Charles Lindsey wrote:
On Fri, 11 Jan 2008 07:42:49 -0000, Frank Ellermann
<nobody(_at_)xyzzy(_dot_)claranet(_dot_)de> wrote:
But SSP introduces unnecessary [FWS], where it
is not required as separator, simplified:
| tag = "dkim" [FWS] "=" [FWS] ( "x" / "y" )
The "=" already separates name and value, you
could as well write:
| tag = "dkim=" ( "x" / "y" )
The world is full of protocols which ask you to write parameters of the
form
<name> "=" <value>
Some of them allow SP/WSP around the "=", some allow folding as well
(maybe even comments), and some allow nothing. The biggest problem is
for users to remember what they are/are not allowed to do in each case.
Our first duty should be not to make that problem worse (supposing we
can discern in which direction "worse" arises).
The nearest and best knwon precedent is MIME <parameter>s, which allow
full CFWS around the "=" (though it requires rather careful reading of
RFC 822 to confirm that). Of course, nobody (FSVO "nobody") goes the
whole hog by putting comments there, nor even spaces, but whitespace is
quite common. Which suggests we should allow at least WSP*. I am happy
to forget the full CFWS, and probably FWS too.
I think I'm mostly with Frank -- WSP between the tag=value seems pretty
dubious to me. DNS isn't the same as 2822 in that manglers are not going
to come in and arbitrarily add FWS. It would be nice to keep a SP/WSP
after tag=value; though for readability.
Mike
_______________________________________________
NOTE WELL: This list operates according to
http://mipassoc.org/dkim/ietf-list-rules.html