Re: Message Etiquette

["darrenp_lock" <darrenlock@xxxxxxx>]

Gregg,

thanks for your comments. No, I hadn't appreciated that a listening
port would be created for each logical device. Therefore, I will
propose the following format for the Logical Address:

DGarage.vxvml.<servername>:.....

Now my question here on convention is should the servername be the
Asterisk Server being managed or the servername where the Gateway is
located (the gateway runs on Windows)? The former would allow for
multiple Asterisk Servers and probably is more intuitive.

I take your point about the dynamic endpoints in the Sub Address. I
was worried about maintaining the UIDs for the folders & messages. So
I guess I can just address them in the message block. Having a device
for the folder meant commands like DeleteAllMessages could be applied
at the Device level without message block processing. However, if
DisplayAllFolders for a MailBox is going to generate a xAP Message
per folder then the context (FolderID) would be in the message block,
so message block processing is going to be required anyway.

An example of message block processing required is when the
DisplayFolders (for Mailbox) command is executed and a xAP message is
issued per folder indicating the MailBox, FolderName and number of
VMail Messages. Ideally, I would like James' HomeSeer Gateway to
create and maintain me a HS Device that represents the number of
Messages in a particular mailbox folder (as you can guess, I haven't
thought that one through yet!). For info, those devices are then
gatewayed out of HS and into MainLobby. An alternative might be to
create a bespoke MainLobby plugin to process my xAP Messages -
however, I loose the benenfit of having 'device' information in HS.

Finally, given that I can never control (unless my Gateway imposes
limitations), the total number of mailboxes, folders and messages and
that they all have the potential to be very dynamic, I believe the
multiple message approach will work best.


Rgds, Darren.


--- In xAP_developer@xxxxxxx, Gregg Liming <gregg@...> wrote:
>
> Hi Darren,
>
> Quoting darrenp_lock (4/11/07 4:26 PM):
>
> > Firstly, addressing:
> >
DGarage.vmxml.<context>.<mailbox>:<folder>.<message>
>
> I'm somewhat assuming you understand the implications (as I know
them),
> but ... just to be clear ... In the above example, each mailbox is
bound
> to a separate listening port and is it's own device.  I'm assuming
that
> you have good/valid reasons for doing so.  But, all things aside, I
> probably would have mapped the server to the device and then
cascaded
> (in whatever fashion makes sense) the endpoints.  Again, I don't
presume
> to understand your design; just want to highlight issues along the
way.
>
> > Where <mailbox> is the mailbox name/number, <context>
allows for
more
> > than one Asterisk VoiceMail Context (I am also making the
assumption
> > that there will not be more than one Asterisk VMail Server on the
> > network with the same context). <folder> is used to address
commands
> > at a particular folder within the mailbox and <message> is
used to
> > perform message related commands on messages within a mailbox
folder.
>
> This reminds me of the debate that I've had w/ Kevin regarding
> transient/dynamic "endpoints".  The folder example is
arguably
> relatively "static" and ok (in my opinion).  The part that
seems
> questionable (again, in my opinion) is use of an inherently
> dynamic/transient concept like a "message" being an
endpoint.
>
> > Now my question is how to best construct the messages. For
example, I
> > have a command to list all folders and their attributes (e.g.
message
> > count)in the mailbox. Should I have a multi block message or
multiple
> > messages per folder.
>
> You can't if you use the above endpoint strategy as the folder
isn't an
> endpoint.  On the other hand, if you forgo the <message> as an
endpoint,
> then that seems reasonable.
>
> > The same goes for listing the messages in a
> > folder. I conscious of low spec devices listening on the network.
> > What works best, lots of small messages or one big message?
>
> It's not clear to me what information you're trying to convey (and,
may
> well be a reason to ignore all of my comments).  But, assuming it
might
> be (1) semi-verbose, (2) you're exceptionally popular and (3) you're
> away on vacation: your transported content might exceed the packet
> length if you're contemplating sending one big message. (I'm
tempted to
> offer comments toward the liklihood of #2 since you're sending to
the
> dev list, but I'll be polite).  Unless there are clear practical
bounds
> that would likely prevent it, I'd opt for more, smaller ones to
avoid
> any likelihood for busting packet length limits.
>
> Gregg
>