Re: Why bound prefixes are an anti-pattern in language design

On Fri, 7 Aug 2009, Toby Inkster wrote:
> 
> It is true that there are some techniques (e.g. microformats) which are 
> simpler still, RDFa has other advantages (such as unambiguity and 
> consistency) which, in my mind, are worth sacrificing a little ease of 
> use for.

Microdata (in HTML5) has unambiguity and consistency without sacrificing 
the ease of use, at least not from the use of prefix binding mechanisms.


> On the other hand, the current HTML5 draft *does* define a number of
> features that *do* seem to require programmer-level knowledge to use:
> 
> 	- the <canvas> element
> 	- the <datagrid> element
> 	- the Window object
> 	- Drag and drop
> 
> Et cetera. If programmer-level complexity should exclude a solution from
> HTML5, then all these should be excluded ahead of RDFa.

These features are intended for Web application authors. If you have any 
suggestions for how to avoid the complexity they present while still 
making the features they expose possible, I would be more than happy to 
change the spec.


> However, <canvas> is an *optional* element. It is not *wrong* for people 
> to create web pages that do not use it. Thus the complexity is only 
> exposed to people who opt into using it.

No, <canvas>'s complexity unfortunately affects everyone. Tutorial 
authors, implementors, testers, people reading tutorials, book writers, 
people reading books, people maintaining pages written by others, etc -- 
every feature on the platform adds to the complexity of the platform. We 
can't just pretend that if people don't use the feature intentionally, 
they aren't exposed to it, and therefore complexity is ok.

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

Received on Friday, 7 August 2009 19:47:16 UTC