Re: [port-srv-reg] we need to make progress

[Joe Touch <[e-mail]>]

On 9/7/2010 4:30 PM, Stuart Cheshire wrote:
>> We only know that nobody registered "www" in your database; we cannot
>> know whether there are no apps that browse for this string.
>
> After about eight years of the whole world using "_http._tcp" you're
> claiming that some hermit somewhere is developing software that uses
> "_www._tcp" instead?
>
> That's about as likely as some hermit somewhere whose TCP stack uses IP
> protocol number 191 to mean "TCP".

Although I agree on the likelihood, simply declaring "there are no apps" 
is not how it's done. We would need to verify.

> Good luck convincing printer companies to update their firmware to
> advertise a different service type to accommodate a non-existent web
> browser that's looking for the wrong service type. If such a web browser
> existed they'd have heard customer complaints for the last eight years
> (but of course they haven't, because this mythical web browser doesn't
> exist).

They'd have heard complaints ONLY if they didn't interoperate with other 
vendors (maybe even just themselves) who do things the same way. And 
even if they had heard complaints, it's equally unlikely you would have 
known about that.

The overall point is we don't change things because of what we *think* 
is deployed; we always try to verify as thoroughly as possible.

>> IMO, there is no such thing as primary. It's ALL and ANY - register
>> ALL, lookup ANY. That's the only way we know things will work.
>
> That's a terrible idea for efficiency on the network, and battery life,
> and I can't see any vendor deciding to do something so pointless.

The whole efficiency of registering is based on the assumption that the 
number of registrations is much smaller than the number of lookups 
(i.e., writes << reads). Given that, there's hardly significant impact 
on network or battery life, and the potential benefit in 
interoperability (since nobody KNOWS for sure what the 'primary' name 
is) could easily be worth the coding effort.

Joe