W3C home > Mailing lists > Public > www-tag@w3.org > August 2007

RE: definition of forward compatible/backward compatible still an open problem [XMLVersioning-41 ISSUE-41]

From: Marc de Graauw <marc@marcdegraauw.com>
Date: Wed, 29 Aug 2007 17:49:29 +0200
To: "'David Orchard'" <dorchard@bea.com>, "'Dan Connolly'" <connolly@w3.org>, "'www-tag'" <www-tag@w3.org>
Message-ID: <1778C6C3018544358D371EB3D831990E@Marc>

David Orchard:

| Right.  Where do we go from here?  You're not comfortable with the
| current definition for reasons you've stated.  It also sounds like we
| don't have a counter-proposal, and further generating a 
| counter proposal
| could involve a huge amount of time in the way of research.

I had been working on a blog item with a different approach some months ago;
this discussion stimulated me to finish it. Maybe it contains some useful or
novel ideas.

I've made a formalization of compatibility based on processor behaviour when
consuming a text instead of the information set conveyed by the text.
Basically my approach is to define two languages L1 and L2 as semantically
backward compatible iff L2 processors behave the same as L1 processors for
the defined set of texts of L1. For texts in the accept set, but not the
defined set of L1, L2 processors may add new behaviour. I'm not much of a
fan of behaviourism in psychology or philosophy, but here it seems quite
natural since often language specs make demands about the behaviour of
conforming processors.

The full formalization is available at:
http://www.marcdegraauw.com/2007/08/29/axioms-of-versioning/

(Printer friendly:
http://www.marcdegraauw.com/files/axiomsofversioning.html)

| Can we weaken the definition and effectively say it's language
| dependent?  
| 
| "I1 is compatible with I2 in a language specific manner such 
| that is not
| generalizable."

I don't know if this is a good idea - often language specs do not contain
formalisms to establish compatibility, so wouldn't this leave compatibility
undefined?

Cheers,

Marc de Graauw

www.marcdegraauw.com
 
| Cheers,
| Dave
| > -----Original Message-----
| > From: www-tag-request@w3.org [mailto:www-tag-request@w3.org] 
| > On Behalf Of Dan Connolly
| > Sent: Friday, August 24, 2007 2:58 PM
| > To: www-tag
| > Subject: definition of forward compatible/backward compatible 
| > still an open problem [XMLVersioning-41 ISSUE-41]
| > 
| > 
| > The July 4 draft has a revised definition of information 
| > compatibility in response to my [21Aug] comments on the 
| > defintion of forward/backward compatibility of languages:
| > 
| > [[
| > * Let I1 be the information conveyed by Text T per language L1. 
| > 
| >   * Let I2 be the information conveyed by Text T per language L2. 
| > 
| >   * I1 is compatible with I2 if all of the information in 
| I1 does not
| >     replace or contradict any information in I2.
| > ]]
| >  -- http://www.w3.org/2001/tag/doc/versioning
| > http://www.w3.org/2001/tag/doc/versioning-20070704.html
| > 
| > But it still defines compatibility of languages in terms of 
| > compatibility of information in a way that doesn't appeal to me.
| > 
| > That definition of compatibility of information reminds me of 
| > the conventional definition of consistency:
| > 
| > [[
| > A theory is said to be satisfiable if it has a model. A 
| > theory is consistent if its closure (under the usual rules of 
| > inference) does not contain a contradiction. One way of 
| > stating the completeness theorem is the following: A theory 
| > is satisfiable if and only if it is consistent.
| > ]]
| >  -- http://en.wikipedia.org/wiki/Model_theory
| > 
| > ... but there a theory is a text. I'm not familiar with any 
| > definition of consistency/compatibility of stuff that the 
| > text refers to, i.e.
| > "the information conveyed by a text."
| > 
| > I was part way working thru mappings back in September:
| > 
| > Re: Re-expressing our formalisation of Language 
| > http://lists.w3.org/Archives/Public/www-tag/2006Sep/0040.html
| > 
| > 
| > But I didn't really see how to finish it. And neither did 
| > Henry nor Pat Hayes.
| > 
| > This still feels like an open research problem, to me.
| > 
| > 
| > p.s. Hi tracker. This is progress on, if not completion of, ACTION-4
| > http://www.w3.org/2001/tag/group/track/actions/4
| > 
| > [21Aug] http://lists.w3.org/Archives/Public/www-tag/2006Aug/0084
| > 
| > 
| > --
| > Dan Connolly, W3C http://www.w3.org/People/Connolly/
| > 
| > 
| > 
| > 
| 
| 
| 
Received on Wednesday, 29 August 2007 15:49:49 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:47 GMT