persistence of user consent

Cantor, Scott cantor.2 at osu.edu
Thu Feb 19 10:56:56 EST 2015


On 2/19/15, 3:53 PM, "David Langenberg" <davel at uchicago.edu> wrote:



>So, it sounds like a good "sane" default would be keeping 7-14 keys or 
>slowing down the rotation?  Taking a 3-day weekend or something like 
>Thanksgiving shouldn't cause the user to have to re-approve?

Well, bear in mind that right now the cookie thing is more a toy for 
experimenting. The size just turned out to be so big that we can't keep 
enough consent records anyway. The local storage thing is what we need to 
get built and are working on.

But sure, as a basic matter, you don't want it rolling over that badly if 
it's not a space issue.

>  So which is the better approach, slowing down rotation or keeping a ton 
>of keys loaded?

A mix probably. The keys don't really "load", it just manually loads a 
non-default key if it sees a need, so it adds a slight overhead of I/O to 
load the key instead of just using the one default key it always has on 
hand. I doubt it's noticeable. So the key history can probably be somewhat 
larger, but I don't know the details of the keystore format or what limits 
there are.

-- Scott



More information about the users mailing list