On Fri, 1 Jul 2022 23:38:51 GMT, Bradford Wetmore <wetm...@openjdk.org> wrote:
>> JDK-8170762 - update Javadoc to indicate ISO10126Padding pads with random >> bytes, though the spec does not require random bytes > > Just a comment, I believe both approaches are ok. The ISO 10126 spec says: > >> When padding with octets...the first subfield (pad fill) shall consist >> of 0 to 7 octets with arbitrary contents. > > Adam's original comment was that because the spec says "arbitrary data," we > could potentially just optimize the code a small bit by not having to fill > with random data, and rather just leave as an uninitialized field. But using > random data to get some arbitrary data it is possibly a little safer since it > will perturb the output a little vs. using an all zero padding. > > So, fine to leave has been fixed. However, I do have further comments on the > doc bug. Yes, @bradfordwetmore. Agreed. Both approaches are valid; hence just taking the "document the choice" approach. ------------- PR: https://git.openjdk.org/jdk/pull/9350
