IRC log from today's meeting - 2012-12-19

[Svante Schubert <svante.schubert@gmail.com>]

Please find the IRC log of today's meeting below.
Our next meeting will be in four weeks on the 16th of January.


Attendees: John, Patrick, Thorsten, Oliver and Svante
Action item: Oliver volunteered to start with the definition of an ODF XML change for insertion and deletion of a paragraph.

[15:40] Svante Schubert: How can we improve our work in the SC?

[15:42] Svante Schubert: Thorsten: Shares the notion that we are loosing ourselves in too many details. We need something done, having some scratchy outline. We should start with some of the "easy" stuff, as we already have the design.

[15:44] Svante Schubert: Agreed. We might start with specifying the insertion of paragraphs and text.

[15:44] Svante Schubert: Patrick: What do we need to specify?

[15:45] Svante Schubert: How a change is represented in XML and second what is the change in the XML.

[15:46] Svante Schubert: Like "insertParagraph" operation with parameter

[15:46] Patrick: Insert an element, element is <p>

[15:46] Svante Schubert: exactly

[15:46] Patrick: name the insertion operation, the element to be inserted and the location of the insertion

[15:47] Patrick: <p-1-previous><p-new><p-2-previous>

[15:48] Svante Schubert: How the label/operation might be serialized in XML, I have suggest as a start idea in the two documents (presentation and basic scenario list) -->
https://www.oasis-open.org/committees/document.php?document_id=47221&wg_abbrev=office-collab
https://www.oasis-open.org/committees/document.php?document_id=45463&wg_abbrev=office-collab

[15:49] Svante Schubert: Talking about testing

[15:49] Svante Schubert: It would be nice, if we could compare two documents if they are the same

[15:49] Svante Schubert: Like ignoring automatic style names and span nesting for instance

[15:50] Svante Schubert: Patrick: The current specification does not need to specify ODF normalization

[15:50] Svante Schubert: It could, but it doesn't

[15:51] Svante Schubert: Lets say we start with the operation insert paragraph

[15:51] Svante Schubert: <add type=paragraph s="/4711"/>

[15:52] Svante Schubert: ^^ this might be a serialization of the insert operation

[15:53] Svante Schubert: Paragraph can not be directly be nested, but it might happen that w

[15:53] Svante Schubert: <add type=paragraph s="/4711/5"/>

[15:54] Svante Schubert: If we read a ODF document with SAX

[15:54] Svante Schubert: we might create component positions

[15:55] Svante Schubert: we might align the component with the root element of the component

[15:55] Svante Schubert: in our case text

[15:55] Svante Schubert: text: p

[15:57] Svante Schubert: How do we specify the change (the insertion of the text) in our specification?

[15:57] Oliver-Rainer Wittmann: i am dialing in

[15:57] Svante Schubert: most likely with written human words 

[15:58] Svante Schubert: Most likely we would define in a generic way how to find the position

[15:58] Svante Schubert: Next meeting on the 16th, no meetin on the 2nd

[16:13] Svante Schubert: Oliver: We need as well the inverse operation

[16:13] Svante Schubert: This would be delete..

[16:13] Svante Schubert: <del type=paragraph s="/2"/>

[16:13] Svante Schubert: ^^ This would be how the label would be serialized

[16:14] Svante Schubert: Counting starts after office:body/*

[16:14] Svante Schubert: Every <text> is incrementing the count

[16:14] Svante Schubert: counting starts only when a known component is being found

[16:16] Svante Schubert: When using SAX (see http://www.saxproject.org/apidoc/org/xml/sax/ContentHandler.html) whenever you pass

[16:16] Svante Schubert: startElement

[16:16] Svante Schubert: you count every known component. In the beginning you only count text

[16:18] Svante Schubert: Oliver states: we need to know about the other elements/components, otherwise we can not reinsert always correctly

[16:19] Svante Schubert: Make it abstract, imagine we have A and B and know only A

[16:19] Svante Schubert: ABABA

[16:19] Svante Schubert: deleting the middle A result into ABBA and we do not know how to insert it correctly back into it

[16:19] Svante Schubert: AABBA or ABBAA would be both wrong

[16:21] Svante Schubert: The most common solution is to count as well the unknown children.

[16:21] Svante Schubert: ^^children^^siblings

[16:21] Svante Schubert: using some kind of placeholder, that yet have no semantics

[16:22] Svante Schubert: How about when ODF 1.4 have new elements. Is there a way to be compatible?

[16:23] Svante Schubert: The schema can tell us all elements being sibling of a text, to solve the insertion problem.

[16:23] Svante Schubert: Siblings as the problem only occurs if you have some unknown on the same level

[16:25] Svante Schubert: Oliver: Do we need to know about all XML?

[16:25] Svante Schubert: No we do not need 'markers' of components (e.g. bookmarks) only components are of interest

[16:26] Oliver-Rainer Wittmann: we do not know every XML element, but the counting need to consider every XML element. Right?

[16:29] Svante Schubert: No 

[16:29] Svante Schubert: Only XML that are component roots

[16:29] Svante Schubert: e.g. table:table, text, etc.

[16:30] Svante Schubert: no text:bookmark, either

[16:30] Oliver-Rainer Wittmann: Then I have to admit I can not start with the specification text, because I do not know how to find the right "position" values

[16:31] Patrick: What about using byte offsets for locations of change operations? The change operations are still specified as XML content but the locations where the operations take place are specified by byte-offsets. Thinking we can define a normalization of the file for byte off-set purposes. To solve the problem of which components do we count when there is no normalized XML stream.

[16:32] Svante Schubert: <del type=paragraph s="/2"/>

[16:33] Svante Schubert: ^^ means you delete the second text we find after office:body/office:text

[16:33] Svante Schubert: add paragraaph

[16:34] Svante Schubert: <add type=paragraph s="/2"/> inserts a paragraph where currently the second paragraph is

[16:34] Thorsten Behrens: sorry guys, gotta run for another meeting in ~10 minutes

[16:34] Svante Schubert: insertion between two tables should not work at the first shoot, okay?

[16:35] Svante Schubert: If you want to be ablet to insert between tables you have to count table:table as well

[16:36] Svante Schubert: as a component

[16:36] Svante Schubert: <add type=paragraph s="/2"/> inserts a paragraph where currently the second component is

[16:36] Svante Schubert: ^^

[16:37] Svante Schubert: Like you have only one paragraph

[16:37] Svante Schubert: it is still <add type=paragraph s="/2"/>

[16:38] Svante Schubert: <add type=paragraph s="/2"/> inserts a paragraph where the second component would be. If there is already a component move it behind.

[16:40] Oliver-Rainer Wittmann: I wish you all a nice Xmas and happy new year

[16:40] Oliver-Rainer Wittmann: bye

[16:40] Svante Schubert: Same to you! bye