[ODFPlugtest] MSO does not read LO files

[Andreas Guelzow]

On Tue, 2011-06-28 at 16:15 -0700, Dennis E. Hamilton wrote:

>  2. I find it strange that we are penalizing any consumer that is
> strict about how it consumes ODF 1.1 Manifests while insisting that
> ODF 1.2 producers make such manifests for no meaningful reason than to
> check off a tiny conformance box, and penalizing the ODF 1.1
> consumer.  

Am I confused. I thought we were talking about ODF1.2 files generated by
LibreOffice 3.4 (and other produceers.) As far as I know LibreOffice 3.4
does not generate ODF1.1 files. According to PDF 1.2 part 3 3.2 the
manifest file required in said package has to follow chapter 4 of the
same document, ie. it has to be an ODF1.2 manifest. Where are those
ODF1.1 manifests fro that you are mentioning? 

As far as I am concerned we have the situation that a ODF consumer
appears to read an ODF1.2 file and falsely claims it to be corrupt. I
think there are two acceptable behaviours:

(1) the consumer upon determining that the file is ODF 1.2 informs the
user that it does not support files in this format.

(2) the consumer tries to make an attempt to read the file (possibly
warning the user that since the consumer does not know the file format,
some information might be lost).

Any claim by a consumer that a file that advertises itself as an unknown
version is corrupt is absolutely not acceptable.

> We can do that.  I don't think it is a responsible, real-world
> approach.
>     Remember, for most of its life, the ODF 1.2 Specification had no
> such provision as <manifest:manifest> manifest:version.  So it is not
> like there was some particular need for it.  It was an afterthought.
> It apparently would not be missed if it disappeared entirely.
> 
> There is *nothing* in the ODF 1.1 specification that says how a
> manifest consumer is expected to be generous, or deal with tags and
> attributes and their values that are not understood and are not
> specified in ODF 1.1, no matter what namespace is used.  I'm pretty
> certain that there is nothing about that in ODF 1.2 Part 3.

All of this has absolutely nothing to do with ODF1.1. 
> 
>  3. My recommendation is that the attribute be omitted *unless* there
> is something in the manifest that is essential and is not provided for
> in ODF 1.1.  That is, the manifest.xml is ODF 1.1 compatible.  The
> attribute is useful where ignore-what-you-don't understand would get a
> down-level consumer in trouble, such as failing to recognize
> additional attributes that apply to decrypting an encrypted document.
> Of course, ignore-what-don't understand down-level consumers are going
> to ignore the attribute anyhow, aye?
> 
> I do think this is an useful topic for the Plugfest.

Several people mentioned that, and I fail to see how it can have
anything to do with a plugfest. 

This is not an issue of having applications agree on a certain
interpretation of ODF. ODF1.2 as currently written unequivocably
requires that attribute. 

Andreas


-- 
Andreas Guelzow <aguelzow at pyrshep.ca>