I concur with Rob's analysis with regard to approved errata.
Going forward, I recall seeing ISO annexes that provided cross-referenced
enumerations of
* all implementation-defined items (and these might include matters of
SHOULD and MAY)
* all conformance statements with identification of the party responsible
for satisfaction of the condition - e.g., format, consumer, producer, and
other cases.
The matter of responsibility might apply to implementaton-defined too,
particularly with respect to SHOULD and MAY, but I am not sure I've seen
that covered. I don't think CAN has to be dealt with because it is a
statement of fact, not permission or requirement, as far as I can tell.
The OASIS recommendations for conformance provisions in specifications
suggest that something along these lines is also desirable.
- Dennis
-----Original Message-----
From: [mailto:]
Sent: Thursday, September 24, 2009 05:57
To:
Subject: Re: [office] Suggestion from the ISO meeting in Seattle
Patrick,
I don't think we can mark something as "implementation-defined" in
Approved Errata, since that would be adding a new requirement (a
requirement for the implementation to specify its behavior) on existing
implementations, something we are not permitted to do under OASIS rules.
However, "implementation-dependent" would be fine.
Of course, in ODF 1.2 we are free to say "implementation-defined" if we
want.
-Rob
Patrick Durusau <> wrote on 09/24/2009 07:49:17 AM:
[ ... ]
