OASIS Open Document Format for Office Applications (OpenDocument) TC

 View Only
Back to discussions
Expand all | Collapse all

Res: RE: [office] Conformance Clause proposal, Version 8

  • 1.  Res: RE: [office] Conformance Clause proposal, Version 8

    Posted 02-09-2009 00:20 
    Advancing a little bit on this discussion, I believe that should be very useful to ODF users if we define a way to clearly (and transparently) define the kind of ODF document they're dealing with (just in case that more than one conformance clause be approved).

With this scenario in mind, I would like to propose that we define different file extensions and/or mimetypes to identify documents with contents based on different conformance clauses.

If Rob's proposal would be accepted, I believe that the nomenclature OpenDocument Extended should be used, meaning extensions as .odxt, odxs, .odxp and so on, but we also should define mimetypes to allow the differentiation on XML (not packages) ODF documents.

I'm waiting your comments (and I bet US$100 that Dennis will have several ones :) ),

Jomar 
------Mensagem original------
De: Rob Weir
Para:office@lists.oasis-open.org
Assunto: RE: [office] Conformance Clause proposal, Version 8
Enviada em: Fev 7, 2009 14:12

Sorry, Dennis.  "Conformable" as a term doesn't work well.  I don't think 
that we should, in an international standard, be using subtle grammatical 
variations to express important distinctions in terminology, especially 
when introducing a novel term.   There is too great a risk that it will 
not stand up to translation to other languages, especially when we already 
have "conforming", "conformance" and "conformant" as terms.

But I'm certainly open to alternatives.  Why not something like "Extended" 
conformance class?  That seems clear enough.

-Rob

"Dennis E. Hamilton"


  • 2.  RE: RE: [office] Conformance Clause proposal, Version 8

    Posted 02-09-2009 06:21 
    Jomar,

I think that is a great idea.  

Because file extensions are not defined normatively, we might need to do
something different there (but make more non-normative suggestions about
extensions that are in use).  There is precedent for decoration of MIME
types to make variations though. E.g.,
application/vnd.oasis.opendocument.text+ext.

I suppose one could decorate the office:version attribute too, with
something like office:version="1.2 ext".  

I also think it would be wonderful if the definition of document types, and
the schema relationship to the different types, were nailed down
normatively. 

I'm not sure that these could be made any more mandatory than the current
mimetypes are, but it would be interesting to nail something like this down.

 - Dennis

PS: Do I win the $100?

PPS: Since the current mimetypes are already in use for documents that are
conformant in ODF 1.0/IS 26300/ODF 1.1 terms, and certainly all spreadsheet
documents have been using at least one foreign namespace, maybe we should go
the other way and make new mimetypes for the new conformant-document level.
That is, "application/vnd.oasis.opendocument.text-ext" and ".odts" or
".odtl" or somesuch?

PPPS: I have already said that I would go along with "Extended Document,"
although a better term would be welcome.


    
                                                
    
					
                                                    

        
                                                
				
    
                                                
    
                                                    
                                                        
                                                        Original Message


  • 3.  Re: [office] Conformance Clause proposal, Version 8

    Posted 02-09-2009 10:57 
    Dennis,

On 09.02.09 07:20, Dennis E. Hamilton wrote:
> Jomar,
> 
> 
> PPS: Since the current mimetypes are already in use for documents that are
> conformant in ODF 1.0/IS 26300/ODF 1.1 terms, and certainly all spreadsheet
> documents have been using at least one foreign namespace, maybe we should go
> the other way and make new mimetypes for the new conformant-document level.
> That is, "application/vnd.oasis.opendocument.text-ext" and ".odts" or
> ".odtl" or somesuch?

I think there is some mis-understanding: Whether a document contains 
declarations for foreign namespaces does not matter. What matters is 
only if this namespace is used as a namespace for an element or an 
attribute. Or to say it differently: What matters is whether a document 
is valid regarding the ODF schema we define. A document that adds some 
namespace declarations remains valid. A document that adds elements and 
attributes is not valid.

That means, even if you take the strict conformance definition that does 
not allow foreign elements and attributes, you can add as many namespace 
declaration as you want, and you can use them in as many attribute 
values as you want.

I have in principle no objections of having multiple Mimetypes for the 
conformance levels, but I'm not convinced whether having multiple 
extensions actually avoids confusion or adds some for the user.

Best regards

Michael


> 
> PPPS: I have already said that I would go along with "Extended Document,"
> although a better term would be welcome.
> 
> 
    
                                                
    
					
                                                    

        
                                                
				
    
                                                
    
                                                    
                                                        
                                                        Original Message


  • 4.  Re: [office] Conformance Clause proposal, Version 8

    Posted 02-09-2009 11:10 
    On Monday 09 February 2009, Michael Brauer - Sun Germany - ham02 - Hamburg wrote:
> That means, even if you take the strict conformance definition that does 
> not allow foreign elements and attributes, you can add as many namespace 
> declaration as you want, and you can use them in as many attribute 
> values as you want.

