[whatwg] Re: Performance of rich internet applications

[Malcolm Rowe]

Hi Brad, 

> I think something we should be careful on is the performance of whatever 
> emulation layer we find is necessary in IE 6 to support these specs.
> I've worked with other cross-browser emulation layers, some written in 
> JavaScript, and while they make life easier they can be really slow.

Note that for Web Forms 2 particularly, the simpler parts are intended to be 
allowed to degrade 'automatically' to less-rich equivalents. 

For example, on IE6, an '<input type="email">' will be treated as '<input>', 
i.e., an untyped text field. So while a Web Forms 2-compliant UA might 
provide a special control for this type of input (possibly allowing the user 
to select contacts from their address list), a non-compliant UA will not 
require any special client-side handling. 

This is the same philosophy espoused by CSS, whose properties are designed 
to be both 'hints' (hence not critical) and backward-compatible (in the 
sense that the default behaviour of the property is usually the one carried 
out by a UA that does not implement the property). But I digress. 

One part of Web Forms 2 that will require client-side assistance is the 
replication model (there may be other parts; I've not checked a recent 
copy). I note that the spec describes creation of a client-side library as 
'an exercise for the reader'. 

I wonder if, instead, development of a BSD-licensed IE6-compatible 
javascript emulation layer for those parts that require it (such as the 
replication model) should be included in the delivery of Web Forms 2? 

Regards,
Malcolm