Re: Cleaning House

On Wed, 2 May 2007, Maciej Stachowiak wrote:

>> I agree with documenting deprecated tags for future UAs but don't 
>> see that that necessarily means this has to be done in the same 
>> specification. It would be a lot cleaner to have one 
>> specification for just the current tags and separate 
>> specifications for deprecated/obsolete tags.
> It sounds to me like this would be more work for the editors, with 
> little practical benefit. I'd rather not ask them to do the extra 
> work of writing multiple specs, where massive cross-references 
> will be needed.

>> It would be a lot more manageable if we were to proceed with 
>> several smaller specs. This would be easier on people from 
>> outside of the HTML WG, and easier when it comes to moving 
>> through the various stages of the W3C Process.
> Having multiple smaller specs is also extra work for the editors, 
> since many cross-references will likely be needed. If we can find 
> self-contained sections where there would be a real benefit to 
> breaking it out (for example, it's intended to be used with other 
> languages besides HTML) and we can find the manpower to move that 
> section forward, it might make sense. But breaking things up 
> arbitrarily likely won't be helpful.

If the spec needs to be so tightly intercoupled that modularity 
isn't feasible then we are in deep trouble as this is likely to make 
it hard to spot detailed problems and will also take much longer to 
reach Recommendation status. Such tight intercoupling will make it 
nearly impossible to provide an complete set of test assertions and 
an associated test suite, would seem like poor practice.

I am however convinced that HTML can be specified in a more modular 
fashion, and that this would allow the hTML WG to prioritise and 
progress modules on a much quicker basis than one giant spec. I am 
sure that many people would like to see a sequence of modular W3C 
Recommendations coming out a regular intervals rather than waiting 
many years for the big bang. The work of the editors will be reduced 
by this approach despite the initial investment in splitting the 
spec into modules, however, I bet that in the process, they and we 
will discover things along the way that will make the combined specs 
much stronger as a result.


  Dave Raggett <>

Received on Thursday, 3 May 2007 08:10:27 UTC