[cgl_discussion] Using XML-RPC for Remote Access to openHpi L
Cress, Andrew R
andrew.r.cress at intel.com
Wed May 7 07:01:18 PDT 2003
OK, but I think this may need more explanation.
In the brief explanation in section 3.7, what appears to be key is that
where the platform provides a proxy or remote interface, such as
CMM-to-blades, HPI sessions should use it seamlessly. However, I'm
concerned when section 3.7 is interpreted so that all HPI sessions opened by
a given library implementation have to go through RPC or similar before
accessing the platform resources, which are often local to the library. I
don't think that was the intent of the spec was it?
If the RPC-like stuff is intended to be skipped for local resources, then I
think we're on the same page.
From: Sean Dague [mailto:sean at dague.net]
Sent: Tuesday, May 06, 2003 3:43 PM
To: Cress, Andrew R
Cc: 'Rusty Lynch'; Steven Dake; David Judkovics; cgl_discussion at osdl.org
Subject: Re: [cgl_discussion] Using XML-RPC for Remote Access to openHpi L
On Tue, May 06, 2003 at 12:22:59PM -0700, Cress, Andrew R wrote:
> I had been thinking that the HPI interface was primarily intended for
> where the element manager (or some HPI-compliant stub) would reside
> and the remote communication would be handled above the HPI interface, not
> within the HPI library implementation as seems to be implied here. If it
> remains above the line, there is a lot more flexibility of LAN/WAN
> communication paths.
HPI is definitely intended to be remotable from my reading of the
specification. Reference Section 3.6 and 3.7 of the HPI spec for more
information on this.
If a consumer of HPI decided to use HPI only in a local context, and do
remoting above that line, that would be valid policy for that user.
However, in general, there exists a need to have remotability of HPI
Sean Dague Mid-Hudson Valley
sean at dague dot net Linux Users Group
There is no silver bullet. Plus, werewolves make better neighbors
than zombies, and they tend to keep the vampire population down.
More information about the cgl_discussion