Re: dispatch: getting rid of QPID_DISPATCH_HOME in python scripts.

2014-06-03 Thread Alan Conway
On Mon, 2014-06-02 at 16:55 -0400, Darryl L. Pierce wrote: On Mon, Jun 02, 2014 at 04:42:21PM -0400, Alan Conway wrote: And how are you going to prevent those irritating users from looking at /usr/bin/qdstat and observing that it imports all its bits from /usr/lib/qpid-dispatch/python/blah,

Re: dispatch: getting rid of QPID_DISPATCH_HOME in python scripts.

2014-06-02 Thread Alan Conway
On Fri, 2014-05-23 at 11:28 -0400, Justin Ross wrote: On Fri, May 23, 2014 at 11:16 AM, Alan Conway acon...@redhat.com wrote: Scripts like qdstat currently load the qpid_python_internal module in a special way by using the value of the env. var QPID_DISPATCH_HOME. This is very weird and

Re: dispatch: getting rid of QPID_DISPATCH_HOME in python scripts.

2014-06-02 Thread Darryl L. Pierce
On Mon, Jun 02, 2014 at 04:42:21PM -0400, Alan Conway wrote: And how are you going to prevent those irritating users from looking at /usr/bin/qdstat and observing that it imports all its bits from /usr/lib/qpid-dispatch/python/blah, and going ahead and using it anyway? This is python, we can't

Re: dispatch: getting rid of QPID_DISPATCH_HOME in python scripts.

2014-05-29 Thread Alan Conway
On Wed, 2014-05-28 at 08:57 -0400, Darryl L. Pierce wrote: On Tue, May 27, 2014 at 10:43:48AM -0400, Alan Conway wrote: On Fri, 2014-05-23 at 11:40 -0400, Justin Ross wrote: In which scenarios are you having to set it? If you produce two builds to two distinct install prefixes, they

Re: dispatch: getting rid of QPID_DISPATCH_HOME in python scripts.

2014-05-28 Thread Darryl L. Pierce
On Tue, May 27, 2014 at 10:43:48AM -0400, Alan Conway wrote: On Fri, 2014-05-23 at 11:40 -0400, Justin Ross wrote: In which scenarios are you having to set it? If you produce two builds to two distinct install prefixes, they should be pre-defaulted with the approprate home dir, so you

Re: dispatch: getting rid of QPID_DISPATCH_HOME in python scripts.

2014-05-27 Thread Alan Conway
On Fri, 2014-05-23 at 11:40 -0400, Justin Ross wrote: In which scenarios are you having to set it? If you produce two builds to two distinct install prefixes, they should be pre-defaulted with the approprate home dir, so you shouldn't have to set it. If I do tests directly against a build

dispatch: getting rid of QPID_DISPATCH_HOME in python scripts.

2014-05-23 Thread Alan Conway
Scripts like qdstat currently load the qpid_python_internal module in a special way by using the value of the env. var QPID_DISPATCH_HOME. This is very weird and surprising. It creates havoc if you have multiple installs or builds of dispatch and assume that you just need to set PYTHONPATH to make

Re: dispatch: getting rid of QPID_DISPATCH_HOME in python scripts.

2014-05-23 Thread Justin Ross
On Fri, May 23, 2014 at 11:16 AM, Alan Conway acon...@redhat.com wrote: Scripts like qdstat currently load the qpid_python_internal module in a special way by using the value of the env. var QPID_DISPATCH_HOME. This is very weird and surprising. It creates havoc if you have multiple installs

Re: dispatch: getting rid of QPID_DISPATCH_HOME in python scripts.

2014-05-23 Thread Justin Ross
In which scenarios are you having to set it? If you produce two builds to two distinct install prefixes, they should be pre-defaulted with the approprate home dir, so you shouldn't have to set it. On Fri, May 23, 2014 at 11:28 AM, Justin Ross jr...@apache.org wrote: On Fri, May 23, 2014 at