RE: [opendocument-users] simple OO.org document goes awry in MS Office 2007 w/SP2 - what went wrong?

[Hanssens Bart <Bart.Hanssens@fedict.be>]

Hi Jan,


> First he should inform the vendors and they should talk to the TC, OIC if
> needed. Becauae effectively the implementors will have to fix their
> products. John can "only" assist by testing, he *could* help patching OOo
> if  needed as it is Open Source, but he would not gain much from talking
> to the TC.

A) At the plugfest we encountered a specific bug with embedded objects (chart in spreadsheet) where the spec was clear how to do it.

In this particular case, the bug is in OpenOffice.org and products based upon it, or following it, while Microsoft Office 2007sp2 does it correct.

So, what vendors can do to increase interop *in this specific case* (since it was all about having a URI with a slash or not, trivial to fix IMHO) is to be liberal in what they accept and strict in what they produce (sounds familiar, doesn't it).

What the OIC *can* do is write a recommendation, sort of like "this is a real world problem, this is what implementer A *can* do when loading invalid docs from implementer B, for the sake of real world interoperability. But in any case implementer B *must* fix the bug" (without calling names)


I'm not saying we can/will do this for each and every specific case, and I'm also not saying that each and every time we should "recommend" (for whatever value of recommend) workarounds. But *sometimes* it makes sense when it solves real world problems and the fix is trivial.



B) On a related note: if I recall correctly, there was a discussion on saving default values like page margins. Not an implementation bug, but an area where the spec isn't clear and/or could be improved.

People expect that their document is rendered similarly when opening it in another application in a similar context (like: desktop -> desktop). There are a lot of issues surrounding fonts etc, but at least your implementation can store "page margin is 2cm" inside the document, instead of relying on the implementation-specific (or user specific) default setting.
The user on the receiving end *may* want to override these settings, but at least the values are stored, so another implementation may reproduce it.


These things can also be documented by the OIC in an "interoperability recommendation", or whatever you'd like to call it, and *at the same time* be sent to the ODF TC, to make sure it will be specified in the next version of the spec.


Best regards,

Bart