[rabbitmq-discuss] Passing null for 'arguments' to queueBind()

[Michi Oshima]

Hi Matthias,


Matthias Radestock-3 wrote:
> 
> There is a problem with all of the above approaches...any routing key / 
> header included in the message will be passed to *all* consumers, thus 
> allowing consumers to learn each other's secrets.
> 

I'm not sure if this is true.  My thinking is that if a consumer doesn't
receive a message, the consumer cannot inspect its properties.  

Does a consumer get all messages, and is it up to some internal logic of the
consumer to filter out messages that aren't meant for the consumer?  I would
have thought this type of logic would be inside an exchange and kept away
from consumers.

For now I'm assuming we aren't back to the drawing board.  Your options #2
and #3 sound promising.  (Although I'd have to learn Erlang first.)


Matthias Radestock-3 wrote:
> 
>> 2) Write a custom exchange type plug-in. This could work exactly like a 
>> headers exchange but not permit empty bindings. Or it could work more 
>> like a direct exchange, with the producer perhaps listing the tokens in 
>> the message's routing key, comma-separated.
> 

I think I might try this one.  The custom exchange type will be basically a
headers exchange.  The only difference is that a null header will match
nothing: that is, it won't permit empty bindings.  

I'm so far failing to dig up good information for implementing custom
exchange type.  If you know of a good link, please let me know.


Matthias Radestock-3 wrote:
> 
>> 3) Write a proxy that prevents empty bindings.
> 

Could you direct me to more information about this also?  This proxy is
something new to me.  And again, I'm failing to locate information about it.  

In violation of mailing list etiquette, I thank you very much for your
replies.

Michi Oshima
-- 
View this message in context: http://old.nabble.com/Passing-null-for-%27arguments%27-to-queueBind%28%29-tp30162963p30182063.html
Sent from the RabbitMQ mailing list archive at Nabble.com.