[rabbitmq-discuss] rabbitmq erlang client

[Ben Hood]

Orion,

On Sat, Nov 8, 2008 at 11:17 PM, Orion Henry <orion.henry at gmail.com> wrote:
> Thanks for the update.  Glad you guys are making the erlang client a
> first class interface.  Its not going to be easy to convince ops to
> let us run the edge version of the server in production.  Is there
> some way I can compile the client against the edge headers but keep
> the official release in production, since I'm using the network/wire
> protocol?
> If not I'm going to stick to old version of the client until rabbitmq
> 1.5 and client 1.0 are released.  So far its been working great. =)

I'm not sure I follow - maybe I'm being thick, but how do the *edge*
headers make a difference to the server version that has been
deployed?

If you are talking about the fact you need to run a non-released
version of the server to get the right version of the common libraries
for the client to work, and you are using the wire protocol, then all
you need to do is to have a different load path for the client and the
server. There are few easy ways to do this:

1) Run the client on a different machine :-)
2) Run the client and server as different OS users, and make sure the
client has the *edge* version of the common libraries in it's load
path;
3) With the same user, do the same as 2, but write your scripts to set
the load path appropriately.

This does however highlight some things we need to think about when we
release the client - Edwin started a thread about splitting out the
common libraries into a separate package (in fact, he commented on
this again today). I think one of the problems is the way that we
target Debian to a certain extent, but I'm not quite sure how it is
supposed to look yet.

HTH,

Ben