Re: References

On 24/12/2011 22:27, Anne van Kesteren wrote:
> On Sat, 24 Dec 2011 18:15:26 +0100, David Carlisle <davidc@nag.co.uk>
>  wrote:
>> On 24/12/2011 10:37, Marcos Caceres wrote:
>>> But for cases where it does not break backwards compatibility,
>>> then I should be able to have it.
>>
>> There are many other reasons than backward compatibility for
>> wanting to refer to a versioned spec. Xpath2 (and soon 3) are to a
>> large degree compatible with 1, but they are also considerably
>> larger languages and many developers (notably browser
>> implementations) have chosen to stay at version 1. It would be
>> utterly confusing if references to xpath in DOM API silently
>> updated to refer to xpath3, if implementations are all at 1. This
>> is the _normal_ case.
>
> I do not think so. For the documents listed on
> http://platform.html5.org/ I think browsers would want to implement
> the latest of them for pretty much all of them, with the exception of
> those that are not fully backwards compatible or have not as much
> interest from the developer community.



Not sure how to read your comment, taking the specific example of xpath.

You could be saying that if xpath had been a "living standard" with the
xpath 2 (then 3) just replacing the xpath 1 spec in place, then
developers would have just all updated. But I don't believe that's true
(or that's what you intended to say).

Or you could be saying that Xpath 2 isn't backward compatible with 1.
There are of course some documented incompatibilities but in fact these
are typically edge cases not affecting real expressions and anyway the
differences are no greater than already allowed differences between
implementations. So a judgement call as always, but having separate
xpath 1 and xpath 2 specs makes it easier to state what judgement has
been made (whether to stay with xpath 1 or move on).

Or you could be saying that the developer community has lost interest.
Well, it may well be that the rather small community of developers that
can affect browser implementations has lost interest in xpath (and XML
generally) and again for them, staying at xpath 1 may make sense.
However that wouldn't have been a reason not to develop xpath further.
Judging by discussions in xslt fora, and development of major xslt
collections such as docbook or TEI, the xpath developer community has
more or less totally moved on to xpath2, the only reason for still
developing using xslt1/xpath1 now is for client side use in a browser.
Again if the browser developers wish to work to a frozen spec,
versioning appears to be a more useful model than "living standard".


David





________________________________________________________________________
The Numerical Algorithms Group Ltd is a company registered in England
and Wales with company number 1249803. The registered office is:
Wilkinson House, Jordan Hill Road, Oxford OX2 8DR, United Kingdom.

This e-mail has been scanned for all viruses by Star. The service is
powered by MessageLabs. 
________________________________________________________________________

Received on Tuesday, 3 January 2012 10:08:38 UTC