<Sunil>Global Unique IDs (GUIDs):
Generating GUIDs is a nasty problem. One easy way to solve would be that
Sender asking the Receiver to generate one for it before it initiates
any reliable message.</Sunil> First, I am puzzled by "is a nasty
problem" statement. Second, even if it was
<SK>
Could you then enlighten me with an errorprone-less GUID generator logic
without a case where clients don't talk a centralized server? Yes,
it's
RFC2822 mandated MessageId, but still to me it is a complex (&
nasty :))
one. Especially in a "reliable" spec., don't you want to be 100% sure
that
the Id is indeed "globally" unique.
</SK>
"a nasty problem" then it looks to me that it would be "even nastier"
to ask and get it from the other side.
<SK>
Before I clarify further, I really meant to say that this be an optional
feature.
for the receiving RMP (Reliable Message Processor). It's been
a long day
and I just missed it.
Yes, the solution proposed could be "nasty" (performance wise) if there
are 100 msgs and if the Sender has to ask the Receiver to send a
GUID everytime. However, if we agree this in principle, we could easily
solve this
by the sender making a single request to (block &) get a batch
of GUIDs
from a receiver and use them for the remaining course until exhausted.
Btw, sometimes a 'nasty' performance problem is better than a 'nasty'
functional problem :)
I hope I clarified.
Thanks,
-Sunil
<SK>
Nikola