Re: [P1619-2] 1619.2 meeting next week?

[Hal Finney <hal.finney@xxxxxxxxx>]

One of my action items from the plenary last week was to convert the
EME2 test vectors into the same format David used for the XCB vectors.
I have done so, and the new test vectors are attached.

One change is necessary to the definition of the test vector struct in
the current draft, Table 5 in Appendix C.2. The size of the  xcb_key
array needs to be increased from 32 to 64, because EME2 uses larger
keys. Also, I suggest changing the name of this field from xcb_key to
key. That way we can share the struct definition for test vectors for
both XCB and EME2.

My mechanical conversion of the EME2 test vectors retained the
ordering from Brian Gladman, which generally speaking started with the
smaller and simpler cases, and then worked up to the larger and more
complex ones. I believe this produces a more readable and more
conventional presentation the data, to have the simpler cases first.

There is a technicality in that the data structure definition attempts
to create a linked list of all of the test cases, and intrinsically
this list gets linked in the reverse order of how the tests are
presented. So by listing the cases in reverse order, most complex
first, one can easily create a test program which will run them in the
opposite order, simpler cases first. I'm not sure how important this
aspect is. My general feeling is that putting the simpler test cases
first would be preferable as that is what implementors will be used to
seeing.

Hal

Attachment: eme2.5.newformat
Description: Binary data