Re: [arch-d] I-D Action: draft-iab-protocol-maintenance-04.txt

["John Levine" <[e-mail]>]

In article <CAOW+2duvc21ss5cxXM1Kr94HW3Fpn5nOYu=[e-mail]> you write:
>The robustness principle is *fundamental* to the operation of the Internet,
>which is about the interconnection of vast numbers of disparate
>implementations. ...

Yes, different implementations need to interoperate, but the best way
to make that happen isn't necessarily to program around other people's
mistakes.

In the e-mail world, I can assure you that SMTP implementations are
ever more strict about what they accept, since we've all observed that
real mail software implements SMTP correctly, and the stuff that makes
mistakes is spamware.  By being stricter about what we accept, we more
reliably deliver the mail our users want.

I think you can also distinguish between being liberal about things
that are ambiguous and things that are wrong.  For example, there's no
guidance about whether to send mail addresses in upper or lower case,
so in practice everyone accepts them interchangably.  (Even this turns
out to have less than ideal consequences since in mail with UTF-8
addresses, you can't reliably do case folding until you know what
language the address is supposed to be in, which you usually don't.)
On the other hand, if a mail client sends something that is just
wrong, leaving out required punctuation, or sending out of sequence,
that's just wrong, and generally is fatal for mail senders.

I think there is a lot to be said for Virtuous Intolerance, in at
least some contexts.  Figuring out which contexts the are will be an
interesting challenge.

R's,
John