Re: [dnsext] SRV and wildcard CNAME

[Phillip Hallam-Baker <[e-mail]>]

2011/2/20 Masataka Ohta <[e-mail]>

> Phillip Hallam-Baker wrote:
>
> >>> When the protocols used at the domain are http and https only,
> >>> nothing break.
>
> > Mark is quite correct here.
>
> Neither your example "_null._random.example.com" nor Mark's
> example "_foo._tcp.bar.example.com" is applicable to the case
> of "protocols used at the domain are http and https only",
> because you are assuming protocols "null" or "foo" is used
> at the domain.
>
> > In other words the domain is advertising
> > services it does not support, this is a failure for a discovery
> mechanism.
>
> That's fine.
>
> Lazy users do not mind if they advertise non existent services.
>


But competent people mind a very great deal. If the client is attempting to
discover which GPS protocols are supported at a site and choose the best,
the wildcard scheme fails because it will advertise everything as being on
offer.

Now I do have some experience in Web Services that is rather relevant here.


Like many protocol design issues, these are very easy if you only deal with
some of the use cases. They become much harder when you actually take into
account all the constraints.


-- 
Website: http://hallambaker.com/

_______________________________________________
dnsext mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/dnsext