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

[Oliver-Rainer Wittmann - Software engineer - Microsystems Inc<Oliver-Rainer.Wittmann@Sun.COM>]

Hi,

here comes the conversion algorithm.

First converting a list defined using <text:numbered-paragraph> into a 
list defined using <text:list>:

Example:
We've got the following list styles in our text document
- L1, L2 and L3

We formed the following list:
<text:numbered-paragraph text:level="1" text:list-id="MyList1" 
text:style-name="L1">
	<text:p text:style-name="P1">Main chapter</text:p>
</text:numbered-paragraph>
<text:numbered-paragraph text:level="2" text:list-id="MyList1" 
text:style-name="L2">
         <text:p text:style-name="P2">Foo</text:p>
</text:numbered-paragraph>
<text:numbered-paragraph text:level="2" text:list-id="MyList1" 
text:style-name="L2">
         <text:p text:style-name="P2">Bar</text:p>
</text:numbered-paragraph>
<text:numbered-paragraph text:level="2" text:list-id="MyList1" 
text:style-name="L3">
         <text:p text:style-name="P2">annex</text:p>
</text:numbered-paragraph>
<text:numbered-paragraph text:level="2" text:list-id="MyList1" 
text:style-name="L3">
         <text:p text:style-name="P2">Another annex</text:p>
</text:numbered-paragraph>

Because all <text:numbered-paragraph> items have the same value for 
text:list-id, they all belong to the same list.

Algorithm:
(1) First, you have to choose a certain list style for the top 
<text:list> element. The best would be to choose the list style, which 
is used most in the list defined via <text:numbered-paragraph>.
In the given example this is list style "L1"

(2) Build up <text:list-item> hierachy corresponding to the text:level 
attribute of the <text:numbered-paragraph> items. For each 
<text:numbered-paragraph> item, whose list style differs from the one 
choosen in step (1), include a text:style-override attribute, whose 
value has to be set to list style, which is used by the 
<text:numbered-pararaph> item.

Applying these steps to the above given example will result in the 
following list defined using <text:list>:

<text:list text:style-name="L1">
   <text:list-item>
	<text:p text:style-name="P1">Main chapter</text:p>
   </text:list-item>
   <text:list>
     <text:list-item>
         <text:p text:style-name="P2">Foo</text:p>
     </text:list-item>
     <text:list-item>
         <text:p text:style-name="P2">Bar</text:p>
     </text:list-item>
     <text:list-item text:style-override="L2">
         <text:p text:style-name="P2">annex</text:p>
     </text:list-item>
     <text:list-item text:style-override="L3">
         <text:p text:style-name="P2">Another annex</text:p>
     </text:list-item>
   </text:list>
</text:list>


Now the other way of the conversion:
Taking the above list defined by using <text:list>.
(1) First, you have to choose an list-id for the list defined by using 
<text:list>. I would suggest to choose the name of list style of the 
top <text-list> followed by an underscore followed by a unique number.
For the given example I would choose "L1_87232343" as the list-id

(2) Transform each <text:list-item> item into a 
<text:numbered-paragraph> item with the following constraints for its 
attributes:
- value of text:style-name is the value of the text:style-override 
attribute, if it exists. Otherwise it's the name of the list-style, 
which is applied to this <text:list-item> item.
- value of text:list-id is the chosen list-id from step (1)

Applying these steps to the above given example will result in the 
following list defined using <text:numbered-paragraph>:
<text:numbered-paragraph text:level="1" text:list-id="L1_87232343" 
text:style-name="L1">
	<text:p text:style-name="P1">Main chapter</text:p>
</text:numbered-paragraph>
<text:numbered-paragraph text:level="2" text:list-id="L1_87232343" 
text:style-name="L2">
         <text:p text:style-name="P2">Foo</text:p>
</text:numbered-paragraph>
<text:numbered-paragraph text:level="2" text:list-id="L1_87232343" 
text:style-name="L2">
         <text:p text:style-name="P2">Bar</text:p>
</text:numbered-paragraph>
<text:numbered-paragraph text:level="2" text:list-id="L1_87232343" 
text:style-name="L3">
         <text:p text:style-name="P2">annex</text:p>
</text:numbered-paragraph>
<text:numbered-paragraph text:level="2" text:list-id="L1_87232343" 
text:style-name="L3">
         <text:p text:style-name="P2">Another annex</text:p>
</text:numbered-paragraph>


Regards, Oliver.


