Re: [SVGMobile12] Comment: Please give specific error-handling behaviour

"David Woolley" <david@djwhome.demon.co.uk> wrote in message 
news:200505180601.j4I61X804307@djwhome.demon.co.uk...
>
>> Ugh, user-specific error handling behaviour is the path to a tag soup
>> world like we have for HTML.
>
> Effectively, fully specified error behaviour ceases to be error behaviour
> and becomes part of the normal behaviour as far as people trying to
> get the most out of the language (or for text media, trying to defeat
> spam filters) are concerned.

I think it becomes normal behaviour for anyone, an error that produces 
consistent identical rendering in all user agents is indistinguishable to 
all concerned from any other behaviour, and as mentioned by Boris, can place 
unrealistic implementation burdens on implementors - being forced to detect 
integer overflow in an unecessary burden, when the range of an integer can 
be constrained.

There is no reason to define all error conditions in user level 
specifications, I can't see any other specification that does it - for 
example CSS 2.1 doesn't constrain the range of integer, and I doubt very 
much any user agent currently supports correct z-index behaviour of z-index 
specified between differences in numbers that differ by 1 in 2^120, neither 
should we really expect to, it simply bloats the specification without 
concrete gain.

> The only specified error behaviours that won't have this effect tend to
> be unacceptable to browser writers because they shame the content
> authors

And penalise the user uneccessarily.

Jim.

Received on Wednesday, 18 May 2005 06:59:55 UTC