Yes, by all means improve the description of the item. I had already suggested a rewording, but it can definitely use more tweaking.
I guess there is also the question as to what to return in cases where there are either multiple processes "using" the endpoint or where the process no longer exists: A) Returning one of the processes (i.e., the process that created or last attached to the endpoint) is useful. B) Returning a process that may no longer exists is fine too, as long as the PID isn't reused by another process (some operating systems don't recycle PIDs or only do so after a very long time). I would argue that it is up to the implementer, perhaps based on what is possible for the platform. - Bernie > -----Original Message----- > From: Erik Nordmark [mailto:[EMAIL PROTECTED] > Sent: Thursday, May 18, 2006 1:39 PM > To: Bernie Volz (volz) > Cc: [EMAIL PROTECTED]; ipv6@ietf.org > Subject: Re: Process id in UDP/TCP mibs? > > Bernie Volz (volz) wrote: > > Obviously that is less than desirable, but just because > some platforms > > may be unable to provide useful information doesn't mean we just > > deprecate it for all. Implementers of SNMP agents for this > MIB will thus > > need to decide whether they return a value for this item or > not, and if > > they do what caveats may exist for it. The text already has > "or zero if > > there is no such process." > > But that text would at least need to be ammended to include > - if there are multiple processes > - if the process can not be determined (for instance, because there > might have been multiple processes in the life of the udp endpoint) > > Erik > -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
