Re: POA & obj key creation ... (fwd)

[Michael Meeks <michael ximian com>]

POA man to the rescue ?

	:-)

		Michael.

---------- Forwarded message ----------
Date: Mon, 23 Jul 2001 14:14:30 -0400 (EDT)
From: Elliot Lee <sopwith@redhat.com>
To: Michael Meeks <michael@ximian.com>
Subject: Re: POA & obj key creation ...

On Sun, 22 Jul 2001, Michael Meeks wrote:

>       It seems something badly broken is (not) going on with object
> key creation: it seems that the length of random buffer created by
> different routines is of strangely different lengths:
>
> . ORBit_POA_make_sysoid uses poa->obj_rand_len which is init to 0
> . ORBit_POA_object_to_okey uses poa->koid_rand_len "  " "    "  0
> . ORBit_POA_new initialised poa->poa_rand_len to 8 - a sensible
> number, used it seems (only) to generate a key for the poa.
>
>       Do we need to add
>
>   poa->obj_rand_len = 8;
>   poa->koid_rand_len = 8;
>
>       to ORBit_POA_object_to_okey ? what is the difference between
> these three numbers to merit their being a different length ?

Hmm, I would guess that some of those rand lens are zero in conjunction
with the persistent object junk. That wouldn't be consistent with
poa->poa_rand_len = 8, though.

poa->*_len definitely need to get set only at:
        POA creation time
        When some POA policies are set (persistent oid's, I think)

It sounds like obj_rand_len is used when generating a random part of
system object ID's, whereas koid_rand_len is used when generating a random
part of all object keys (system oid's vs all object keys), and then
poa_rand_len is used to generate the random poa part of all object keys.

-- Elliot
If a single monkey presses random keys on a single typewriter for a single
second, it will discover that it needs to be subscribed to gnome-gui-list
to post.

-- 
 mmeeks@gnu.org  <><, Pseudo Engineer, itinerant idiot