The IP over ATM Mailing List Archive by date

[jhalpern@newbridge.com (Joel Halpern)]

James Watt wrote:
>3) I would like to see us make an attempt to reduce the distinction
>   (implied or inferred !!) between the mesh and server cases.  We need
>   to ensure the end-station behaviour is as simple to understand as we
>   can make it.

In response to which, Tim Smith wrote:
> Just for clarification, could you help me understand item 3. Are
> you saying that an end-station behaves differently depending on
> whether an MCS or a mesh is used?  I was under the impression that
> end stations were oblivious to the fact.

Since I first brought this up in two of my comments, I will attempt to
clarify.  No, the ATM Attached client host does not behave any differently
sending to (or receiving from) mesh vs server groups.  However, in reading
the document this is not clear.  The way the document is organized, I
found myself continually wondering through the first half of the document
"How does this work for servers".  For example, I wondered what the meaning
would be if a client got multiple addresses in the server case.  Much later
in the document there is a note that says that such a case is for further
study.

In general, the specification is laid out like a book instead of like a
spec.  Dealing with the mesh case completely, and then describing the
changes to MARS behavior for the server cases is a nice style, but it
produces unclear specification.  One has to integrate two separate parts
of the document in order to determine what the MARS does.

Like, the matters of passive voice, and the confusion about what is the
common header and what is specific to individual requests, these are all
matters of clarity.  Yes, it would be better if we could have made these
kinds of comments earlier.  For myself, it was not until recently that I
was able to put my finger on why I continually had trouble reading the
internet-draft.  

Yours,
Joel M. Halpern				jhalpern@newbridge.com
Newbridge Networks Inc.