XML Memory breakout, TPAC 2012 -- 31 Oct 2012

<scribe> scribe: liam

<scribe> scribenick: liam

introductions

Jim Fuller: Senior engineer at MarkLogic, interested in XML!

Jirka Kosek: also interested in XML!

David Lewis: co-chair of multilingual web LT WG, at Trinity College Dublin, interested in provenance too.

Liam Quin: XML Activity Lead at W3C

Robin LaFontaine: from DeltaXML, it's our business focus

Steven Pembertom: chair of XForms, prev. chair of XHTML, on ODF committee too

scribe: interested to know if there's a standard here

Background

<Robin_La_F> http://www.w3.org/wiki/User_talk:Rlafonta

<Steven> http://www.w3.org/wiki/TPAC2012/SessionIdeas#XML_Memory_and_Change_Tracking

Robin: this work originates from ODF where they want improved change tracking...

we came up with something more generic than just for ODF

principle here is, could XML benefit from having the ability to record previous versions or variants, track changes

What other useful areas might there be that this could apply to?

In the document area it's fairly obvious

e.g. if it's a technical manual I might be primarily interested in what's new

If you look at data, e.g. configuration files, would be nice to roll back to a previous version, or see changes

Audit trails, who changed what, when

So, is it worth having a standard?

ODF, DITA, DocBook, StrategicML (project planning)

Standard might also be useful for processing change

and XML editors

all tend to have their own mechanisms for tracking change, but they don't want to do something different from DITA and DocBook

daveL: another use case, localization, e.g. to know what new stuff needs to be translated

Daniel: work for paypal, picked this as I know the least about it!

Alex Russell: Work at Google, Here to understand what you're doing

Norm Walsh: produces tools (MarkLogic) and would like to have fewer ways of tracking change

Robin_La_F: we're not starting with a blank sheet, we have a proposal...

I don't say it's complete, but it represents maybe a person-year

Robin_La_F: Principles

1. must be easy to get the latest version

[see http://www.w3.org/wiki/User_talk:Rlafonta ]

2. validation... if you have the latest document, and it's schema-valid, and undo the latest change, and that's schema-valid, it's OK

3. ability to group changes, e.g. changes from this version to that version

or if you delete a table row, an XML element, but deleting a column is harder, lots of changes

4. managing dependencies... e.g. 2 changes to a paragraph and then delete it, can't undo the changes unless you undo the paragraph first.

Liam: multiple dependency, e.g. I change a schema _and_ a document, tracking dependencies between documents

Robin_La_F: yes, I think that's OK

Jirka: not sure it's a real use case because e.g. docbook schema never changes

Jim: is this in scope, a higher-order binding, tracking between documents?

Robin_La_F: it's in scope in that the principle of validation would work, out of scope in that what validation might mean isn't part of this

Liam: different example, dependency between two chapters, always want the corresponding versions

Robin_La_F: our proposal is only about a single document; the relationship is part of the CMS

Daniel: syntactic version semantic validaton difference and only handle the syntax?

Robin_La_F: a content management system could handle both, e.g. using the same mechanism, but here we're looking at a single document.

Norm: experience suggests that doing this with a single document is hard enough, let's start with that!

Jim: reminded of Dana Florescu's Time Axis for XQuery

Liam: you could have a document that represented multi-document relationships and track changes in it.

Robin_La_F: proposal has two levels

1. sufficient to track any change but not always optimal

e.g. add and delete of text and attributes

2. add or remove elements without touching content; split and merge elements; move elements

Level 2 is a lot more complicated, but necessary in some situations

e.g. in level 2 you can split a paragraph, or merge two paragraphs

Jim: do the levels layer?

Robin_La_F: level 1 is a subset of level 2

and you can always convert a level 2 changelog and convert to level 1

Daniel: would it be fair to say that level 1 operates at the DOM level and level 2 operates at the content level
... example - ruby in Japanese.

Robin_La_F: there's a history & community in XML editor area of using processing instructions rather than namespaces, because it's ignored by schema validation

I think we can accommodate both of those if there's a defined [transformation between them].

Robin_La_F: the 3rd area that comes in is, do we want an external representation or only in the document?

e.g. so I can send you the diffs.

Next Steps

Steven: should we recommend this as an important area for W3C to adopt?

Jim: I think it's a perfect CG candidate

Steven: a CG is fast, can create one in 15 minutes, but it doesn't get you a Recommendation

Alex: this is an important area for HTML

Liam: a WG can have liaisons with other groups.

Jim: worried that a WG might get pulled apart, a fast CG might be able to produce a coherent design

Jirka: a CG can also be joined by people not W3C Members

Daniel: DAV has relevance here too

want to make sure we don't preclude DAV that content management systems use today.

CONSENSUS: start with a CG

Jirka: and a paper at XML Prague in February would be great!

CG: Change Management Markup Community Group

<Jirka> CFP for XML Prague is open till end of November http://www.xmlprague.cz/call-for-papers/

<Steven> http://www.w3.org/community/groups/proposed/#change

<Steven> FtF at XML Prague

[adjourned]

<Steven> s/groups.proposed.#//

<Steven> CG now created http://www.w3.org/community/change/

- DRAFT -

XML Memory breakout, TPAC 2012

31 Oct 2012

Attendees

Contents

introductions

Background

Next Steps

Summary of Action Items

Scribe.perl diagnostic output