Re: uri handling of hosts is too restrictive

Hello  Roy, others,

At 15:41 04/02/10 -0800, Roy T. Fielding wrote:

>On Friday, February 6, 2004, at 12:11  PM, Stephen Pollei wrote:
>>Also at issue is that the uri
>>spec SHOULD be neutral as to what particular host name lookup
>>technologies and restrictions a particular uri resolution implementation
>>may choose to use. I might use DNS, Host Tables, yp/nis/nis+, etc.
>
>Done.  This was implemented as part of removing hostname productions
>in favor of general registered names.

The restriction of hostnames to DNS was discussed and agreed on
at the San Francisco IETF based on interactions with IRIs.

The argument was that conversion from IRIs to URIs (defined in the
IRI spec) should take care of conversion from non-ASCII characters
to punycode in the DNS part. But this is only possible in a scheme-
independent way if it's possible to know what is a domain name
and what not.

The alternative, which Roy has chosen now, and which I think is
architecturally cleaner, is to have IRI to URI conversion use
%-escaping. The main impact of this is that proxies may receive
%-escaped domain names, and may have to convert them to punycode
if they want to support IRIs. [for Roy's benefit: I volunteer
to work on a module that does this for Apache]

I can go ahead and change the IRI spec to do things this way,
but before I do this, I'd prefer to get a confirmation that
this is the last change on this topic; flip-flopping on an
important part of a spec is no fun.


Regards,    Martin.

Received on Wednesday, 11 February 2004 18:49:57 UTC