Full feedback for merged SCE-MI 2 draft

This is a re-send - first send may not have worked ...

Greetings ITC Techies,

I have my full feedback for Brian's merged SCE-MI 2 draft -
both the main body and appendix.

I'm sending it in the form of a marked up document
to Brian as a separate e-mail but my changes are summarized
below.

Before I mention them, I have general concerns about the
separation of SCE-MI 1 from SCE-MI 2 and favor a
slight reorganization of the ordering of the chapters
and appendices to make this separation clearer. Right
now, to an uninitiated user, there does not emerge a clear
view that SCE-MI 2 and SCE-MI 1 can be used independently
of each other, or together, which is a very important notion.

I see there were some e-mails on the reflector on this
subject so I'll respond to them separately in more detail with
my views.

But, aside from this issue, here is a summary of my
detailed feedback on the current draft from Brian.
I had reset change tracking from Brian's draft so
that only my feedback is highlighted.

- All references to try_flush() function on the HDL side
   removed. Some were mistakenly left in from before even
   though the functions themselves were removed as per discussion.

- Section 4.8.6 had some inaccurate bullets which I removed. I added a new
   one to replace them that is more accurate.

- Section 4.9 - improvement of wording.

- Section 4.9.1 - Minor corrections, improvement of wording, request for
   clarification

- Section 5.7.4.1 Changed scemi_pipe_c_can_send/receive() to remove
   num_elements arg which is redundant. These functions return num_elements
   instead.

- Add HDL side can_send(), can_receive() which had been been in
   the Cadence "ok_to_put/get" proposal but removed. Only the
   ok_to_put, ok_to_events should have been removed. can_send(),
   can_receive() can be kept because they are, as Cadence pointed
   out, quite useful for non-blocking polling status of a pipe
   without committing to any operation as is done with
   try_send()/receive(). Other than this, can_send()/receive()
   have similar non-blocking semantics to try_send()/receive()
   and so should be kept.

   So, I've reinstated can_send(), can_receive() to make the
   HDL pipe i/f's complete.

- Reworded Shabtay's section (5.6.6) on SCE-MI 2 support for
   DPI utility functions as per my AI from last meeting.

- Minor correction in TLM proxy example in appendix D2 to remove
   2nd argument from scemi_pipe_c_can_send() call as described
   above.

-- johnS
______________________________/\/            \     \
John Stickley                   \             \     \
Mgr., Acceleration Methodologies \             \________________
Mentor Graphics - MED             \_



-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.