On Fri, Jan 22, 2016 at 11:04 AM, Patrik Fältström <paf(_at_)frobbit(_dot_)se>
wrote:
On 22 Jan 2016, at 16:55, Phillip Hallam-Baker wrote:
The only problem with this approach is that you don't always want to
burn a port on the local machine. Let us say that Bob has permission
to deploy multiple Web Services. Alice is willing to cut as many SRV
records as he likes but she doesn't want to give him multiple ports
because she has to reconfigure the firewall for each one.
That is what makes use of .well-known the natural adjunct to SRV.
This is why we in 2005 started working on URI that is the extension of SRV
records in the form of being able to redirect same owner as SRV but to a URI
and not only port and host. That way multiple services can run on same port
(for the named service) but with different path.
Patrik
And that would have been a perfectly reasonable approach if it had
been made instead of SRV. Or the IETF had published the RFC in 2005
rather than 2015.
Unfortunately, virtually every registrar supports SRV now but support
for URI is nowhere near as common. To use URI with my provider, I have
to run my own DNS.
The other problem is that URI supplements rather than replaces SRV.
And discovery is much cleaner if there is one mechanism regardless of
the service in question rather than having to know the mechanism for
the particular service.
As Joe T. pointed out, there are also some services that have prefixed
TXT records used in the discovery process and some that don't as well.