--On Montag, Mai 31, 2004 01:47:01 -0700 Greg Connor <gconnor(_at_)nekodojo(_dot_)org>
wrote:
[...]
1. The poll is pretty much useless for our purposes. I would really like
to see more choices, so we can get to the heart of the issues and
possibly solve them.
* Opposed to any XML, due to bloat, patent/IP issues, hate MS, other
(list any that apply)
opposed because:
- XML bloats data size (acceptable for HTTP, unaccepatble for DNS)
- full parser necessary for extensibility
- no clear advantage as XML only allows for extension of syntax, but we
would also need a mechanism to express the intended semantics of the
extension - implying in effect a programming language and/or executable
code to impement the desired semantics
[...]
* If extensibility is a requirement, XML wins. It at least has syntax
extensibility. So far it has hand-waving in place of feature
extensibility, but that can be sorted out easily.
But you only get syntactic extensibility. What can you use an extension for
if you do not know the semantics? And if have to implement the semanticvs
anyway, a minor modification to a parser to implement a new keyword or
mechanism is IMHO less work than adding a full blown XML parser right from
the start.
* Yes, extensibility IS a requirement, and version numbers don't cut it.
They cut it extremely well as they also connect the syntax to the semantics
- which are defined in the RFC and not in any kind of XML data.
I do not belive that there is a way to make any kind of specification truly
extensible, including extension of semantics. And without the semantics you
don't need syntactic extensibility.
We don't talk about optional parts here, if someone uses an extension that
you don't understand, evaluation should deliver "unknown" (SPF should be
fail safe). So either you risk an unknown result by using that extension or
you must be sure that a significant majority of users has implemented this
extension.
I believe that this goal is best served by successive versions of SPF that
implement new mechanisms when they are needed - and _only_ then.
Ralf Doeblitz