Re: Report on testing of the link relations registry

On Sep 6, 2010, at 1:45 AM, Henri Sivonen wrote:

> David Singer wrote:
>> I think that formal registries and Wikis are both useful.
>> 
>> Formal registries can have well-defined entry criteria, expert review,
>> stability, references/owners/specifications, and so on (if they wish).
>> They can be trusted, stable.
> 
> Trusted by whom and for what purpose? Not by browser implementors or Web authors.
> 
> Who benefits from trusting that image/svg+xml doesn't exist (and presumably "cannot" be used) or from trusting that ISO-8859-1 isn't an alias for Windows-1252 when decoding?

Any media type can be used, whether it is registered or not.  But there
is no point in registering a type that has no corresponding specification
of how it is to be processed.

ISO-8859-1 is not an alias for windows-1252, anywhere.  Please do not
confuse errors in labeling due to one browser's rendering choices
with defining a new meaning for the label.  Working around content errors
does not imply they are not errors.

....Roy

Received on Tuesday, 7 September 2010 19:53:45 UTC