Re: Current state of DOM2 Traversel and Range

On Tue, 8 Sep 2009, Morten Barklund wrote:
> On Mon, 7 Sep 2009, Ian Hickson wrote:
> > On Sun, 6 Sep 2009, Morten Barklund wrote:
> >>
> >> What would be the purpose of a rewrite and what would be the work 
> >> involved?
> >
> > The purpose would be to make it so well-defined that implementations 
> > could be interoperable without reverse-engineering each other. The 
> > work involved would be writing a new spec from scratch, probably about 
> > 10-20 hours a week for four months, followed by maintenance and bug 
> > fixing, probably about 5 hours a week for six months, followed by 
> > waiting for someone else to write a test suite, helping them out where 
> > possible and fixing problems they find, probably a few years with 
> > minimal work on average but with the occasional 20-hour week when an 
> > implementor or tester finds a bunch of problems at once.
> 
> This actually does not sound as intimidating to me, as it might could 
> have.

Cool!


> What would be the expected coverage of the two new specifications? Would 
> they cover only what the single specification covers today (but with a 
> lot more detail of course) or should many new things be added to them? 

Traversal would be more or less what's in the traversal part of the spec 
now.

Range would likely quickly get new additions, but would start off the same 
as what the spec has now.

I would encourage you to just take one of these (probably Traversal) to 
start with, rather than worrying about doing both, if you are in fact 
interested in doing this.


> E.g. if some implementor extended his implementations with extra 
> features, could they be deemed interesting enough to creep into the 
> specification?

Depends on the extension, but yes, they could.


> I will try to do some reverse engineering (if I can't find it done by 
> someone else) on different implementors to pinpoint key differences for 
> starters. Are there any important implementors besides browsers, that I 
> should be aware of?

I believe several server-side implementations exist, but it may be that 
the spec needs to fork into "Web" versions of the spec and "Server" 
versions of the spec, if the implementations aren't compatible. (This is 
already happening to some extent with DOM Core.)

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Tuesday, 8 September 2009 11:46:48 UTC