[rfc-i] URL checking

[Paul Hoffman]

At 5:38 AM -0700 10/18/05, Joe Touch wrote:
>Rather than continuing the debate on whether URIs are persistent or how
>persistent, let me ask a different question:
>
>- would a test that says "page found" be sufficient? I.e., anything
>other than a connection error or page-not-found is fine?

That is a good start, yes.

>If so, it's fairly simple to script-extract all URLs and match them,
>provided they're full URLs (http://). We can't necessarily test short
>ones (e.g., listed as 'www.postel.org'), since many are just DNS names
>that need not start with www.

Those aren't URLs anyway, and I don't think we see them often in the 
references sections of RFCs.

>Would that kind of test be sufficient?

It's a reasonable start. A better method would be to have the human 
who is doing the sanity check on the references in the RFC-to-be 
actually look at the page and see if it has the same title as the one 
purported in the reference itself. If it doesn't, this is certainly 
something the RFC Editor might ask the document authors about, just 
as if the document authors had a reference to RFC 3819 and called it 
"IANA Considerations for the Point-to-Point Protocol (PPP)".

--Paul Hoffman, Director
--VPN Consortium