Re: Rep (2) : QUIPU vs X.500 (was: A tool for...)

[[e-mail]]

Colin,

We will try to send you a new report as detailed as possible.

About the QUIPU problem and the fact that it needs or not a
major revision or just a bug fix, we are not in a position
to answer, you are the real experts knowing about QUIPU
architecture. What we observe is that QUIPU is at least
extremely specific in some reference handling. 

It seems that for correct interworking there is need for
handling all the mandatory type of references.
We suspect that the current QUIPU reference choice and handling
is totally tied with the basic EDB choice assuming
that all the subordinates of one entry are by default
hold by the same MTA. Without having any knowledge of
QUIPU internals or even architecture I tend to believe
that Steve Kille is right when stating that this cann't be changed
without a major architectural revision, but again
you are in a much better position than us to know this.
A major interworking problem is the fact that QUIPU insists
on the fact that the master entry of a relative root of a subtree
is held by a different server than its direct subordinates.
This is
  . in contradiction with what other implementations have implemented
      and prevents full interworking when other implementations are
      not in a leaf position in the servers tree
  . seems to induce some strange behaviours about the reference
  returned in somes cases (reference to a server not holding
  the object concerned but to a one holding the subordinates)


hope this helps

Bruno will send to the paradise partners list a detailed
description with logs and PDU analysis, but it will not be different
from what we already sent you, just hopefuly better explained.
We would appreciate that you couldspend time enough to confirm
or infirm our analysis (of the problem, not on the impact on QUIPU)


regards

-- PAP