Gentlemen,
Based on the comments I have received and distributed to date, for which
I thank you all, I have taken a stab at updating the draft VSI-S in
accordance with my best judgement of concensus. The updated draft
is attached in both ps and pdf format. In summary, I feel that this
draft presents a communications model which is both more robust and more
internally consistent than the original draft, though it has become a bit
lengthier than I originally imagined. Following is a list of the
some of the issues and how I addressed them in the update. I invite
you to read this updated draft carefully and, as usual, offer
constructive criticism and point out errors, omissions, etc.
Regards and thanks, Alan
-
1. Target of VSI-S
Issue: Should VSI-S be targeted only at
machine-to-machine control of the DTS, or should it include a user
interface?
Recommendation: Based on the comments I received, I judge
that VSI-S should be targeted at machine-to-machine communications.
The inclusion of check codes and sequence numbers, which I think are good
ideas, simply do not lend themselves to simple user interaction. It
is certainly possible that a similar user interface could be created
which would allow simple operation from a dumb terminal (by dropping
checksums, sequence#'s, etc), and VSI-S could make some non-binding
recommendations. So, the bottom line is that VSI-S is targeted at
machine-to-machine communications. With this decision, some of the
other problems, such as typability, simple disappear
-
2. Checksums, sequence numbers, timeouts, retries
Recommendation: Adopt Crestech suggestions. Time-out is
0.5 seconds (except special cases), max 2 retries. Time-sensitive
commands are exempt from retries.
-
3. Handling of Unsolicited responses
Recommendation: Allow system to be used in either half-duplex
or full-duplex mode. For half-duplex, adopt Crestech/Ferris
recommendations for special status command to retrieve unsolicited
messages. Full-duplex system has no problem with unsolicited
messages.
-
4. Additional commands
Recommendation: You will note several additional
commands and queries have been added, in accordance with suggestions by
several people. The DTS can now be queried for DTS ident,
media ID, station ID, source ID, software version#; and a diagnostic
self-test can be requested.
-
5. Ethernet port number
Recommendation: A particular Ethernet port number is
suggested as a 'standard' registered port number, in the range
1024-49151. though we don't yet have the number. (Anyone want
to volunteer to register a port number for VSI-S?)
-
6. Comments and informational commands
Recommendation: Non-operational commands are
called 'informational commands' and can have special interactions
with P/QDATA. A comment is just an informational command.
-
7. Multiple commands
Recommendation: With the new packet structure and
sequencing, multiple commands creat an unnecessary complication.
Drop them.
-
8. PDATA/QDATA
Recommendation: For consistency, only VSI-S-defined
commands (in VSI-S packet form) are allowed on P/QDATA lines.
Various options exist for passing ignoring or acknowledging PDATA command
packets, and for generating QDATA command packets.