ietf-mta-filters
[Top] [All Lists]

Re: Stripping leading/trailing spaces in relational draft

2005-10-11 17:30:23

On Tue, 2005-10-11 at 19:16 -0400, Barry Leiba wrote:
Well, I think *all* comparisons should strip leading and trailing
white space

would you want the base spec to say that this stripping should be done
for _all_ tests?  my feeling is that we may regret that in the future,

Yes, that is, indeed, what I said (and meant).  Both sides of every
equation.

that's not what "relational" says, it only strips whitespace from values
taken from the message, in other words the left hand side.

Why do you think we'll regret it?

I don't know, just a gut feeling.  sooner or later someone will
introduce a test where keeping whitespace is the "intuitive" behaviour?
"string" may be such a test, but even "address" is to some extent.

recall that <"kjetilho"@ifi.uio.no> is semantically equivalent to
<kjetilho(_at_)ifi(_dot_)uio(_dot_)no> (RFC 2822, 3.2.5).  therefore, it is 
IMHO natural
that Sieve simplifies the localpart to 'kjetilho'.  example:

  address :localpart :is "From" "kjetilho"

this test should return true for both addresses.  consider then <" odd
address"@example.com>.  given the canonicalisation rule above, the
"address" test can not test for this value.

this is clearly an edge case, and it may be acceptable to sacrifice it
simply to ensure that all the relevant documents[*] agree on the
behaviour and specify a simple rule.  in my book, a simple rule should
not specify different behaviour for the two "sides" of a comparison.

[*] base, relational and variables
-- 
Kjetil T.

<Prev in Thread] Current Thread [Next in Thread>