Robert Greig wrote:
On 26/07/07, Jonathan Robie <[EMAIL PROTECTED]> wrote:
I agree with Rajith and Gordon here - the current URLs do not seem to
follow common conventions, and are hard to remember because of that.

I think it is worth pointing out that with 0-8 the set of hosts to use
for failover is configured by the client. Hence the brokerlist in the
URL.

I don't think that Qpid has the right to register an amqp name scheme
for URLs, we  have the right to use a qpid name scheme. I'm not sure
which group should create the URL format. (OK, I'm a standards geek,
it's a curse I live with ;->)

Like many things we do, we intended to get it included in the standard.

This is a bit of a problem now as the defined "amqp:" URL scheme for 0-10 is not compatible with our use of "amqp:".

I suggest we use the conventions discussed in
http://tools.ietf.org/html/rfc3986. As I interpret it, AMQP can use any
of the following forms:

1. amqp://[EMAIL PROTECTED]/virtualhost?param=value,param=value
2. amqp://localhost/virtualhost?user=jrobie,param=value
3. amqp:localhost?virtualhost=hostname,user=jrobie

Qpid would need to use a qpid name scheme:

Why? http is a protocol and we have http://foo.com/ not httpd://foo.com

1. qpid://[EMAIL PROTECTED]/virtualhost?param=value,param=value
2. qpid://localhost/virtualhost?user=jrobie,param=value
3. qpid:localhost?virtualhost=hostname,user=jrobie

Can you provide an example that includes the brokerlist for failover
and the client id?

I agree that our "URL" is not entirely like other URLs but I am not
convinced that any alternative that actually includes all the things
that are required is any clearer.

I think the goal is not necessarily to make the complex cases look simple, but rather to make the simple cases look normal, i.e. there is no reason we can't choose an URL syntax that looks normal when you're only specifying a single broker rather than a list of brokers.

Maybe we should create something more like a corbaloc address?

I believe the 0-10 URL scheme is actually modeled on corbaloc.

--Rafael

Reply via email to