On Mon, 28 Feb 2011 18:23:51 -0700, Alex Rousskov wrote:
On 02/28/2011 04:44 PM, Amos Jeffries wrote:
In the final URL and protocol design we have AnyP::ProtocolType
which
enumerates the registered protocols accepted in URLs (lower case)
and
the request protocol field (upper case). Then URLScheme which
expands
that to allow other protocol names accepted in URLs.
Perhaps this is too abstract to allow a meaningful discussion right
now,
but why not simplify and just have one AnyP::Protocol class that
stores
the ID and the image of any protocol? Can that cover both URL scheme
(lowercase) and MIME message first-line (upper case) uses, among
other
things?
I'm build-testing a patch which includes the missing changes to
URLScheme which does the split nicely without much alteration to
existing code.
Will have that for you in a few hours.
For now there are places using local representations in strings etc of
the protocol, or using protocol_t where it should be URLScheme. Future
polish will clear those up correctly once this patch straightens out the
two. Then we can move on to supporting unknown URL protocols with
pass-thru.
Amos
