[whatwg] We should not throw DOM Consistency and Infoset compatibility under the bus

Wed Mar 6 09:55:27 PST 2013

On Wed, 6 Mar 2013, Simon Pieters wrote:
> On Mon, 14 Jan 2013 09:40:56 +0100, Henri Sivonen <hsivonen at iki.fi> wrote:
> 
> > Would it be terrible to make attempts to mutate the 'is' attribute 
> > throw thereby teaching authors who actually try to mutate it that it's 
> > not mutable?
> 
> We already have several attributes that are immutable but don't throw or 
> anything when they're changed.
> 
> e.g. <script src>, <html manifest>.
> 
> In the case of manifest, we omitted the IDL attribute to signal that it 
> shouldn't be changed.
> 
> As another example, <input type> was immutable in old IE and that made 
> people think of it as immutable.

I think the content attributes should always be mutable, for consistency 
and to avoid unexpected crashes in code that just tries to set every 
attribute indiscriminately. (Much like how the DOM tree needs to be a real 
tree, not a graph, so that code can walk it without fear of loops.)

It's ok to have attributes that set some other variable that is itself 
immutable, though. For example, <html manifest> sets the page's manifest 
URL, but you can't change the manifest URL, even though you _can_ change 
the element's attribute after the fact.

In a sort of similar way, <script src> is mutable, it's just that it's 
only read at one point in the script processing model and it's the value 
at that point that matters.

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