Some WGs have actually done this as RFCs: e.g.
http://www.ietf.org/rfc/rfc3665.txt
In some cases, it's no problem doing this (even if out-of-scope) as an
ad-hoc activity on the WG mailing list, though that's up to the chairs
and they could always suggest the activity move elsewhere. I suppose
if there's effectively a code-base rather than an Internet-Draft, it
might be nicer to use a open-source collaboration site with SVN as
well as forums instead of this mailing list anyway -- so it depends
also on participation and activity level as well as how the corpus is
structured.
Lisa
On Thu, Apr 9, 2009 at 1:16 AM, Robert Burrell Donkin
<robertburrelldonkin(_at_)gmail(_dot_)com> wrote:
i'm much happier with the current sieve-in-xml draft. we hope to ship
JSieve soon with experimental preview partial support with the hope
that usage by developers may help to understand whether any issues
have been missed, with a complete implementation would be targetted
for the next release.
however, to confidently ship a library claiming full support, a
reasonably large and fully debugged corpus of integration tests is
needed: basically, example sieve scripts together with corresponding
sieve-in-xml documents. IMO this is the sort of data which is usually
best developed and debugged collectively.
from the sieve working group charter:
<blockquote cite='http://www.ietf.org/html.charters/sieve-charter.html'>
(5) Produce one or more informational RFCs containing a set of test
scripts and test email messages that are to be filtered by the scripts,
and the expected results of that filtering. This will serve as the basis
of a interoperability test suite to help determine the suitability of
moving the base specification and selected extensions to Draft status.
</blockquote>
as i read this, a collection of example sieve scripts and sieve-in-xml
documents is out of scope for this group (hopefully people will jump
in if my reading is incorrect)
so, i was wondering whether there is any interest from members of the
group in collaborating offshore (http://code.google.com, say) on
creating and debugging such a corpus as an open source project under a
very permissive license (MIT, say), or whether there are any other
good ways that such a corpus could be created and made freely
available
- robert