Florian Reuter schrieb:
> Hi Oliver,
> 
> your proposal/suggestion/comment seems to violate the statement explicitly made in the ODF spec:
> "A list in <text:list> representation could be converted into a list in <text:numbered-paragraph> representation and
> vice versa."
> 
> I therfore disagree with your proposal/suggestion/comment.
> 
> ~Florian
> 
> 
>>>> Oliver-Rainer Wittmann - Software Engineer - Sun Microsystems <Oliver-Rainer.Wittmann@Sun.COM> 01/09/07 12:14 PM >>>
> David Faure wrote:
>> On Monday 11 December 2006 17:22, Michael Brauer - Sun Germany - ham02 - Hamburg wrote:
>>> Hi,
>>>
>>> The original ODF list concept is that we have <text:list> elements in the content, and a 
>>> list style. Each list has its own numbering. That is, if I have two <text:list> elements 
>>> they both start with 1. 
>> This seems sensible. Note that it contradicts Oliver's interpretation ("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.")
>>
>>> When using <text:list> elements the list styles only specifies the  
>>> layout of the list. The lists themselves are identified by <text:list> elements, but not 
>>> by the style name. 
>>> The only problem we have is that we cannot change the style within a  
>>> list. The style:overwrite attribute that has been proposed would solve this.
>> ... so you would agree with the style:overwrite attribute being a text:list-item attribute rather than text:list
> attribute, right?
>> Otherwise you need two <text:list>s to change the style and this breaks the structure.
>> E.g. my 1/2/3/A/B example should be modelled as only one text:list, not two.
>> Another example would be differently-numbered subsections, like:
>> 1.
>>   1.1.
>>   1.2.
>> 2.
>>   2.A)
>>   2.B)
>> In this example it really is just one list, otherwise the numbering will end up being wrong.
>> This example also shows why we (koffice) dislike the "define 10 levels in one list style".
>> Not all items in this list use the same style. style-override at the item level solves this indeed,
>> although it requires that one style is the "normal" one and the other one is the "override"
>> With text:numbered-paragraph and one level per list style, the above would simply 6 numbered
>> paragraphs pointing to whichever style they want to use.
>>
>> ====
>>
>> To be fair, if OOo is supposed to be implement text:list correctly, then I'm still very confused
>> by the semantics. For instance if I increase the level (e.g. using the tab key) I get two paragraphs
>> with the same number, see attached document. For 4 paragraphs of depths of 1/1/2/2, I would 
>> expect to get A/B/a/b or A/B/1/2 or A/B/<nothing>/<nothing> or A/B/A/B if the same style is 
>> defined for level 2, but certainly not A/B/B/C !?
>>
>> Why does ODF define 10 levels in one style, while even the OOo GUI doesn't do that? "Numbering 4"
>> is "A) B) C) D)", but doesn't say anything about what happens at level 2 and below.
>> Not that I would want it to :) But then I would assume that A) B) is used at level 2 as well, unless
>> I select another numbering style for those paragraphs.
>>
>>> We later introduced <text:numbered-paragraph> elements. The problem we face here is that 
>>> we don't have a <text:list> element, and therefore no way to decide to which list (or 
>>> numbering) they belong. One option would be to define that they all belong to the same 
>>> numbering, and to add an attribute that starts the numbering at a certain paragraph. Other 
>>> options are to re-use the list-style, or to use list-ids. I don't have a clear opinion 
>>> what is the best solution, but I think it is a different issue than the one for the list 
>>> elements, and we should discuss it independent of it.
>> I agree, that the issue is different and should be solved independently. 
>>
>> "All belong to the same numbering" is actually more or less what we do in koffice.
>> For any numbered paragraph, we go up and find other numbered paragraphs of the same level
>> to find out what this paragraph's number is; and any upper-level (or outline) paragraph stops
>> the search.
>>
>> However the notion of a list-id for a bit more structure sounds nice, and from what I recall,
>> MSWord lists also have the equivalent of a list-id, this is where this idea mostly comes from.
>>
> 
> Hi all,
> 
> ok, my interpretation of the list style is some how OpenOffice.org driven.
> I agree that each <text:list> element starts a new list. OpenOffice.org 
> uses attribute text:continue-numbering="true" to achieve the 
> interpretation, that all lists using the same list style belong together.
> 
> I propose that all <text:numbered-paragraph> by default should belong to 
> *one* list. Thus, by <text:numbered-paragraph> a list with different 
> list styles is possible.
> 
> I propose to introduce the attribute text:style-override for the 
> <text:list-item> element to support different list styles inside a list 
> defined by the <text:list> element.
> 
> I propose to introduce the attribute text:list-id for the 
> <text:numbered-paragraph> element to define a group of 
> <text:numbered-paragraph> element to belong to an own list.
> 
> Regards, Oliver.
> 
> 
>