On Mon, Oct 21, 2024 at 9:58 AM John R. Levine <[email protected]> wrote:

> On Mon, 21 Oct 2024, Brotman, Alex wrote:
> > FYI, the ticket has been closed without merging, and the existing
> regular expressions will remain in the current document.
>
> We all agreed that the current regular expressions for IP addresses are
> wrong and don't match what people send.  The question was whether to try
> to replace them with a long RE that exactly matches valid addresses, or a
> short RE that just lists the valid characters.  In the first case, it's
> hard to get long RE's correct, and several people said we're not in the IP
> definition business.  In the latter case, the RE will let some invalid
> addresses through, presumably to be rejected later by address parsers.
>
> It seems to me that address parsers are already going to reject addresses
> that are semantically invalid, e.g., 127.1.2.3 or ::123, so also allowing
> some syntactically invalid ones won't matter.
>
> I suggest [0-9a-f:.]+ which lists the valid characters and is very short.
>

Does this have to be done with a regexp?  Is it enough to say
"syntactically valid per RFC xxxx" (and cite something appropriate)?

I ask because after poking around a bit, I found that there is (amazingly)
no standard for what constitutes a syntactically valid IPv4 address.

-MSK
_______________________________________________
dmarc mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to