Andrei Robachevsky <[email protected]> wrote:
>I was reading some pretty recent drafts and noticed that authors
>continue using IPv4 in their examples. This is probably more convenient,
>but is not really forward thinking. Also, the prevalence of IPv6
>examples will send a strong message that IPv4 is essentially a legacy
>protocol.

I agree that examples in RFCs should be required to use IPv6 examples (and 
optionally also include IPv4 examples). If IPv6 is really the successor 
specification, then continuing to publish RFCs that do not provide examples 
using IPv6 is actually a disservice to the community. It seems like a sensible 
step.

One small quibble, though. I'm not sure that the section on IPv6 addresses with 
ports accurately reflects RFC5952 and RFC3986. The bracket notation MUST be 
used if it represents a URI literal. In other instances, it SHOULD be used (as 
the draft indicates), but other formats are acceptable. People may use this 
document when developing examples without referring to the reference, so the 
explicit requirement should be noted. (And I think the convention is that the 
'SHOULD' should be capitalized.) So perhaps something like the following text.

"IPv6 addresses including a port number SHOULD use the bracket notation style 
[RFC5952]. URI examples containing an IPv6 literal MUST use enclose it in 
brackets [RFC3986]."

And then provide the example. I'm sure it could be phrased better than the 
above. It's early yet. But the idea is that someone reading this document 
should know there are situations where the bracket notation is actually 
required.

Scott 
_______________________________________________
sunset4 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/sunset4

Reply via email to