On Wed, 15 Sep 2021 11:12:06 -0700, Charles Mills wrote:

>I am not sure I really understand but wouldn't the web site be computing the 
>hash on the ASCII representation of ABCabcAB12345678 while the mainframe would 
>be computing the hash on the EBCDIC representation of ABCabcAB12345678? Those 
>would be different hash values. The hash computation is on the "bits" of the 
>key not the "letters" as you see them.
> 
What does a "HASH" have to do with a "seed"?  Isn't a hash algorithm such as 
SHA-1
deterministic, repeatable, so that (e.g.) CSNBOWH will produce the same result 
for
a given message every time?  (I verified the availability of CSNBOWH by passing 
it
"Hello, World!' and verifying the output.)

Does ICSF's random number generator support seeding?  This seems undesirable for
cryptography.  A colleague once suggested that for Monte Carlo numerical 
analysis
it's valuable to have a repeatable random number stream while varying other 
parameters
of a model.  But that depends on the algorithm's not relying on 
acceptance-rejection.

Is ICSF's PRNG truly "psudo"?  I understand that on recent z hardware folds in 
true
randomness when available (comparing the output of two unstable oscillators.)

And is ICSF's RNG local to a job, not susceptible to any other job's stealing 
part of
the stream, making the behavior unpredictable even for a known initial state?

And the suggestion of translating the message to hex and hashing the hex stream
can fail depending on whether the hex is represented in ASCII or EBCDIC.

And newline conversion: <CR>, <LF>, or <CRLF>?

And I've suffered from transfer agents that pad or strip trailing blanks or 
discard
empty lines.

-- gil

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to