Re: xPL announcement/description protocol -- was xPLDiag

["Mal Lansell" <mal@xxxxxxxxxxx>]

I'm not proposing this in order to change the way hubs discover
applications and handle config - I think the existing system is
adequate, and as you point out, once set up, each app has its config
and starts straight away.

The changes I'm talking about are an addition to the existing
system, and are required in order to support the diagnosis of xPL
communications problems; to try to make things easier for newbies
who might miss a step, and to provide a detailed report so that the
more experienced user can see what is going on.

If we are to have an xPL diagnostic application, we don't want to
have to wait up to 9 minutes to discover all the devices on the
network - we need a way to check who is out there and report back
straight away.

To be most effective, this tool would also need more information
than is carried in a regular hbeat message.  The version number of
the application would help enormously when trying to debug a
problem - how many times have we told people to reinstall
the "latest version", because we can't tell what version they are
running to start with?

Gerry's version (extended hbeat rather than about.basic) does not
appear problematic - the proposed change is backwards-compatible and
would enable me to write a diagnosis tool that would make all our
xPL lives easier.

Mal


--- In ukha_xpl@xxxxxxx, "Tony Tofts" <tony@x> wrote:
> Aren't we over engeering stuff here?
>
> Firstly an app doesn't (shouldn't) take up to 5 minutes to start
working
> without a hub. An app is supposed to bind it's hub port regardless
and start
> using it immediately, even if there's no hub
>
> Only config can take up to 5 minutes at start, and this only
applies to
> app's/devices that don't do the increased 1 minute "config
me" at
start up
> and can't store their own config.
>
> Almost all app's store their config, so in the vast majority of
app's there
> is no delay whatsoever once a hub is started theyre all up and
running
> (regardless of the boot order)
>
> If the above isn't happening for some people, theyre's something
wrong
> somewhere - and this should be addressed before anything else
surely.
>
> I personally can't see any reason why there should be any need for
any
> changes, the protocol works as long as app's/device's are adhering
to it,
> and no amount of diag/ping messages should have any bearing on the
working
> of xPL
>
> If we want an about.basic to provide other information fair
enough, but this
> shouldn't be used to patch over badly acting app's/devices
>
> Regards
> Tony