[Libre-soc-bugs] [Bug 1159] poly1305 unit tests

[bugzilla-daemon at libre-soc.org]

https://bugs.libre-soc.org/show_bug.cgi?id=1159

--- Comment #6 from Luke Kenneth Casson Leighton <lkcl at lkcl.net> ---
(In reply to Sadoon Albader from comment #5)

> When I said "added" what I really meant was I was messing around with the
> code and *replaced* the original values :P

yyeah i did that too :)

ah this is a... it just occurred to me, of course the tests will be
very specific to the algorithm internals, it will be quite fragile.
as in: if you don't *exactly* faithfully replicate the poly1305-donna
algorithm (even from the original c) then you could end up missing
something important, as it is so rare an occurrence.

hm i tell you what, i will create a python wrapper around poly1305-donna.h
so that *that* can also be added to the test matrix. i won't do the
unit tests themselves (other than a real basic one), just to get it
not to segfault ok?

then you can run huge numbers of tests very quickly, on the c-based
module *and* the poly1305-donna.py conversion, and at least have a
degree of confidence that i did the conversion correctly!

we *must* make sure that there is a "chain of confidence", absolutely
no more than one "unknown" at any one time, no matter how small and
"seemingly ridiculous" ok?  just like in that word game.

-- 
You are receiving this mail because:
You are on the CC list for the bug.