Date: Tue, 20 Sep 2011 14:54:46 -0400
Message-ID: <CA+8xBpdF88tTHOT40=-9enrb4hsekexELSrctdHDK8QqWxGVXw@mail.gmail.com>
From: Jeff Garzik <jgarzik@exmulti.com>
To: Alex Waters <ampedal@gmail.com>
Content-Type: text/plain; charset=ISO-8859-1
Cc: bitcoin-development@lists.sourceforge.net
Subject: Re: [Bitcoin-development] Issue / Pulls timers
Precedence: list

On Mon, Sep 19, 2011 at 8:20 PM, Alex Waters <ampedal@gmail.com> wrote:
> I would like to propose adding timers to the pulls / issues:
>
> https://github.com/bitcoin/bitcoin/pull/523
>
> "From time to time a pull request will become outdated. If this occurs, and
> the pull is no longer automatically mergeable; it will be closed after 15
> days. This can be avoided by rebasing the commit. Pull requests closed in this
> manner will have their corresponding issue labeled stagnant.
>
> Non-bug issues with no commits will be closed after 15 days from their
> last activity.
> Issues closed in this manner will be labeled stale.
>
> Requests to reopen closed pull requests and/or issues can be submitted to
> QA@BitcoinTesting.org. "


We need to avoid a user/contributor experience of:  "my pull request
was abruptly closed with no warning"

Contributors might not track the state of the tree on a day-to-day
basis.  Thus, following the example of bugzilla.redhat.com and many
other "tracker" applications, outdated issues first initiate an
automated warning email -- usually by adding a comment to the bug
report -- that describes the policy, why the policy (closing outdated
reports) exists, and how to avoid automated report closure.

In our case, this means a "we will close pull req, unless you update
this commit in 15 days" comment should be added to the pull req.  The
comment should describe in broad terms, with links, how to rebase a
commit, what standard expectations are, etc.

Closing with no warning should be avoided.

-- 
Jeff Garzik
exMULTI, Inc.
jgarzik@exmulti.com