[odf-discuss] A story of TIFF

[Daniel Carrera]

On Sat, 2007-02-03 at 00:31 -0800, marbux wrote:
> Do you have proposal for mapping all of the ODF to Microsoft Office
> that does not require MIcrosoft's cooperation? Or any alternative to
> declaring an MS Office interoperability subset that can enable
> interop?

I think you missed the point. And others have answered your questions.
I'll answer below anyways, but I don't want to convey that you actually
got the point of the email because you didn't.

Here is the point:  If every application starts adding its own
extensions, and its own subsets, ODF will go the way of TIFF.
Interoperability will be lost. ODF will become little more than a
container format that can contain anything. Sure, each application will
be able to read its own files with perfect fidelity, but interop goes
out the window. And what's the point in that? If you are going to turn
ODF into some container format, little better than zip, you might as
well just let every application use its own proprietary format and save
yourself the hassle.

Now, to your off-topic questions:

* If there are features in an ODF file that MS can't render well, that
sucks for MS. It is not OK for each application to define which bits of
ODF they like and label it "Interop Subset". I don't want to see an
Abiword Interop Subset, and a KOffice Interop Subset, and a Writely
Interop Subset.

* If there are binary blobs you don't know how to convert to ODF, the
solution is not to extend ODF with binary blobs. Peter recommends one
of: (1) Print to PDF. (2) Keep 1 computer with a copy of MS Office. (3)
Retype the document. He has more relevant experience than anyone I know
on this list, and it sounds sensible, so I think I'll go for that.

> Also, I think your analogy to TIFF is inapt. You describe a
> specification that grew like topsy and became so complex that full
> support became infeasible.

And that's the way you'll take ODF if everyone starts adding their own
pet extensions. Especially if those extensions contain binary blobs. AT
LEAST the TIFF extensions were fully specified. The ones you are adding
are not.

> I see no facts stated that support your
> conclusion that it was the declaration of interop subsets that led to
> TIFF's downfall. The way you have described it, it seems to have been
> the infeasibility of full application support for the spec that was
> its downfall.

Infeasibility of full application support led to interop subsets.
Because of the mismatch between what each vendor chose to support, today
you can't use any of the interesting features of TIFF and expect interop
because I betcha that there is some vendor that doesn't support that
particular feature. It is all about "interop subsets". What happens when
Writely, AbiWord, KWord and OOoWriter each defines its own pet interop
subset? The only file you can produce as ODF and expect to be
interoperable is one that lies in the intersection of all of those. As
the number of applications grows, and more applications decide that it's
OK for them to define their own pet subset, the number of features you
can use and expect interop grows smaller.

Right now both OOo and KOffice developers are trying to fix their own
applications so they can support the full range of ODF. Praise to them.
That's the way to go. What you are doing, however, is a bad precedent.

I know you are about to repeat your question "what if I really need to
preserve this file and can't do it with ODF?". I will copy and paste
what I said above:
<daniel>
* If there are binary blobs you don't know how to convert to ODF, the
solution is not to extend ODF with binary blobs. Peter recommends one
of: (1) Print to PDF. (2) Keep 1 computer with a copy of MS Office. (3)
Retype the document. He has more relevant experience than anyone I know
on this list, and it sounds sensible, so I think I'll go for that.
</daniel>

Daniel.
-- 
May you live in interesting times.
May people in high places take notice of you.
May all your wishes come true.
  -- Chinese curse.