[omniORB] parsing stringified IOR

Stephen Crawley crawley@dstc.edu.au
Tue, 30 Nov 1999 13:00:23 +1000


> I think I disagree.  From a diagnostic standpoint, such as in a log that a server might
> generate, I as a sysadmin , or user, would find it useful to know when/how/why a server
> died/restarted/migration.
> 
> That is my need.  I do not want to parse the IOR as part of the normal interface at all.
> Simply a diagnostic routine.

Rob,

You can disagree all you like!!

The >>fact<< is that there are no standard APIs for parsing IORs, and
if there were, they wouldn't tell you anything about where the corresponding
server actually is at the moment ... or why / when / how it crashed.
All of that stuff is >>deliberately<< not standardised by the CORBA specs,
and with good reason.

If you want to get a handle on how IORs work, and (by inference) why it
is not a lot of use knowing what is in them, try reading Michi Henning's 
article on CORBA object location.

  http://www.triodia.com/staff/michi/binding.txt

This having been said, if you are prepared to make sweeping (and probably
unsafe) assumptions and use non-standard, non-portable APIs, you could in 
theory find out where an object is currently instantiated.  But it would
be hard work, not to mention flakey, non-portable, etc, etc.

-- Steve