This wasn't on any mailing list as far as I can tell, but
here's a historical snapshot of the DISCUSS question. I'm
hoping Bill won't mind me forwarding this without permission
ten years later...
-------- Forwarded Message --------
Subject: Re: DISCUSS percentage question
Date: Fri, 29 Jul 2005 11:53:24 -0400
From: Bill Fenner <fenner(_at_)research(_dot_)att(_dot_)com>
To: brc(_at_)zurich(_dot_)ibm(_dot_)com
CC: klensin(_at_)jck(_dot_)com
Is there a way to answer this question with reasonable accuracy:
Determining the accuracy requires spot-checking results, which
I don't have time to do, but I did come up with an answer for
which I think the methodology is sound.
What % of standards track and BCP documents pass IESG ballot
without attracting a DISCUSS? (Over the last year, say.)
I got just about 30%:
+---------+-----------------------+
| ballot | count(approved.docid) |
+---------+-----------------------+
| NULL | 62 |
| discuss | 149 |
+---------+-----------------------+
Here's my methodology:
1. Create a temporary table with all discussed documents:
create temporary table discuss
select distinct docid,ballot from comments
where ballot='discuss'
(comments is my copy of the comment log, and ballot is whether it was a
regular comment, a ballot comment, or a ballot discuss)
2. Create a temporary table with the set of BCP or standards track
documents (docintend of [20..23]) that were approved
(newstate of [50..59]) since 8/1/2004 (changedate)
create temporary table approved
select distinct docs.docid from docs, dochistory
where
dochistory.docid = docs.docid
and
( docintend >= 20 and docintend <= 23 )
and
( newstate >= 50 and newstate <= 69 )
and
( changedate >= '2004-08-01' )
3. Join the two temporary tables together and count the result:
select ballot, count(approved.docid) from approved
left join discuss on discuss.docid = approved.docid
group by ballot
Bill