<div dir="ltr">Ben,<br><br>Thanks so much for looking into this and giving it your consideration. I'm happy that you found it of some value. I was a bit worried that you would have concerns about the performance of the modified code, given that it's no longer a direct module call. I read your comments below with interest, and I am watching and learning. I'll be very curious to see the actual finished code when it's released. I'm glad the direct driver problem was just a commented-out line of code - I admit I did do it all in a hurry as a proof of concept. <br>
<br>I have in the meantime created my own little rabbitmq_common application, used only by the Erlang client, and it is all working famously and nicely decoupled from the server. I trust that the changes you are talking about will come out in the next release of RabbitMQ, so when that happens, I'll change my code to use the "official" client and whatever the common app is called.<br>
<br>Regards,<br>Edwin<br><br><div class="gmail_quote">On Wed, Sep 10, 2008 at 6:54 PM, Ben Hood <span dir="ltr"><<a href="mailto:0x6e6562@gmail.com">0x6e6562@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Edwin,<br>
<div class="Ih2E3d"><br>
On Mon, Sep 8, 2008 at 8:19 AM, Ben Hood <<a href="mailto:0x6e6562@gmail.com">0x6e6562@gmail.com</a>> wrote:<br>
>> I'd be very interested to hear your thoughts on this approach, and whether<br>
>> or not you can get the direct driver test to work.<br>
<br>
</div>I've started a branch for this in hg, it's called bug19373.<br>
<br>
All that I've done so far is to apply your patch and fix the line that<br>
you commented out that made the direct test fail (you un-exported the<br>
start/2 function, which is required for the direct client).<br>
<br>
I do have a few observations that I'll just work into that branch -<br>
<br>
1. The string building is probably unecessary - you could just pass<br>
the module name in 1:1<br>
2. I think you took the comment in the unregister/2 function a little<br>
too literally, I don't think it need any driver specific<br>
differentiation. If it did, then the correct thing to do is to<br>
encapsulate this in a callback and let the driver handle it.<br>
<br>
Also, this will all be dependent on another bug that I've started to<br>
do the necessary splitting in the server code base.<br>
<br>
Just wanted to give you a heads up to let you that we're not<br>
completely slacking off here.<br>
<br>
As for time lines, ATM this is about priority 2 behind some major<br>
server refactoring, so I'll let you know as soon as I can.<br>
<br>
HTH,<br>
<font color="#888888"><br>
Ben<br>
<br>
</font></blockquote></div><br></div>