OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

ws-rx message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: Head to head comparison of all three MakeConnection variants


In an effort to clarify where/why BEA stands on PR001 I have written the
attached article comparing all three variants off the MakeConnection
feature (Sequence-ID, RM-Anon, and PR001 proposal). I scoped the
comparison to the two (count them!) use cases BEA actually cares about.
I concentrated on two major implementation aspects:

1.) How does the server-side RMS assign message(s) to the server->client
sequence?

2.) How does the MakeConnection message select which message(s) should
be returned over the back-channel.

For those of you too busy to read the article here's the conclusions:
-------
The only effective difference between the PR001 Mechanism and the
Sequence-ID Mechanism is that the former allows for the dynamic creation
of new server->client sequences. Since none of the use cases of interest
to BEA require this capability, this difference is irrelevant in the
scope of this analysis. Given that the PR001 Mechanism is more
complicated than the Sequence-ID Mechanism, BEA sees no compelling
reason to replace the Sequence-ID Mechanism with the PR001 Mechanism.
Furthermore the RM-Anon Mechanism is preferable to both the Sequence-ID
Mechanism and the PR001 Mechanism due to the fact that it is more
flexible with respect to assigning response messages to "back-sequences"
and does not require the server-side RM handler to track the
relationship between request and response messages.

- gp

 <<MC-Head2Head.zip>> 

MC-Head2Head.zip



[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]