[ODFPlugtest] MSO does not read LO files

Dennis E. Hamilton dennis.hamilton at acm.org
Tue Jun 28 17:58:02 CEST 2011


Since my subscription request is hung up in a moderator queue, my moderated post is going to break the conversation thread.  Sorry, it is not in my power to deal with that.  (It would be great if the SSL certificate on the archive didn't fire up my browser warnings though.)

First, yes, I do propose that ODF 1.2 implementers *not* use <manifest:manifest> manifest:version="1.2" *unless* some feature of the manifest that is defined only for ODF 1.2 is used in that manifest.  Then you need to know that.  As far as I know, the only critical case is if extensions to encryption provisions beyond what was provided for in ODF 1.0/1.1 are used.  That's a very localizable case.  

I suspect a validator won't mind, unless it cross-checks for a manifest:version="1.2" when it finds an office:version="1.2" attribute in the actual XML content.  If it were my validator, that would be a warning.  (It just falls into one of the various conformance crevasses in ODF 1.2.)

Secondly, I disagree with Andreas about this.  The ODF 1.2 specification is not an OASIS standard yet and although the likelihood of it being so before the end of this year (and an ISO International Standard next year) is quite high, I say that the making of that attribute and its value mandatory was a blunder, since there was no previous versioning of the manifest.xml.  I would hope we can deal with that as an errata and that we certainly deal with this better in ODF 1.3.

Furthermore, ODF 1.1 said *nothing* about provisions for potential up-level issues (nor does 1.2) so retroactively imposing such a condition is unforgiveable (and I have a stronger word for it in my private thoughts).  Furthermore, ODF 1.2 says nothing (and can say nothing) about what ODF 1.1 consumers do with ODF 1.2 documents.  The fact that some implementations got lucky and one that was following the rules didn't is happenstance and not an occasion for celebration or dismissal of where the breakage occurs.

Yes, I think interoperability trumps slavish adherence to a specification (defect).  So, I suspect, do the users who run into this.

This is not about who will support ODF 1.2 and when they will do it, since that is independent of this particular question.  It has no impact on anything else about ODF 1.2.

 - orcmid





More information about the Plugtest mailing list