Re: Send-N draft: draft-bellis-enum-send-n-02 published

1000 entries in a database is laughably small.  Downloading it to a phone as
the country code is dialed is even easy (although I suspect you wouldn't do
that to a wireless phone with current data rates, you might with 4G data
rates).

Here is a more or less concrete proposal:
The regulator publishes, in the existing ITU document, a URI of its current
dial plan.  The URI yields an XML data structure that looks like:
<dial-plan globalprefix="+1">
...
<assignment operator="ATT" use="wireless" prefix="202555" min=4 max=4/>
...
</dial-plan>

The operators do the same with codes that are different min/max in the
regulators data.  So, let's say there was an assignment like
<assignment operator="BT" use="wireline" prefix="456789" min=3 max=7/>

Then BT may have
<dial-plan globalprefix="+444567898">
...
<assignment prefix="123" min=1 max=1>
...
</dial-plan>
Which means that +445678981232 is a good TN.


Now, that's quite verbose.  It would be very simple to turn that data into
an MGCP digitmap, or a Finite State Machine algorithm, or some really
compact binary tree thingy that encoded the same information in far fewer
bits.

I chose the XML because it closely represents the form the data is now
provided.   Any other simple to understand form would be fine.

What is important to me is that it be something the authoritative source of
the data can reasonably publish, and that it work for the cases which are
easy as well as the cases that are hard, using one of my favorite dictums
that "simple things should be simple, complex things should be possible".

Brian

> -----Original Message-----
> From: [email protected] [mailto:[email protected]]
> Sent: Friday, June 27, 2008 10:19 AM
> To: Brian Rosen
> Cc: [email protected]
> Subject: RE: [Enum] Send-N draft: draft-bellis-enum-send-n-02 published
> 
> > What you want is a dial plan.  Fine, we should standardize a way to get
> a
> > dial plan for every TN.  In nearly every case, that comes from the
> > regulator.  There are conditions where what you get from the regulator
> is a
> > dial plan that determines the end of a "prefix" and you have to go
> somewhere
> > else to get the rest of the dial plan.  Fine, we can do that.
> >
> > 99% of the telephone numbers in the world have a pretty simple dial
> plan.
> > That's the case we should optimize.
> 
> I don't believe it's as simple as you say.
> 
> The UK dial plan alone contains 1000s of entries.  Germany's is another
> 100+.
> 
> > Overlapped dialing decoding is walking through a dial plan.  We should
> > provide the means to do that.  All of the hundreds of millions of phones
> you
> > are talking about work fine without Send-N.
> 
> But those existing phones are connected to switches that *do* use Send-N
> in C7  (which of course is what this is named after).
> 
> > They have the dial plan for the home country burnt into the switch, and
> they use
> > a time out for international numbers.
> 
> The large number of dial plan entries for the UK means that for some
> number ranges (in particular +44800) they're having to use inter-digit
> timeouts, and they're not happy about it.
> 
> The UK carriers are therefore actively trying *not* to have the UK dial
> plan burnt into their switches, and this is the proposed implementation.
> 
> > There is value in improving the dialing experience of everyone, but not
> so
> > much that we need make it really optimal.
> 
> The carriers consider any use of time outs to be unacceptably sub-optimal.
>  Our proposal avoids them completely.
> 
> > You have brought us a problem I
> > think is worth solving: determining dial plans for international
> numbers.  I
> > don't like your solution, but I'm happy to work on the problem.
> 
> The more the merrier :)
> 
> cheers,
> 
> Ray