[Top] [All Lists]

Re: Revisiting RFC 2822 grammar (Subject field)

2004-01-19 05:42:12

Russ Allbery wrote:

I didn't mean to do that.  I am actually arguing the exact opposite by
saying that RFC 1036 is *also* not a well-defined standard to follow since
it won't interoperate with Usenet as it exists today.  I am arguing
exactly that Usenet is simply not well-defined, by RFC 1036 or anything
OK, I see. Presumably 1036 was reasonably accurate when issued, since it described the operation of a widely used implementation. The fact that actual use has deviated with no revision to the documentation says more about Usenet than about 1036.

You're saying that if you ignore RFC 1036 you have to ignore Usenet.  This
is obvious nonsense, since there are reams of software written to work
with Usenet articles that ignore RFC 1036.  My point is twofold:  one
cannot simply write to something that claims to be a standard because
sometimes it's badly obsolete, and it is actually possible to write
interoperable software without a standard (it's just very annoying).
I'm saying it's one option. Picking and choosing is another. Making things up with no supporting standard is another -- and you're right about that being annoying.

It would be difficult to present a convincing argument that Usenet isn't
the least important application that uses the text message


This is a completely different discussion that I'm not going to have on
this mailing list.

It simply means that given the alternatives (pick and choose, and justify the choices made, or make things up and justify them, it's far simpler to say "Usenet? Sorry, not supported."
That isn't to say that that's a great choice; it's the least of evils.

 I would prefer that this group *did* ignore Usenet
except for the small and very specific places where exactly following RFC
2822 makes it very difficult to use with Usenet-like applications, such as
allowing whitespace in the middle of message IDs.  Other than those very
specific issues, RFC 2822 work would be best served by ignoring Usenet
because Usenet would be best served by ceasing its attempts to tread off
into uncharted territory and moving back to using the same message format
as e-mail.
I agree that there are some issues where 2822 could make some things easier for some uses of the text message format. And I wholeheartedly agree that Usenet
should continue to use the common text message format (indeed, RFC 1036
(section 2) goes to great lengths to indicate that is is in fact the same format).

Unfortunately, some provisions of RFC 1036 have an impact on that common
message format, and the Subject field hacks are such provisions.