Re: [office] list-override proposal

["Florian Reuter" <freuter@novell.com>]

Hi Oliver,

/me is also disappointed that I don't really understand your proposal. However /me has not yet given up ;-)

Regarding the schema you proposed in the name of Thomas and David I have the following question:

In order to be backward compatible with ODF1.0/1.1 the list-id would have to be optional I guess. Is this mandatory by accident or by intention?

The other thing I have problems with is this:
I assume that adding a list-override to the numbered-paragraph would solve the same problem as the list-id, right?

So from a practical point of view I consider a list-id to be more harmfull for backward compatibility than a list-override.
By using a list-override the worst thing which can happen is that the number is formatted wrong.
By using a list-id the worst thing which can happen is that the actual number is wrong.
This is why I prefer a list-override.

So Thomas and you have argued that a list-id is "better" for reasons of "you like it more". And I agree. If we would build ODF1.0 I would agree. However there has been a lot of promise around ODF beeing stable for decades and even centuries.   And I just think that a list-override serves this issue better.

For example. Let suppose we want to express the following numbered paragraphs
1. Par1
B. Par2
3. Par3

With list-id we would get
<numbered-paragraph list-id="1" style:name="L1">Par1</numbered-paragraph>
<numbered-paragraph list-id="1" style:name="L2">Par2</numbered-paragraph>
<numbered-paragraph list-id="1" style:name="L1">Par3</numbered-paragraph>

and with list-override we would get
<numbered-paragraph style:name="L1">Par1</numbered-paragraph>
<numbered-paragraph style:name="L1" list-override="L2">Par2</numbered-paragraph>
<numbered-paragraph style:name="L1">Par3</numbered-paragraph>

Correct?

So let's look what an ODF1.0/ODF1.1 reader would make of this:

<numbered-paragraph list-id="1" style:name="L1">Par1</numbered-paragraph>
<numbered-paragraph list-id="1" style:name="L2">Par2</numbered-paragraph>
<numbered-paragraph list-id="1" style:name="L1">Par3</numbered-paragraph>

interpreted by a ODF1.0/ODF1.1 conforming reader would result in
1. Par1
A. Par2 (wrong number, right style)
2. Par3 (wrong number, right style)

and 

<numbered-paragraph style:name="L1">Par1</numbered-paragraph>
<numbered-paragraph style:name="L1" list-style-override="L2">Par2</numbered-paragraph>
<numbered-paragraph style:name="L1">Par3</numbered-paragraph>

interpreted by a ODF1.0/ODF1.1 conforming reader would result in
1. Par1
2. Par2 (wrong stlye, right number)
3. Par3

So in on case the style is preserved and in the second case the actual number is preserved. Its just my feeling that the number is more important than the style.

I'm actually suprised that most of the TC members consider the style as more important than the actual number. I was not aware of this. I honestly always assumed that we had an understanding that the number is more important.

So under the assumption that
a) we make the list-id optional and 
b) we really consider backward compatibility with the style to be more important than backward compatibility with the right number
I consider to give up my objections. Althought I still think that the right number should be more important than the right style ;-)

~Florian

>>> Oliver-Rainer Wittmann - Software Engineer - Sun Microsystems <Oliver-Rainer.Wittmann@Sun.COM> 03/09/07 1:55 PM >>>
Florian Reuter wrote:
> Here is my proposal for the list-override enhancement.
> 
> It should also cover all use cases us the "list-id" proposal, so (in my opinion :-)) there is no need for a list-id. In fact I discourage the use of "list-ids" because:
> * No backward compatibility: A reader which does not understand the text:list-style-override will still be able to correctly number all paragraphs. The only difference would be a different formatting. For example instead of 1. ii. 3. an old reader would generate 1. 2. 3..
> * Using a list-id approach would cause a serious backward compatibility break, sin The association of list styles with a “counter domain” is usually sufficient. By using a style-override the style of the number formatting can be changed. Splitting this relationship between a list style and a “counter domain” would cause unneeded redundancy, since in the “normal” case the list-id and the style-name had to be emitted.
> * By introducing a list-id the transformation between text:lists and text:numbered-paragraphs can be quite complicated. This would be a burden for the implementation.
> 
> I also included some "normative examples" for ODF1.1 lists to avoild from misunderstandings of how lists currently work.
> 
> ~Florian

Hi,

as I told you already on Wednesday, I didn't share your concerns about 
the existing proposal - see 
http://lists.oasis-open.org/archives/office/200702/msg00172.html - in 
your serious given form.

I also suggested to be more complete and precise in your proposal. 
Namely state, what exactly is the impact for your proposed 
text:list-style-override on the <text:list> element and on the 
<text:numbered-paragraph> element.
I also think, that introducing the "override" attribute at element 
<text:list-item> provides much more flexibility, than to introduce it at 
element <text:list>.

Additionally, I suggested to include the clarifications (2), (4), (5) 
and (6), which are given in the existing proposal - see 
http://lists.oasis-open.org/archives/office/200702/msg00172.html - in a 
similar form.

The existing proposal also includes the clarification about the counter 
domain for numbered paragraphs - see (1) in 
http://lists.oasis-open.org/archives/office/200702/msg00172.html -, 
which is currently not clear in the ODF 1.1 specification. See 
http://www.oasis-open.org/archives/office/200702/msg00046.html and the 
following thread, which discussed this unclear point in the ODF 1.1 
specification. The proposed new attribute text:list-id has the elegance 
to clarify this in ODF 1.2.
Thus, I suggested, you should include a similar clarification about this 
in your proposal, too.


I'm a little bit disappointed that you haven't got considered any of my 
suggestions, yet.
I think your proposal in its current state isn't complete enough in 
order to compare it with the existing one. Thus right now, I don't think 
your proposal is a true alternative to the existing one.

Under the assumption, that you consider my suggestions, I want to ask 
you, if you can again consider the existing proposal and think about it. 
Please check, if you can support it. It's a compromise, that is 
supported by most of the TC members.


Regards, Oliver.