Because it's sufficient to generate the ASCII version once on
publication and then keep it.
Keeping the source is essential for completely separate tasks (meta
data extraction, document revision, generating other formats such as
HTML or PDF).
You are using "the ASCII version" as a proxy for a printed page
(and as I once complained, it is not a faithful proxy on all platforms).
However, the problems we wanted to solve were precisely those where
is not sufficient, namely graphics and equations, and so we need
to return to the printed page as the final word.
So far the IETF hasn't done that. The format is ASCII.
One good reason to use a specific XML grammar is that when the only
thing that's available is a presentation-oriented format
(such as Word or PDF), it gets *much* harder to do meaningful things
with the source.
You are making assumptions about presentation formats.
It is quite easy to do meaningful manipulations on TeX source.
And that's one of the reasons why volunteers maintain xml2rfc (both
the format itself and various implementations).
And here is precisely where we are expending efforts.
I too enjoy coding, but why are we recreating for the XML2RFC
mechanisms that exist in available tools?
The zenith of these efforts is a script to extract TeX-style math
expressions, run TeX on each separately,
create images, and then embed them into a PDF created from the xml.
... and all that just to avoid using TeX.
Ietf mailing list