[mythtv-users] How to handle schedule pre-emptions?

[Brad Templeton]

On Fri, Apr 29, 2005 at 02:39:07PM -0700, Brad Templeton wrote:
> Well, if some brave volunteer wanted to offer a service to notice pre-emptions
> and broadcast database updates for the networks, I could produce code
> to read these off the web or an e-mail to update the database.
> 
> A mailing list seems the most efficient, but it means that each user
> has to know how to redirect a mail or an alias into a process (such as
> a perl script.)  That changes from system to system so it's not really
> possible to have it slotted in out of the box.
> 
> Far less efficient would be polling a web address.  That could work out
> of the box but it's a _lot_ less efficient and depends on you polling in
> time.
> 
> But the main thing is somebody willing to be the trusted party who
> sends out updates.    You would write up the updates in something
> similar to the SQL that would be executed.

With a bit more thought, I came upon some additional realizations.

One more secure approach would be to send the zap2it style XML with
updates, and feed that into mythfilldatabase.

(One could also consider just a signal to tell the system to run
mythfilldatabase with an argument telling what site to fetch data from.
However, that would still need to be secured, as otherwise it could
cause a DOS attack)

However, it also occurs to me that unlike me, many people may not have
a mail server running anywhere that can connect to their SQL database.
While one could tunnel or play other tricks, the e-mail route might
leave a number of users unable to easily use the system.

The alternatives aren't great though.  One would be to develop an "alerts"
daemon in Myth.  This daemon would listen to a port (could be any port)
for a very simple command set.   One would need to open a firewall/NAT hole
to the port, and because of the security risks of doing so, the command
set for the port would be very small, and the commands probably signed
with a crytographic digital signature.   The commands would probably not
contain any data, they would be more along the lines of "You should
resync your database now" or "go get database updates from this IP address."

(Alerts could also still come via email for those who could handle that.
The alert could just trigger the same thing or even talk to the hidden
alerts port.)

While one obvious app is real time database update notifications, this
port could also exist for future collaborative or P2P applications in
the system.

This is all a lot more work.  E-mailed mythfilldatabase XML or SQL would
be vastly easier to implement, but I presume there are many who can't
get E-mail anywhere in range of their myth box.  (You could install
an MTA on your myth box but that's pretty drastic.)  E-mail, unlike an
alerts port, has lots of extra reliability measures to assure delivery.