RE: Will Atom have a "don't process invalid documents" rule?

[Asbjørn Ulsberg <asbjorn.ulsberg@xxxxxx>]

Bill Kearney wrote:

> I certainly favor "must reject".

+1. Let's not get a new HTML-situation where virtually *everything*
goes. The validator is strict and conformant, but it doesn't change
the fact that nearly 99% of all HTML-documents out there doesn't
validate.

And we have to differentiate the levels of validation;

  1. Wellformed XML. '&', '>' and '<' etc must be escaped, every
     element must be closed, attributes needs quotes, etc. Non-
     wellformed XML just _can't_ be allowed.

  2. Valid XML. The XML validates after a given DTD or schema.

The first level is crucial, the second is important. If the first
level of validation goes wrong, a BIG RED «IDIOT» with a loud
«SCREECH» should be thrown at the sinner. If the second level
fails, we can be kinder, but point them in the right direction
and *don't* pass their feed as "ok", because it's not.

What I think we should and must to, is to provide a list which
Atom-readers and -producers can follow, not unlike GNKSA[1].
If they don't follow this list, then they won't get GNKSA-
approved either.

We would have to have the same type of system as GNKSA has, where
people can submit[2] applications, which then gets processed[3] by
one of us, and when finished processing ends up on the "Evaluated
Software"[4] page.

Not being on the "Evaluated Software" page, or being there, but
not passing, should be embarrassing to the developers. They should
blush when seeing their application fail the process, and users
should NOT be encouraged to use non-passing software.

Also, I can't really see the problem. Do people really still use
print()-methods for generating XML? If they do, this has to be
aggressively addressed. I don't know of any programming environment
today that doesn't have an XML library, so theres no excuse for not
using one.

Passing the data to an XML library before serving it to a client
SHOULD be enough for both levels of validations, but might only
be enough for level 1. But getting feeds past level 1 at least
gives Atom-readers the possibility to treat the data as XML; having
to parse it as pure string, doing all kinds of replace() and regex-
stuff on it is perishable.

____
[1] <url: http://www.gnksa.org/>
[2] <url: http://www.newsreaders.com/gnksa/cgi/gnksa-form.cgi>
[3] <url: http://www.gnksa.org/in-progress.txt>
[4] <url: http://www.gnksa.org/gnksa-evaluations.html>

-- 
Asbjørn Ulsberg           -=|=-          X-No-Archive: No
"He's a loathsome offensive brute, yet I can't look away"