RE: [oic] interoperability issue with google-docs - mimetype

["Dennis E. Hamilton" <dennis.hamilton@acm.org>]

I already had an ODF Text document in Google Docs, so I downloaded it to see what the .odt package was like.

It shows up with correct mimetype file and it is the first file in the Zip stream.  It also has a correct <manifest:file-entry> for "/" with correct media-type.  

It is an ODF 1.1 document however.  There is this interesting comment added immediately after the <office:document-content> start tag in content.xml:

  <!-- Generated by Aspose.Words for Java 4.0.3.0 -->

It will be interesting to see what is different when a spreadsheet is produced.  I also wonder if there is a difference when the spreadsheet is uploaded from an user file and when it is created directly in Google Docs and then exported.

 - Dennis


-----Original Message-----
From: Dennis E. Hamilton [mailto:dennis.hamilton@acm.org] 
Sent: Sunday, October 09, 2011 14:19
To: 'Andreas J. Guelzow'; 'OIC TC List'
Subject: RE: [oic] interoperability issue with google-docs - mimetype

Andreas, which validator are you using, so the results can be confirmed and explored with additional spreadsheet document?

It seems that there is some reason to examine validators more closely as well.  What <office:document> office:version attribute value is on the export from Google Docs?  It matters because ODF 1.2 is much more restrictive concerning mimetype than ODF 1.0/1.1.

The mimetype file has never been mandatory directly.

ODF 1.0/1.1 MIMETYPE RESTRICTIONS

In ODF 1.1, a mimetype package file SHOULD be in the package and it SHOULD be the first file in the Zip stream. There are also mandatory limitations on the mimetype entry in the Zip concerning compression and the absence of an extra field.  In ODF 1.0/1.1 there is an implied relationship between the <office:document> office:mimetype attribute value.  I have not found any expressed condition whatsoever concerning a <manifest:file-entry> manifest:full-path="/" and its manifest:media-type value.


ODF 1.2 RESTRICTION AMBIGUITIES

In ODF 1.2 part 3 section 2.2.1, the statement is that there SHOULD BE a mimetype file for a conforming ODF 1.2 package.  This should produce at best a warning.  In section 3.3, however, the mimetype file SHALL be present if there is a <manifest:file-entry> with manifest:full-path="/" and their values must agree.  (This contradicts language in 3.2 which is permissive about manifest:full-path="/" in the absence of a mimetype file.) Also, section 3.3 says the mimetype file SHALL be the first file in the zip and it preserves the ODF 1.0/1.1 restrictions on the form of the mimetype entry in the Zip data stream.

However, in ODF 1.2 part 1 section 2.2.4, there is the condition that there SHALL be a mimetype file and the allowed values SHALL be one of the two prescribed for an OpenDocument Spreadsheet.  (Oddly, it need not match the value required for the <office:document> office:mimetype value, although both must be chosen from the same set of allowed cases.) 

It looks like there is a basis for some Errata and perhaps even an advisory or two with respect to these confusing interdependencies between Parts 1 and 3 and within Part 3 all by itself.

 - Dennis

-----Original Message-----
From: oic@lists.oasis-open.org [mailto:oic@lists.oasis-open.org] On Behalf Of Andreas J. Guelzow
Sent: Saturday, October 01, 2011 17:26
To: 'OIC TC List'
Subject: [oic] interoperability issue with google-docs

Hi,

perhaps somebody on this committee knows whom to contact with respect to
the fact that google-docs (the spreadsheet application) creates files
that it claims are ODF but that have lots of issues, the main one being
that it omits the mimetype file from the package (beside a bunch of
other schema violation):

upload:///GD-spread.ods[-1,-1]:Error:The 'mimetype' file have to be
contained in the ODF XML Schema package 'upload:///GD-spread.ods'!

[ ... ]