[Mailman-Developers] email schemas

[emf]

John Dennis wrote:

> O.K. that makes sense, but I guess it boils down to a design choice.
> 
> 1) Well defined DTD/Schema, but awkward to use in practice.

Another approach would be something like:

<header><name>To</name> bob at dobbs.org</header>
...
<header><name>X-Foo</name> blarg</header>

I'm not against that.

> 2) Easy to use, but no standardized DTD/Schema to be used for
> validation.

Well, schemas have the singular merit of being able to loosely specify 
something, so you could still have a schema, just not a DTD.

It also is problematic to have an arbitrary set of tags; I can imagine 
someone using characters illegal for element names in an X-header.

The above approach has the advantage of keeping things safely in CDATA 
segments, so it's probably better than attributes.

>  Most authors suggest strict DTD/Schema is mostly appropriate
> for documents with complex structure or where the document receiver has
> no apriori knowledge of the document structure. I don't think either of
> these apply in this circumstance. 

As I understand it, any user agent is free to throw on any X-header 
their little heart desires, so that strikes me as a lack of a-priori 
knowledge. For archiving/moderation purposes, we must accept anything 
that mailman does.

> BTW, is the intention this XML document is going to have full blown
> parsing all the way through all the mime (sub)parts?

I was angling for something along those lines, yes.

~ethan fremen