I also support this idea.
In the proposal at hand, there are two places where implementation-defined
is appropriate. One has to do with saying
"Alternatively, a namespaced value may be used for implementation-defined
preferred views."
The other might be to say
"The behavior of application software is implementation-defined for cases
where manifest:preferred-view-mode is absent or the value is one not
provided for."
I suppose the first case applies to emitters and the second seems to apply
to accepters.
- Dennis
-----Original Message-----
From: [mailto:]
http://lists.oasis-open.org/archives/office/200810/msg00161.html
Sent: Wednesday, October 29, 2008 14:09
To:
Subject: Re: [office] preferred view mode upon opening document
> 2. What about a sentence on the specification like: "It is
> recommended that all custom view mode specification documentation is
> made available to enhance documents interoperability" ?
>
The issue Warren raises is a general issue, not exclusive to view mode
alone. One way to address this globally is two add two things to the ODF
text:
1) Where ever something like this occurs, explicitly call the behavior
"implementation-defined". Use that as a uniform label of areas where ODF
allows implementations to implement their own behaviors. So we say
something like "Additional view modes may be used, but they shall be
prefixed by a namespace qualifier not defined by this standard. The
behavior of such view modes is implementation-defined".
2) Then in the conformance clause, we add language like: "An
implementation shall be accompanied by a document that defines all
implementation-defined and locale-specific characteristics and all
extensions." (That is the language directly out of ISO/IEC 9899:1999 C
Programming Language. You can see here what one implementation created to
define its "implementation-defined" features:
http://gcc.gnu.org/onlinedocs/gcc/C-Implementation.html
We don't need that exact language, but you get the idea. A standard can
allow implementation-defined behavior, but can also require that these be
documented. If we're going to do this, I'd rather we do it pervasively in
the standard rather than just in the view mode attribute.
What do TC members think?
[ ... ]