This is a rather strange notion of strict conformance, isn't it?

 in not allowed, but


  • 5.  Re: [office] Conformance Clause proposal, Version 8

    Posted 02-09-2009 11:47 
    On 09.02.09 12:09, David Faure wrote:
> On Monday 09 February 2009, Michael Brauer - Sun Germany - ham02 - Hamburg wrote:
>> That means, even if you take the strict conformance definition that does 
>> not allow foreign elements and attributes, you can add as many namespace 
>> declaration as you want, and you can use them in as many attribute 
>> values as you want.

I should have been clearer. Of cause, one can use namespaced values only 
for those attributes that allow namespaced values, like the 
table:formula attribute.
> 
> This is a rather strange notion of strict conformance, isn't it?
> 
>  in not allowed, but
>


  • 6.  Re: [office] Conformance Clause proposal, Version 8

    Posted 02-09-2009 11:54 
    On Monday 09 February 2009, Michael Brauer - Sun Germany - ham02 - Hamburg wrote:
> one can use namespaced values only 
> for those attributes that allow namespaced values, like the 
> table:formula attribute.

OK, good.


PS: I will probably be late for today's meeting, not sure by how much.

-- 
David Faure, faure@kde.org, sponsored by Qt Software @ Nokia to work on KDE,
Konqueror (http://www.konqueror.org), and KOffice (http://www.koffice.org).


  • 7.  RE: [office] Conformance Clause proposal, Version 8

    Posted 02-09-2009 17:01 
    I also think we are having a language problem.

I wasn't referring to how many xmlns:foo=... and xmlns=... declarations
there are but only if a foreign one is actually used, that is, an element,
attribute, or value that depends on that namespace occurs in the ODF
document.  Also, though I failed to mention it, I am also excluding those
foreign namespaces that are only used, via CURIES, to compose URIs in RDF
attributes, and not actually used as namespaces. 

I didn't mean to suggest it was the declaration of the namespace that was
important, but the reliance on the namespace as a namespace (excluding use
in forming URIs via the RDF CURIE mechanism, where there is no reliance on
the namespace URI as anything but a string).

Also, of course, the special use in ODF of QName prefixes within attribute
values is not subject to schema validation, as far as I know, since I think
we always say those attributes are of type string.  I suppose we can take
that up when we look at those cases, including table:formula, to see exactly
what is in and what is out at what level.

Just the same, I *do* count the use of QName prefixes in attribute values as
an use of a namespace.  It is not something trivial like a CURIE, it is a
determination of how the rest of the value is to be interpreted.  Sounds
like application/use of a foreign namespace to me.  We've not even required
that such a namespace usage be documented in any way, but it does determine
the syntax and other rules for the value of the attribute and it does
require that there be a related xmlns declaration.  If it walks like a
namespace, quacks like a namespace, and is called a namespace, and if it is
not an ODF namespace, why isn't it a foreign namespace?  Of course, ignoring
the attribute breaks the formula dependency of the value in the cell, but we
have given no guidance on what else to do that might work.  I have seen one
ODF 1.1 implementation that ignores any prefix in table:formula,
substituting one of its own choosing.

 - Dennis

PS: I really do like the idea of having new MIME types as a promise to
accompany the new stronger conformance level.  I think Jomar's idea is a
good one.  It is not necessary to use them, of course, since the stronger
conformance level is a subset of the weaker one.  I would object to using
the existing MIME types to signify the stronger conformance level because of
forward/backward compatibility issues.  Backward compatibility is probably a
reason not to monkey with the MIME types at all, though.  Because of that,
the current MIME types should not differentiate conformance level.  Ah well
... .


    
                                                
    
					
                                                    

        
                                                
				
    
                                                
    
                                                    
                                                        
                                                        Original Message


  • 8.  Re: [office] Conformance Clause proposal, Version 8

    Posted 02-09-2009 11:29 
    On Monday 09 February 2009, Dennis E. Hamilton wrote:
> Because file extensions are not defined normatively

They are, actually, but not in OpenDocument - they are defined normatively at IANA,
based on a submission by this TC.

> application/vnd.oasis.opendocument.text+ext.

I don't like this - it will break existing applications. And I haven't seen a justification
of _why_ we should do this. What is the goal there?

> I suppose one could decorate the office:version attribute too, with
> something like office:version="1.2 ext".  

Again, I have to ask about the purpose of this.

On the one hand we are saying that there is no real need for foreign elements/attributes
(my examples of existing foreign attributes have been answered with better
conformant ways of doing it), only for maybe style-*-properties extension,
and on the other hand we want to define a whole class of documents
(including a different mimetype and a different extension) for documents
with foreign elements/attributes? This doesn't make sense to me.

Unless I'm missing something, there's a huge contradiction between
"extensions are not needed" and "let us define extended documents".

Or is this about style-*-properties extensions? I have to name koffice-produced
files *.odxt just because I save a harmless koffice:frame-behavior-on-new-page
style attribute in the style properties, and because of this file extension/mimetype
those documents will not be opened by any of the existing OpenDocument
processors out there? This makes no sense. We are not changing the mimetype
for each version of ODF either, even though the contents are slightly different,
that's the whole point of XML's extensibility: a ODF-1.1 processor can give a shot
at parsing a ODF-1.2 document, so it can also give a shot at parsing a ODF-1.2
document "with a few extensions", can't it? A new mimetype prevents that completely
(and will confuse users for no apparent benefit).

