Re: Why "Platform Core" and "HTML5" are in the same spec

On Fri, 21 Nov 2008, Nikunj Mehta wrote:
> > 
> > My impression is that Nikunj volunteered to edit the API definitions, 
> > not the Web SQL language.
> 
> I was signing up to edit the SQL language, the offline cache, which in 
> my opinion is spec bloat (unnecessary sync algorithm and format) that 
> misses a key primitive, and server-sent events, which also creates spec 
> bloat (by defining a specific and limited format for events). All three 
> pieces, especially the latter two, should be on the table.

I would be happy for you to edit a WebSQL language specification. My 
apologies for not realising that you were including this in your proposal. 
Please go ahead and create such a specification. If you need access to 
dev.w3.org, I believe Mike Smith can hook you up.


Regarding the offline cache and server-sent events features, they are IMHO 
too tightly integrated with the language to really deserve their own 
separate specification. The offline cache feature is really just a 
processing model for the <html manifest=""> attribute, and the server sent 
event feature is just the processing model for the <eventsource> element. 
Both are therefore part of HTML5 proper, and I don't think we'd want to 
split these out. (One could maybe split out the server sent events format 
and API, actually, but I don't see much point. They couldn't advance to 
REC before the main spec, so it wouldn't really help anything, and I can't 
really see anyone wanting to implement it outside of HTML.)

Note in particular that being editor does not mean you get to specify what 
you want. Contrary to what several people have suggested, very little of 
HTML5 reflects my own personal desires. For example, the offline cache 
feature, server-sent events, WebSocket, <video>, etc, have all been 
dramatically changed from what I originally proposed, in many cases to the 
point where none of my own personal preferences remain. (I believe people 
think that I'm getting my own way because whenever I reject some feedback, 
I do so by taking the devil's advocate position, even if that position 
isn't really my own personal preference, so all they see is me arguing 
against them.)

For example, the original server-sent events feature, the way I designed 
it, was a much more generic mechanism. It was scaled back based on 
technical feedback (that I disagreed with, but my position was not 
technically sound compared to the arguments put forward). As editor, you 
have to put other people's technical feedback on an equal footing to your 
own, and must be able to make objective decisions between them.

What support is there from implementors for the changes you propose to the 
offline cache and server sent events features?

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

Received on Saturday, 22 November 2008 00:06:10 UTC