ietf-822
[Top] [All Lists]

Re: empty quoted strings and other oddities

2002-10-04 20:52:40

At 07:54 -0400 02-10-04, Gary Feldman wrote:
I don't understand the question.  Computer science students have
long been taught that they must reasonably handle invalid input, at
the same time that they document precisely what the input must be.
How is this different from web servers still accepting http 1.0 or
C compilers still accepting K&R C?

The problem is that the golden rule actually specifies three
standards:

A conservative standard, which implementors should use when
producing new data.

A liberal standard which implementors should use when
accepting incoming data.

The real standards text, which is somewhat inbetween.

It is difficult for an implementor to understand the
difference between these three standards, since most
standards text does not say what is the difference
between them.

Should you for example accept
From: John F. Kennedy <jfk(_at_)afterlife(_dot_)net>

Standards have a long time accepted the "K." even
if the user-friendly name is not quoted, but implementors
have not been told that accepting such a "." followed
by space is in the liberal category of what you should
accept but not produce.

The obs- syntax of RFC 2822 is an attempt at clarifying
the difference between these three standards.
--
Jacob Palme <jpalme(_at_)dsv(_dot_)su(_dot_)se> (Stockholm University and KTH)
for more info see URL: http://www.dsv.su.se/jpalme/