Re: Percent encoding

On Wed, Feb 24, 2010 at 9:19 PM, Jack Jansen <Jack.Jansen@cwi.nl> wrote:
> Does anyone know what the canonical reference is for percent-decoding? Because we seem to assume it can be done late, but I just typed <http://www.youtube.com/watch%3fv=FwJduPtCvSM%26feature=rec-r2-2r-1-HM> into my browser URL field, and the ? and & were decoded...
>
> This means that our whole discussion about getting rid of he quotes and using only percent escapes may be moot.....


Ultimately, I think a server can decide what URLs it wants to accept
and which not.

http://tools.ietf.org/html/rfc3986#section-2.4 only states "under
normal circumstances" only the name-value pairs are encoded, not the
rest. But if the server can interpret the URL either way, there's
nothing telling him not to. Servers of companies are usually much more
lenient towards "mistyped"/misconstructed URLs than standards.

I don't think that should influence our specification though. We can
go with the standard and propose the name-value pairs are percent
encoded. That in turn doesn't force us to remove the quotes either,
but it might make it possible. We should put some examples together to
give us a better impression what problems we may be introducing.

Cheers,
Silvia.

Received on Wednesday, 24 February 2010 12:36:13 UTC