Unit test for Mersenne primes #60
Conversation
|
Unit testing now takes much, much longer; it increased from ± 2 seconds to ± 24 seconds on my machine, for a single Tox environment. Please reduce the size of the numbers, and the number of numbers, to the minimum that's needed to perform the tests. |
|
Could you please test execution time again? I just removed testing for product of primes, and Travis times have returned to normal. |
|
On my little Linux server (which I use for development when my desktop is booted to Windows and I don't want to reboot) this patch still increases unittest time on a single Tox environment (py34) from 13 to 21 seconds. Are all these test cases really necessary? |
|
I chose these test cases for testing primes up to the 4096-bit range! Reducing it to 2048-bit range saves 4 cases. Actually, we could remove last test case and still remain in 4096-bit range (because of Testing times on my hardware (
IMHO, correct prime generation and validation is essential in RSA, so I think it worths the testing overhead, especially having Travis to handle unittest. |
|
Changes Unknown when pulling 20c1693 on adamantike:test-mersenne-primes into * on sybrenstuvel:master*. |
|
Changes Unknown when pulling 0b8c914 on adamantike:test-mersenne-primes into * on sybrenstuvel:master*. |
|
Yes, let's go for "Proposed PR 4096-bit", as I fully agree we should test 4096 bits! |
|
Merged in a72efaa |
Checks Mersenne primes' primality. Tests bigger numbers, so it also checks Miller-Rabin's correct result for bigger bitsizes (and less rounds after merging #58).
It could help with #54, by giving more testcases and reducing the random factor affecting test coverage.9a9e08c fixed this.