RE: [office] Proposal for lists/numbered paragraphs

["Philip Boutros" <Philip.Boutros@stellent.com>]

Hi David

On reviewing the schema I see you are correct that paragraph style can
reference a list style which allows (but does not require) a connection
to be established between para and list styles. This brings up another
issue for me. Doesn't this produce redundancy and potential conflict
between the list style associated at the text:unordered-list and the
list style associated at the paragraph level. OpenOffice currently
generates XML like this...

<office:automatic-styles>
   <style:style style:name="P1" style:family="paragraph"
style:parent-style-name="First line indent" style:list-style-name="List
4"/>
</office:automatic-styles>

        <text:unordered-list text:style-name="List 4">
            <text:list-item>
                <text:p text:style-name="P1">One</text:p>
            </text:list-item>
            <text:list-item>
                <text:p text:style-name="P1">Two</text:p>
            </text:list-item>
            <text:list-item>
                <text:p text:style-name="P1">Three</text:p>
            </text:list-item>
        </text:unordered-list>


Notice that "List 4" is referenced both by the text:unordered-list and
by the "P1" paragraph style. What if "P1" referenced a different list
style? How would I be required to interpret this? 

Since the paragraph style already contains the list style information,
from a rendering standpoint in this example the text:unordered-list is
completely redundant and the text:list-item is simply defining a list
level. List level could be easily done with an attribute (which could
default to 1) producing the following alternative XML. 

<text:p text:style-name="P1">One</text:p>
<text:p text:style-name="P1">Two</text:p>
<text:p text:style-name="P1">Three</text:p>

This all seems like a lot of extra syntax just so HTML generation can
eaisly produce <OL> and <LI> tags. 

-Phil



-----Original Message-----
From: David Faure [mailto:faure@kde.org] 
Sent: Friday, March 21, 2003 10:57 AM
To: office@lists.oasis-open.org
Subject: Re: [office] Proposal for lists/numbered paragraphs


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Tuesday 18 March 2003 21:34, Philip Boutros wrote:
> Hi David
> 
> When we "add" structure we are also removing the connection between 
> the paragraph style and the list style/counter. Let say in Word I have

> a style called "phil" to which I have associated a list style that 
> makes the each paragraph that has the  "phil" style display with a 
> letter in front of it (A, B, C, etc.) When going to openoffice the 
> list style is preserved but would have to be decoupled from the 
> paragraph style.

Are you sure this is what happens? If your style called "phil" has all
the information necessary for those paragraphs, then the paragraphs will
have the "phil" style, there will be no additional "list style" created.
So there would be a single style, no "connection lost".

The OO file format calls "style" any collection of settings, with a
distinction between automated-styles and user-created-styles. Your
"phil" style would be one of the latter group, but that's all. I don't
see why a separate "list style" would be created, if the "phil" style
already says "number those paragraphs with A, B, C in front".

> Plus each paragraph of the "phil" style would have to be wrapped in 
> <list> elements.

Well, each set of consecutive paragraphs of that style, more precisely.
The point the OO people make is that most of the time they will be
consecutive indeed. Except when people use numbered paragraphs in a case
where they should 
use headings, of course.... I found that OO makes that difficult,
probably on purpose, but other word processors don't.

> This is a loss of information because what I want as a user is that 
> anything I style
> with "phil" has A, B, C, etc. in front of it. 

Sure. But since the "phil" style says so, it would work.

> Bigger Picture
> --------------
> What openoffice has done with the <list> element is attempt to take 
> the very
> simplest case of "numbered paragraphs" in word processors and turn
them 
> into structure so they can be easily processed. 
Yes.

> In doing so they have complicated all the other cases of "numbered 
> paragraphs" and were forced to add a whole new mechanism to deal with 
> "numbered headings".

Yes, I totally agree here. This surprised me much too.

However the elements for specifying the kind of numbering etc. are
shared, so it's not really a "whole new mechanism".... Only a different
top-level tag.

> I again recommend we abandon this well intentioned but problematic 
> scheme and
> model "lists" as the major word processors do. This not only unifies
all 
> "numbered paragraph" in the schema but also allow better transfer of
the user's intention 
> from the major word processing formats.

This was also my initial opinion....
But if we can express everything with both mechanisms, i.e. if they are
totally equivalent on a semantic level, then it doesn't really matter
which way we 
choose, it's only a matter of "whatever makes the implementation easier"
(of the word processors, and of the filters - with a conflicting goal
there ;).

Since I'm not convinced by your example of a semantic difference (loss
of information), I'm still with the position that both representations
are equivalent, for now...

- -- 
David FAURE, faure@kde.org, sponsored by TrollTech to work on KDE,
Konqueror (http://www.konqueror.org), and KOffice
(http://www.koffice.org). How to write a Makefile.am for KDE/Qt code:
http://developer.kde.org/documentation/other/makefile_am_howto.html
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE+e0Rq72KcVAmwbhARAnbkAJ4uxKyBXqRBTepMkF7NvrM3MniXlACgsesb
Wj4sY25OwHx14EE+b+Rup9s=
=AkLW
-----END PGP SIGNATURE-----