leerho commented on issue #34:
URL: 
https://github.com/apache/datasketches-rust/issues/34#issuecomment-3679833001

   **WRT public/ internal Hash functions**
   In Java we made the hash functions public so that they can easily be 
leveraged across multiple packages.  I don't see any harm in that.
   
   **WRT Compute Seed Hash**
   See java's 
[computeSeedHash(...)](https://github.com/apache/datasketches-java/blob/main/src/main/java/org/apache/datasketches/common/Util.java#L958-L974)
   
   The Theta, Tuple, and CPC sketches allow the user to operate with a 
different seed for the hash function.  This was based on strong requests from 
our users at the time (circa 2013).  The reasons were varied:
   - There was belief among our users that some seeds provided better accuracy 
than others.  (We later proved that this is not true).  They also felt is was 
OK to have a different seed than was already established in other groups.  
(perhaps better security or privacy?). On reflection, this was misguided.  
Later on, when the two groups merged, they were not able to read the other 
group's data.
   
   **WRT Security & Privacy (related)**
   This is a major topic and deserves a lot of discussion.  Nonetheless, keep 
in mind that the sketches in the library do not qualify as cryptographically 
secure and were never intended to be. 
   
   Also, the sketches in the library, by themselves, do not satisfy the 
requirements for "Differential Privacy".  However, under certain conditions and 
in the proper system environment some of these sketches, or perhaps slight 
variations thereof, can be used in a Differentially Private context. (See 
[paper)](https://arxiv.org/pdf/2203.15400) by Charlie Dickens and Justin Thaler.
   
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to