[Top] [All Lists]

Re: Revisiting RFC 2822 grammar (Subject field)

2004-01-18 04:45:29

Russ Allbery wrote:

RFC 1036 is not a standard and this is not best practice for Usenet
software.  Please do not include this in the grammar.
RFC 1036's title is "Standard for Interchange of USENET Messages".  While it
has no standing as an Internet Standard it certainly claims to be a standard, and moreover it is current and is THE RFC addressing message format w.r.t. Usenet
(RFC 850 having been superseded by 1036, and 1036 not having been amended or

I agree that a number of things in 1036 are questionable practice; Subject field hacks are one of those questionable practices. But questionable or not, that's
what 1036 requires.

RFC 1036 is not a standards track document and therefore should not need
to be explicitly repudiated by another document.
As a self-proclaimed standard for message format, and with a number of discrepancies
with other message format RFCs, 1036 presents a problem for implementors; an
implementor apparently has the following choices:
a) ignore RFC 1036, which means ignoring Usenet, since 1036 is THE RFC dealing
   with Usenet message format
b) take 1036 into account, which imposes structure on the supposedly unstructured
    Subject field
c) pick and choose bits and pieces from the various standards

Option a is certainly a viable choice. Option b leads to the present discussion. Option c can lead to incompatibilities and lack of interoperability; it is the option of
last resort.

Repudiation of the questionable items may provide an out for the hapless implementor. Maybe not [old messages still exist, and may need to be parsed in accordance with old

One of the lessons to be learned is that it can be extremely difficult to recover from
poor design decisions.