[rfc-i] Potential RFC format approach: HTML

Joe Hildebrand jhildebr at cisco.com
Sat Mar 24 10:46:18 PDT 2012

On 3/24/12 10:01 AM, "Phillip Hallam-Baker" <hallam at gmail.com> wrote:

> W3C has already addressed the same set of problems and has developed a tool
> set.

We should definitely learn from their experience.

> The main constraint to be addressed is that the W3 want to use their
> own style sheet. This means that they have to strip out any other
> formatting introduced by Word etc.
> That is not difficult to do.

And it seems completely reasonable from the point of the consistency of
their brand.

> The main constraint when using HTML is going to be deciding what
> formatting features are going to be supported in the stylesheet. The
> stylesheet itself is rather less relevant as the reader can even
> change it if they like.

As long as we're generating good HTML, writing alternate style sheets should
be trivial.

> So the only features of HTML that I think we would end up using in the
> source document are <H1..5>, <p>, <dl>,<ol>,<ul> and tables, stuff
> that hasn't really changed in ten years. That plus the attributes
> style= and id=.

This is the right discussion to have, and we can argue over precisely which
choices are correct.

As a starting point, <div> rather than <p>, and class= instead of style=,
though. :)

> The other set of considerations I see would be outside the remit of
> the IETF standard but matter a lot to some people writing specs,
> particularly specs with XML or other schemas or with worked examples.
> The way I generate those is to have an HTML document with just the
> running text with <include> directives specifying the places where
> segments are to be pulled from another source.
> This approach makes spec writing a lot easier as it means that every
> time I build the spec I can run the regression tests to generate and
> check all my examples and then pull in a completely checked and
> consistent set of schemas, examples etc.

You could always use <iframe> while you're writing, and use a couple lines
of perl to do the substitution into a single doc before submitting.

Joe Hildebrand

More information about the rfc-interest mailing list