Re: [office] consensus suggestion

[David Faure <faure@kde.org>]

On Monday 19 March 2007, Florian Reuter wrote:
> Hi David,
> 
> The "global" list-id table is needed to get unambigious roundtrip between text:list and text:numbered-paragraph:
I would simply use the style of the first numbered-paragraph of each list:id, as the style you have put in the list-id-table.
After all, an override is there to override previous stuff, so the first paragraph doesn't need to be an override.

> <text:list text:style-name="L1">
> <text:list-item text:list-override="L2"><text:p>P1</text:p></text:list-item>
> <text:list-item><text:p>P2</text:p></text:list-item>
> <text:list-item><text:p>P3</text:p></text:list-item>
> <text:list-item><text:p>P4</text:p></text:list-item>
> </test:list>
OK, good example, to counter my above idea ;) ... but see below.

> would map into
> 
> <text:numbered-paragraph text:list-id="id1" text:style-name="L2"><text:p>P1</text:p></text:numbered-paragraph>
> <text:numbered-paragraph text:list-id="id1" text:style-name="L1"><text:p>P2</text:p></text:numbered-paragraph>
> <text:numbered-paragraph text:list-id="id1" text:style-name="L1"><text:p>P3</text:p></text:numbered-paragraph>
> <text:numbered-paragraph text:list-id="id1" text:style-name="L1"><text:p>P4</text:p></text:numbered-paragraph>
> 
> which then would map back to (according to Oliver's/Thomas' proposal)
> 
> <text:list text:style-name="L2">
> <text:list-item><text:p>P1</text:p></text:list-item>
> <text:list-item text:list-override="L1"><text:p>P2</text:p></text:list-item>
> <text:list-item text:list-override="L1"><text:p>P3</text:p></text:list-item>
> <text:list-item text:list-override="L1"><text:p>P4</text:p></text:list-item>
> </test:list>

Hmm. I would say "so what?". This list and the initial list are exactly equivalent in how they
appear to the user. There might be differences when editing this list later on, but well, I can find
a thousand examples where opening a OOo file in kword, saving, and reopening it in OOo
leads to a document that looks the same, but that behaves differently during editing.
I always considered that "same rendering" was a requirement, but not "same behavior
while editing". After all the user is having control while editing and can adjust what he's
doing to how the application reacts (e.g. in the above case he can choose whether he wants
to continue the list with L1 or L2 paragraphs). Same rendering is the important part.

> With global list-id table
> <list-id-table>
> <list-id-definition list-id="i1" style-name="L1">
> </list-id-table>
> 
> this disambiguity would be solved.

I would make it clear that the only purpose of list-id-table is to improve
compatibility with the <text:list> lists, and I would make it completely optional.
This way, someone generating a simple document (that isn't really going to be modified, but
just printed) with a perl script and using numbered-paragraph, doesn't have to care about
generating <list-id-table>. 

But wait...... an implementation like KOffice which is based on numbered-paragraph,
does not have the notion of which style to put in that table! This is the main reason
why this table is, well, a solution to a very theoretical problem. Yes it is needed for
full roundtrip, but an application which treats numbered-paragraphs independently
is never going to have the information about which style is the "main style" in 
the first place.

So in which case is the full roundtrip compatibility going to be useful?
I don't object to list-id-table (assuming it's fully optional) if there's a case where it might
be useful, but I'm stressing the fact that most documents created with numbered-paragraphs
are never going to have that table since they can't, by design, know what to put there.


BTW on the call there was a concern that Oliver's proposal was conflicting with yours,
and that it had to be one or the other but at least on this topic they seem quite complimentary... 
list-id-table adds well to style-override, as you show by using style-override in your examples ;)

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