Re: late dtls-id request

Hi,

>>But, I assume we can make that assumption also with 255.
>
>For that, 20 would be plenty.
>
>Note that you only require 32 bits of randomness (which fits in the
>minimum of 6 characters), which is not quite enough to provide
>global-scale uniqueness.  20 pretty much is going to do that (assuming
>6 bits per character, which base64 would allow).
>
>FWIW, I would be happier if you moved the 32 bit requirement to 120
>and raised the minimum to 20, but that's a much bigger change.

If people think such change would be useful I personally have no problem
implementing it.

But, I¹d like to hear what Roman & others think.

Regards,

Christer