Re: Rewriting the subject (Was: [sgnus v0.83] Followup Subject: typos in message.el (patch))

["Sudish Joseph" <sudish@VNET.IBM.COM>]

Greg Stark <gsstark@MIT.EDU> writes:
> Do we really want to construct the new header as:
> Subject: Re: Re[145]: re^4: ...
>
> It seems we're only making the problem worse, the header is now even
> further from the structure intended by the RFCs.  To some degree i

I dunno.  It seems to me that the header is finally fully compliant
with So1036.  Note that doing this isn't promoting further growth, any
compliant reader following up to that article will leave it alone from
this point.

> standards, we would only make those authors claim, "well look what
> happens when people strictly follow the standards; standards
> restrict my ability to support cool new features and make stupid
> headers like these."

Which is precisely what standards are there for--to prevent people
from adding cool new features that make interoperability an impossible
task.

Add features that are permitted by the standards.  It's obvious that
most new features will be in areas not covered by the standard; except
where said features don't have anything to do with interoperability--
all the cool colors in GNUS are new features that don't hinder
interoperability.  MIME is a feature that affects interoperability,
but it was added to an area not covered by the existing standards of
the time and in a fashion that lets older UA's still interoperate with
MIME-aware UA's.

> And now, what if i write a hook to do my "manual" rewriting of the
> Subject automatically, is that ok? What if Lars wrote it? What's the
> difference?  I don't think you can change the world by making life

There's no difference between you writing it and Lars writing it, *for
your personal use*.  The moment you make it GNUS *default behaviour*
or even something that is part of GNUS, you break our compliance with
standards and some non-standards.

Incidentally, neither RFC1036 nor So1036 is a standard (821/822 are,
but not 1036)--if you're throwing out the Seal as a non-standard, why
not do away with RFC1036 formatting of messages? :-)

(Uh, that was a lame joke. Sure, 1036 isn't really a standard, but
it's our last line of defense against Chaos and Other Bad Things.)

> harder on other packages, you can make life easier by demonstrating
> what headers are supposed to look like.

Sure, as long as we're talking about headers not covered by standards.

-Sudish