Re: HTML plan

> On Jan 20, 2016, at 08:04, Tab Atkins Jr. <jackalmage@gmail.com> wrote:
> 
> On Tue, Jan 19, 2016 at 11:20 AM, Léonie Watson <tink@tink.uk> wrote:
>> One approach to test modularisation is to encourage people working on a
>> specific section to split it out from the "main" HTML specification, move it
>> independently to Recommendation, so that it can be referenced normatively
>> from the base specification. This way we can get some experience of the
>> process without undertaking a massive project before we really know the
>> costs and benefits.
> 
> This is "the CSS process", and it's worked well for the past decade
> (with CSS 2.1 serving as the big monolithic base, and modules
> gradually carving chunks of it out and levelling them independently).

Well, sorta.

Assuming "the base specification" in "so that it can be referenced normatively from the base specification" means the monolithic HTML5.x spec, there's a difference between "the CSS process" and what I'm reading in Léonie's mail.

CSS 3+ Modules reference (and often override) CSS2.1 and move along independently on the REC track, but they do not get referred to normatively FROM CSS2.x. It's the other way around.

Modulo errata, CSS2.1 is meant to stay as it is. Other modules get written on (mostly) self contained topics. Whether they merely complete the CSS2.1 base specification, or whether they replace some sections of it is made explicit somewhere near the beginning of each document.

There is no particular need to refer to these new independent modules FROM the base spec. Eventually, the CSSWG will produce a CSS2.x spec with all the bits that have been overridden by newer modules removed, and it will probably include pointers to where these parts have moved to, but won't be meant as the base spec from which you can find everything.

 - Florian

Received on Wednesday, 20 January 2016 04:37:39 UTC