Fix #38: improve dfbuyer random algorithm #39
Conversation
trentmc
changed the title
pdr_backend/dfbuyer/main.py
Outdated
| return [k] | ||
| num = random.randint(1, k) | ||
| return [num] + numbers_with_sum(n - 1, k - num) | ||
| numbers = [random.randint(1, k - n + 1) for _ in range(n - 1)] |
There was a problem hiding this comment.
You should use random sampling without replacement.
And to do this, use random.sample().
Examples of usage: https://pynative.com/python-random-sample/
There was a problem hiding this comment.
Randint is not the way either. You shouldn't have to write a loop. Just use random.sample()
There was a problem hiding this comment.
Just to clarify, in this line I'd like to generate n-1 numbers that are between 1 and k - n + 1. I don't understand how random.sample is going to work in this case. Could you elaborate on that please?
There was a problem hiding this comment.
Something like
random.sample(range(1, k - n + 1), n - 1)It might be that the first argument needs to be 1 larger or smaller. And same for second argument. I defer to you to test.
There was a problem hiding this comment.
I updated it to use random.sample, could you take a look?
|
Thanks @trentmc for the review! I've added docstring, inline comments, tests and type declarations. Also reverted the tx arguments changes. |
pdr_backend/dfbuyer/main.py
Outdated
| return [k] | ||
| num = random.randint(1, k) | ||
| return [num] + numbers_with_sum(n - 1, k - num) | ||
| numbers = [random.randint(1, k - n + 1) for _ in range(n - 1)] |
There was a problem hiding this comment.
Randint is not the way either. You shouldn't have to write a loop. Just use random.sample()
…er-random-algorithm
trizin
deleted the
issue38-fix-encrypted-tx-arguments-and-dfbuyer-random-algorithm
branch
Fixes #38
Changes proposed in this PR: