Re: [gupnp] [PATCH] Respond to M-SEARCH requests for urns specifying an eligible version

["Zeeshan Ali (Khattak)" <zeenix gmail com>]

Hi,

On Tue, Jul 13, 2010 at 5:29 AM, Stephen Depooter <stephend xandros com> wrote:
>
> If I have a device that implements versions 1, 2, and 3 of Media Server, and as per section 2.3 of the spec, includes and declares version 3 as the "highest supported version" of the MediaServer device type, then only the versions 1, 2, and 3 of MediaServer are supported.  Clearly the device must respond to any search request for a MediaServer on the network when the request is for a supported version.  Since by definition version 4, 5 and 6 have new functionality compared to versions 1,2 or 3, then they are not supported.  Thus any request for a version 4, 5, or 6 is not supported and should not be responded to.  Since the control point that is dong the searches receives no responses, it MAY fall back and search for a lower version if it is behaving according to the details of the 1.3.2 section of the spec.  Clearly the expectation is that if a control point is requesting a version of a device that is not supported it will not get flooded by irrelevant responses by every device that happens to implement only lower versions of the spec.
>
> Any without providing a copy of the UPNP standards test tool, I can't hit you with the clue bat that clearly shows the expected behaviour as tested by the standards committee.  As written, the section 1.3.2 of version 1.1 of the UPNP standard, does not say that a device MUST not respond to any goddamned irrelevant M-SEARCH request, it only says that it must respond to requests for devices/services that ARE SUPPORTED.
>
> I don't know how the DLNA certification body's test tool tests this aspect or even if it tests this, but if you want to support UPNP body certification, then you need to pass the UPNP standards body's test tool.
>
> And with that, I'm done.  I won't waste anymore of my time trying to get this patch set accepted, feel free to merge the changes
>from the posted m-search branch on gitorious.

   First of all, I already said that your set of changes to check for
sanity of the M-SEARCH are good. The only reason they didn't get
merged was that you have put them together with this other change that
is different and was under discussion.

   Secondly, I got the wrong impression that you want to change the
check for version because of some specific requirement in the spec.
Now that you have made it very clear that that is not the case and
provided a compelling argument, I'll be willing to accept this change.
However, since we are making use of regex, I would prefer if you could
come-up with an alternate patch that changes the regex to only match
equal or lower version but not higher.

-- 
Regards,

Zeeshan Ali (Khattak)
FSF member#5124
--
To unsubscribe send a mail to gupnp+unsubscribe\@o-hand.com