Re: [office-comment] Re: my comments on the numbering spec proposal discussed by David F. andFlorian R.

[David Faure <faure@kde.org>]

On Monday 11 December 2006 14:42, Oliver-Rainer Wittmann - Software Engineer - Sun Microsystems wrote:
> Hi,
> 
> Thomas Zander wrote:
> > I read the mailing list archives and wanted to comment on this post from 
> > Oliver.
> > http://lists.oasis-open.org/archives/office/200611/msg00108.html
> > 
> > Oliver wrote Thu, 30 Nov 2006 13:50:34:
> >> The text:style-override approach seems to more natural to me. I've got
> >> some numbered paragraphs in a certain list and wants to apply another
> >> label format to a part of these paragraphs.
> > 
> > There is a big assumption in this paragraph that I'd like to address.
> > You state that you start out with 'numbered paragraphs in a certain list'. And 
> > go from there.
> > In reality there is no way to explicitly say that a number of paragraphs 
> > belong to a list in the examples given (using text:numbered-paragraph). The 
> > assumption seems to be that all numbered paragraphs that belong to a certain 
> > style belong to the same list.
> > 
> Yes, that's my interpretation of the list style. All paragraphs, which 
> apply list style "L1" belong to the same list.

This is the current spec, but this interpretation would in fact make very little sense to the end user.
Imagine you have a very large document, with many headings, and at page 3 under 1.1.2 you have a simple
a/b/c/d list, and at page 100 under heading 18.5.6 you have another simple a/b/c/d list.
How is that "the same list"? Conceptually those are certainly very distinct lists, despite the fact
that they use the same a/b/c/d style.

One could certainly define two styles L1 and L2 that both use a/b/c/d, but this isn't about style
anymore then, it's about identifying lists, and this is what I propose to separate out into list-id.

Styles are supposed to be reuseable, to avoid having to define two styles with the exact
same properties. For e.g. text properties this is exactly the case already. But for styles,
if reusing a given style for two lists isn't possible, then it's inconsistent and messy.

> That's already the case for the headings in the text document, which all 
> by default apply the outline style - the default list style for outline 
> numbering.

Outline numbering is a bit different since it is by definition scattered across the
whole document, and there's only one list (although it would be useful to be able
to switch styles on it, for the case of 1 / 2 / 3 / A / B annexes).

> > Now, this means that having 5 lists, and I want to structure my ODF document 
> > to make it clear there are 5 lists, that I would need 5 different styles.
> > That completely breaks the concept of separating content from markup.
> > 
> I didn't understand your conclusion here.
> Why does the usage of different list styles for different lists breaks 
> the concept of separating content from markup?

Because the style ends up defining the structure of the document (which paragraphs
belong to which lists). Even <text:list> doesn't really define that, since multiple text:list
elements can in fact be part of the same list, conceptually, right?

> > I think that the suggestion that Florian made is based on visual appearance. 
> > Which is correct if we want ODF to be used as a fileformat for a text. But it 
> > breaks down if you want to use the text for more structural purposes. Like 
> > datamining it. Why? Because 2 separate lists using the same style can no 
> > longer be separated as their style is their 'list id' in the 
> > text:style-override proposal.
> > 
> If you want to use the same list style for two separate lists, then you 
> have to define for the second list, that the numbering has to restart.
That's what shouldn't be necessary, it looks like a hack to make the current concept work.
See my example above of two completely unrelated lists.

> But, in my interpretation - which is somehow technical - of the list 
> style the paragraphs of these two list still belong to the same list.
To the user, and to me, (and to most people who will generate ODF xml), they don't.

> As state in the ODF standard, the <text:list> way and the 
> <numbered-paragraph> way are equivalent. Thus, I think we should choose 
> a solution that fits both on the same way. Otherwise the equivalence 
> isn't as obvious as it is now.
Well they are almost equivalent. But <text:list> defines structure (when it's not misused
like OOo misuses it, cf older mail from me), whereas <text:numbered-paragraph> by definition
doesn't define structure (since it's about independent paragraphs), so having a way to provide
structure for text:numbered-paragraph using list-id, makes sense.

> What about my concern about the backward-compatibility of the <list:id> 
> approach?
Given the current non-implementation of <text:numbered-paragraph> outside of KOffice,
I'm much more concerned about getting it right than about backward-compatibility.
OOo doesn't care for text:numbered-paragraph currently so backwards compability
as an argument from OOo people makes no sense ;)

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