Re: [office-comment] Foreign elements and attributes

[rjelliffe@allette.com.au]

> Your anecdotes do not support your argument.  If I'm reading it correctly,
> you are saying that we should allow extensions because extensions are
> useful to integrators.

Yes.

The current committee draft text is correct in not allowing readers that
don't cope with foreign elements. The idea that foreign-namespace markup
is necessarily dangerous is wrong. Indeed, it also has positive uses:
foreign namespace markup is an escape valve for integrators. It maybe a
tradeoff, except that the remedy of banning foreign markup does not
actually remove the danger, so it looks like a misdirection of effort. It
will look like something has been done, without anything actually being
done.

For example, does ODF prevent XML processing instructions PIs? These are
part of XML, and they can be used almost anywhere in a document, with
absolutely no constraints imposable by schema languages (actually,
Schematron can.) PIs are intended to provide processing-specific or
application-specific instructions. They are yet another reason why merely
blocking foreign elements and attributes will not actually achieve the aim
of somehow making ODF extension-proof or circumventable.

> But then you list three examples of how you were
> not able to accomplish what you wanted in ODF 1.1, even though ODF 1.1
> allows these extensions.

No. I wrote
"I have had three use cases where this issue was a showstopper for ODF,
due to the incorrect implementation of this feature."

I.e. the old standard was good in what it said (even though they were
terribly expressed), but because of the failure of OpenOffice in
particular to implement the standard (quite possibly because of the
inadequate drafting), the feature could not be used.

By the way, Rob originally asked the question "So if it is such a useful
feature, how come in 3 1/2 years no one has bothered to find a use for it
or to support it in their products?" which I did not take to be a
rhetorical question but something that could be answered by real uses
cases (what Rob calls "anecdotes".) I hope the TC is run on speculation
and pure logic rather than real use cases.

Cheers
Rick Jelliffe