On 04/23/2013 11:41 AM, Christian Huitema wrote:
You might argue that, clearly, the autoconf prefix needs to be part
of the seed (I'd personally not expect developers to figure this
one out).
This is the kind of attitude that really does not go well with actual
developers. You are basically saying that you expect developers to be
too dumb to understand what they are doing,
You're misreading me.
I'm saying that developers work across across many different areas and
are busy people, and I cannot expect them to figure out stuff what *we*
were lazy or dumb enough to leave unspecified.
If I'm the guy producing a spec, my goal is to produce a spec that is
clear as possible, and only leave open those bits that are necessary to
leave open.
and thus you feel
compelled to specify an algorithm in its most minute details, even
That's incorrect. For instance, that's why we don't mandate a specific
PRF: because for that particular case, developers are in a better
position of judging what's the best PRF to use.
though there are dozens of equally good ways to achieve the same
result.
So far you have only mentioned one (recording the Prefix/IIDs pairs),
which IMHO is not elegant. Any others?
FWIW, you'll realize that e.g. RFC 4941 has even more detail than this spec.
The net result of such over-specification is that developers
will discard the spec as arrogant, and implement what they fell like
implementing.
I cannot control people's thoughts. However, I'd say that I'm fairly
well-known for socializing stuff among developers, well-before a
document is published as an RFC.
Much of the work that I've done over these years have resulted in
patches from a variety of OSes, because I cared to discuss them with
developers. Of course, documents that I've authored have benefited from
input from developers during that discussion process.
(Not to mention that the contents of the documents I've written have
generally been very influenced by the input of a number of
participants... anyone who has ever sent me feedback on a document I've
authored/co-authored knows I do my best to address all feedback, no
matter whether that implies re-writing large portions of the document in
question.)
Make the spec as clear as possible save time and energy on all sides.
Cheers,
--
Fernando Gont
e-mail: fernando(_at_)gont(_dot_)com(_dot_)ar ||
fgont(_at_)si6networks(_dot_)com
PGP Fingerprint: 7809 84F5 322E 45C7 F1C9 3945 96EE A9EF D076 FFF1
--
Fernando Gont
SI6 Networks
e-mail: fgont(_at_)si6networks(_dot_)com
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492