RE: RFC 2397 vs relative references

No URI scheme definition BNF includes the definition of
the fragment identifier, because the fragment identifier
is relative to the content-type of any retrieved value,
not relative to the scheme itself.

Putting HTML in a 'data:' URI is a bit of a stretch,
because of the necessity of escaping%20all%20of%20the%20spaces,
but I suppose you could do it.

I suppose this should be clarified in the various documents
on URI scheme registration, because there have been some
attempts in the past to define scheme-specific fragment
identifiers due to this misunderstanding.

 
>   There is some constant confusion around relative references and
> references to fragments using the data: URL scheme. In particular,
> some people argue that references of the form data:...#fragment
> are not allowed (because the BNF in RFC 2397 does not mention this)
> and if the following fragments are part of the data part in e.g.
> data:text/html;charset=utf-8,...
> 
>   ... <a href='#fragment'> ...
> 
> or
> 
>   ... <a href='text/html;charset-utf-8,...'> ...
> 
> this is either not allowed either and/or processing of it is not
> defined. This confusion is caused in part by the assertion in RFC
> 2397 that 'The "data" URL scheme has no relative URL forms.' Could
> you clarify which of these forms of references are allowed and how
> to process them? There are some other issues in the RFC, ideally
> someone would make a second edition of it. What do you think?

If there are other issues, please specify. I think the
fragment identifier issue is more general than 'data:' though.

Received on Wednesday, 19 April 2006 18:10:51 UTC