[mythtv] Caller ID

[Shawn Rutledge]

On Thu, Sep 18, 2003 at 08:54:28PM -0500, Kevin Elliott wrote:
> > The entirety of a "caller ID" notification packet is, in fact, about a
> > Caller ID event.  If you want to send other kinds of notifications on
> > the same UDP port, there's nothing stopping you from doing it; and
> > clients which are interested only in caller ID info can look at the
> > first tagname within the packet to decide whether they're interested or
> 
> Certainly you're not referring to the initial block. The initial block
> should almost always be a container tag describing the generic contents
> of the xml content within.
> 
> Frankly, I think creating a higher level mythrequest protocol would be
> more appropriate, to facilitate for future advancements in the rpc 
> (i.e. controlling myth remotely, other forms of rpc like data fetching,
> web services, etc)
> 
> For example:
> 
> <MythRequest version="1.0">
>   <alerts>
>     <alert type="callerId">
>       <name>Jack Johnson</name>
>       <phoneNumber>650 444 4444</phone>
>     </alert>
>     <alert type="message" delay="5">
>      <message>You have received 5 unanswered phone calls today</message>
>     </alert>
>   </alerts>
>   <execCommand name="flushLogs"/>
> </MythRequest>

Yes but you're talking about what kind of data to send to Myth, and I'm
talking about what kind of data can notify numerous client programs on
computers and terminals all over the house (or office) that the phone
is ringing.  I'm already using such a system at home:

http://ecloud.org:8080/~ecloud/caller-id/

My clients are Linux machines running X, and 3Com Ergo Audreys.  These
are not Myth boxes and there's no reason they ought to be interested in
MythRequests.  But Myth ought to somehow be able to display the same
Caller ID packets that the other machines on the home network can
display.

The consensus we seem to be coming to is that there needs to be a
translator daemon of some sort, which receives the Caller ID packets and
sends Myth a different notification tailored for on-screen display
purposes.  The message going to Myth should probably be done by a local
IPC method rather than via a UDP packet; but UDP packets are eminently
suitable for notifying everybody all at once that the phone is ringing,
as quickly as possible.

-- 
  _______                                                Shawn T. Rutledge
 (_  | |_)                      e_cloud at yahoo.com * http://ecloud.org:8080
 __) | | \________________________________________________________________