On Wed, Feb 27, 2008 at 6:27 AM, Martin Ritchie <[EMAIL PROTECTED]> wrote:

> On 26/02/2008, Rajith Attapattu <[EMAIL PROTECTED]> wrote:
> > Hi All,
>
>  [...]

> That is at least on the M2.1 branch. I have only had a brief look over
> your new parser on trunk which is a way bigger change than the three
> lines I suggested to correctly parse wildcard topics in the the
> routingkey value if it is in the options.


Martin please feel free to review the code and suggest improvements.
I ensured that all tests in the DestinationURLTest pass and the other 3 test
failures.
The remaining test failures  were there before the  destination changes. (I
am working on getting them resolved as well)

Perhaps it doesn't require
> the / exists between queue and destination.
>

I wasn't sure about this so currently this is allowed. Fixing this is
trivial.


> >  We need to make this less ambiguous to the end user.
> >  I propose that we always interpret the following format as QueueName
> >  direct://amq.direct/mydest?option='1'.
> >  If the user wants to only specify destination then they must explicitly
> use
> >  the following format.
> >  topic://amq.topic/myDest/?option='1'
> >
> >  Comments and suggestions are most welcomed.
> >  Besides can somebody please explain what the destination really means?
> and
> >  also how it used?
>
> Destination is the name lifted from the Java Client for what is the
> routingkey because JMS does not have the concept of routingkey.


IMO JMS doesn't really need to worry about the routingkey, exchange etc . at
all.
The Destination class provides that nice abstraction.
So I am not really sure why we need that extra field destination as it
currently doesn't map very nicely to any AMQP concept.
However I ensured that the current code fucntions exactly the way it used to
treat "destination" before.


> So what does destination really mean. In the M2.1 code it means the
> routingkey and is only used by the Topic and Headers Destinations.
>

Currently for producers we allow you to use any exchange with any
routingKey.
For consumers we allow them to bind a queue to any exchange with multiple
binding keys.
So for 0-10 "destination" doesn't really make much sense. But it is fully
supported in the code for backward compatibility.

Rajith


> >
> >  --
> >  Regards,
> >
> >  Rajith Attapattu
> >  Red Hat
> >  blog: http://rajith.2rlabs.com/
> >
>
>
> --
> Martin Ritchie
>



-- 
Regards,

Rajith Attapattu
Red Hat
blog: http://rajith.2rlabs.com/

Reply via email to