On 2/24/11 7:28 AM, Andrew Sullivan wrote:
On Thu, Feb 24, 2011 at 10:18:10AM -0500, Worley, Dale R (Dale) wrote:
I would expect that the idnits rules change only very slowly. What is the real
story on that?
They seem to change more quickly than people realise. Also, of
course, a whole bunch of changes can go by between a given
contributor's last I-D, and a current one. Not all the changes are
intentional; they're often pretty clearly based on heuristics that as
often as not look to me like they're derived from what certain tools
do. So then if someone submits something that doesn't do exactly what
those tools do, they get a false failure.
Note that none of this is to attack the tools developers. I've ranted
before about the absurdity of checking the formatting of early drafts
for perfection, so I won't bother again.
Off-list, I asked Henrik which errors the idnits step finds. A summary
of his answer is "only two non-boilerplate errors (non-ascii characters,
and lacking an Abstract), and many boilerplate errors (mostly bad IETF
Trust text, some about the boring first-page boilerplate)". So, I
suspect that the large percentage of the non-00 drafts getting kicked
back are due to the IETF Trust requirements changing over time. We can
argue (and have argued!) about all that, but I think saying "idnits" is
likely just shooting the messenger.
Ietf mailing list