Let us define one thing: a document is conformant if it validates against
the schema (=> easy yes/no result), and in the schema we allow extensions
in style-*-properties and metadata.

-- 
David Faure, faure@kde.org, sponsored by Qt Software @ Nokia to work on KDE,
Konqueror (http://www.konqueror.org), and KOffice (http://www.koffice.org).


  • 9.  Re: [office] Conformance Clause proposal, Version 8

    Posted 02-09-2009 12:19 
    On 09.02.09 12:28, David Faure wrote:
> On Monday 09 February 2009, Dennis E. Hamilton wrote:
>> Because file extensions are not defined normatively
> 
> They are, actually, but not in OpenDocument - they are defined normatively at IANA,
> based on a submission by this TC.

This is correct, except that it is OASIS who submits them to IANA.
> 
>> application/vnd.oasis.opendocument.text+ext.
> 
> I don't like this - it will break existing applications. And I haven't seen a justification
> of _why_ we should do this. What is the goal there?

After having thought about this for a while, I agree.

At least the extensions for documents that do not contain foreign 
elements should remain the same.

And for those that contain foreign elements: Well, I'm not sure whether 
one mimetype would be really sufficient. A vendor A may want to differ 
documents that use its own foreign elements from the documents that 
contains other foreign elements. So, every vendor may actually want to 
register its own mimetypes and extensions.

Please note that this is permitted anyway, regardless whether we define 
a "loose" conformance mode or not.

> Or is this about style-*-properties extensions? I have to name koffice-produced
> files *.odxt just because I save a harmless koffice:frame-behavior-on-new-page
> style attribute in the style properties, and because of this file extension/mimetype
> those documents will not be opened by any of the existing OpenDocument
> processors out there? This makes no sense. We are not changing the mimetype
> for each version of ODF either, even though the contents are slightly different,
> that's the whole point of XML's extensibility: a ODF-1.1 processor can give a shot
> at parsing a ODF-1.2 document, so it can also give a shot at parsing a ODF-1.2
> document "with a few extensions", can't it? A new mimetype prevents that completely
> (and will confuse users for no apparent benefit).
> 
> Let us define one thing: a document is conformant if it validates against
> the schema (=> easy yes/no result), and in the schema we allow extensions
> in style-*-properties and metadata.
> 

I would like to keep the questions whether there should be an 
extensibility mechanism for *-properties separate. For details, see one 
of my previous mails.

Best regards

Michael


-- 
Michael Brauer, Technical Architect Software Engineering
StarOffice/OpenOffice.org
Sun Microsystems GmbH             Nagelsweg 55
D-20097 Hamburg, Germany          michael.brauer@sun.com
http://sun.com/staroffice         +49 40 23646 500
http://blogs.sun.com/GullFOSS

Sitz der Gesellschaft: Sun Microsystems GmbH, Sonnenallee 1,
	   D-85551 Kirchheim-Heimstetten
Amtsgericht Muenchen: HRB 161028
Geschaeftsfuehrer: Thomas Schroeder, Wolfgang Engels, Dr. Roland Boemer
Vorsitzender des Aufsichtsrates: Martin Haering


  • 10.  RE: [office] Conformance Clause proposal, Version 8

    Posted 02-09-2009 17:01 
    No, the definition of file extensions is not a normative part of the IANA
registration.   I looked at the form.  The file extensions section is
optional and simply identifies common file extensions that are in use.  It
does not reserve them or anything.

Jomar's and my observations about this are only material in the case that
there are two levels of conformance defined for ODF 1.2, but I agree that
changing MIME types will completely screw up backward compatibility and we
should not do that.

 - Dennis

PS: I know that OO.o, for example, does not care what the file extension is.
If it is a Zip package, it uses the MIME type to figure out what kind of
document it is and then it opens it properly.  The file extension, on
Windows, will determine whether OO.o is launched automatically, but OO.o has
its own way of deciding what it has been fed.


    
                                                
    
					
                                                    

        
                                                
				
    
                                                
    
                                                    
                                                        
                                                        Original Message


  • 11.  Re: Res: RE: [office] Conformance Clause proposal, Version 8

    Posted 02-09-2009 11:01 
    On Monday 09 February 2009, Jomar Silva wrote:
> If Rob's proposal would be accepted, I believe that the nomenclature OpenDocument Extended should be used, meaning extensions as .odxt, odxs, .odxp and so on, but we also should define mimetypes to allow the differentiation on XML (not packages) ODF documents.

This is the best way to break interoperability completely (with existing
versions of the OpenDocument applications).

-- 
David Faure, faure@kde.org, sponsored by Qt Software @ Nokia to work on KDE,
Konqueror (http://www.konqueror.org), and KOffice (http://www.koffice.org).


Related Content