On Thu, 7 Mar 2019, Ray Bellis wrote:
On 07/03/2019 16:57, Petr Špaček wrote:
Like this one?
https://dougseven.com/2014/04/17/knightmare-a-devops-cautionary-tale/
Have you perhaps got anything constructive to contribute to the discussion
instead of pure hyperbole?
It is not hyperbole. It is an example of what can happen when people
overload options. Your proposal is a bad overloading option.
You are suggesting to introduce an option code point to convey blobs in
DNS. So different parties can send and receive blobs. You think or hope
that these parties will interpret this blob the same. But you have no
guarantee this is true.
If you have a specific use case, get a code point for that specific use
case. Than you know for sure what the blob means and that it will be
interpreted by all parties in the same standard RFC way.
If your use case is too private/secret or non-standard, then use a
code point from the "Reserved for Local/Experimental Use" range. Other
implementations then do not need to worry about misinterpreting the
meaning of the blob if more than one common use case started happening
on this code point, since they can ignore private use code points. If
your use case is experimental, go experiment and come back to us for a
real code point once the experiment is a success.
This draft is not a good idea.
Paul
_______________________________________________
DNSOP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dnsop
