On Thu, 22 Apr 2004, Joe Touch wrote: > The "tel:" part is sufficient to get you to VOIP - in fact, that's what > tel: ought to mean -- no more, no less. If you want IBM to differentiate > the switchboard from the headquarters, try: > > tel://ibm.com/hq - headquarters-specific > tel://ibm.com/ - default switchboard > > > Similarly, I'd expect that tel://ibm.mobi gets me to the IBM Cellular > > switchboard. > > tel://ibm.com/cell > > I.e., that's a decision IBM gets to make; others can decide, e.g., that > cell and regular calls all go through the same VOIP gateway. Well, why not just run everything over http and have an application-type header to destinguish the different formats? Possible, but not quite ideal, since then you are stuck with an HTTP server in the middle. > > I recall that gte internetworking used gte.com for internal corporate > > addresses and gte.net for customer addresses. Some companies use > > subdomains for such purposes. > > Sure, but that is different than the above. In the tel: cases, all the > addresses are for internal corporate, not for a service IBM runs for its > customers. Umm, I don't see how. Its just using naming, and specifically different TLD's to make a distinction between corporate and customer functions. No matter what you do, you are probably going to use naming anyway, either as a url path component, or as a subdomain, or as a new TLD. I don't see any significant operational difference(*), nor any particularly significant technical differences to them. Except, I'd probably prefer not to send everything through http, so that tends to go against the use of url pathnames and application-types as discriminators. What difference does it make to the IETF whether there are more TLD's or less? (*) Having more TLD's ought to improve the reliability of the internet in general by distributing load off the overused com and net domain servers. Whether you have 259 TLDs or 2500 isn't going to make a great deal of difference to the root servers, as these aren't particularly large zones. 2500 TLDs would no doubt make some measurable impact on the roots, but it would make a hugely big difference in directing traffic off the com and net servers which have millions of entries. And we aren't talking about 2500 TLDs, though. We are talking about just adding a handful. This isn't a significant difference, and may not even be measurable. On the other hand, I would also note that the com and net domains are commercial, and charge fees for their operation. I don't see that the IETF or ICANN necessarilly cares either way how much that operation costs, or that it should be concerned very much with reducing those costs with more TLD's. But I'd also say that extreme disregard for costs would be a bad thing. So I'd say that's a wash. _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf