Re: [TLS] New drafts: adding input to the TLS master secret

[James Manger <[e-mail]>]

> I propose a design rule:   A PRNG should never be used to obtain a unique value.
> Instead a counter or a date should be used.



I propose a different design rule: A PRNG should always be preferred over a counter when a collision-free value is required.

Counters need their state to be retained to ensure uniqueness and maintaining state is very often a challenge for computer programs (across outages, across server farms, across saving VM images and restarting duplicates, across hot backups, across correcting the clock...)

For a PRNG output to be unique you just choose an appropriate output size.


Revisit the rule iff a new crypto result emerges undermining all known PRNG designs. 

--
James Manger