Talk:RSA code: Difference between revisions
→Solutions are starting to diverge: there's time...
(→Solutions are starting to diverge: concencus?) |
(→Solutions are starting to diverge: there's time...) |
||
Line 17:
--[[User:Dgamey|Dgamey]] 02:49, 27 April 2011 (UTC)
: Deliberately so. :) With the task in draft status I thought I would mix things up with some different ideas, and thought I would start with some code as an illustration. So in no particular order here, I was thinking a number of things.
:* The task has potential to get uncomfortably large and multi-faceted. I wanted to see if I could write something compact. (Well, as Go code goes. I know Go tends to run long in LOC.)
:* I think big numbers are an important part of RSA, but when I tried RSA-129, it looked a little messy, requiring either literals split across multiple lines or else lines much longer than a punch card.
:* Real encryption involves a number of details in the blocking. There's not only the splitting, but there's usually a few steps of padding and hashing too. This seems a distraction from the interesting number-theoretic algorithm. Still, the task as worded wanted text-to-text encryption to decryption, so I tried a minimal way of doing that, just laying out bytes and not even mapping the character set. (It could even be done more directly but I wanted to make my technique as obvious as possible.)
:* Also I don't bother converting the encrypted number to something transmittable like a printable string or byte array. Is it enough to show the encrypted number? Should the task require sequencing blocks into a byte stream?
:* Encryption and decryption are symmetric--it's the exact same function. I thought I would write a common function that I would call for both purposes. Then I thought, that's silly, why repeat? Maybe the task should just give an encrypted message and ask to decode and print it. Then I realized there's nothing to write--the exact function, the exp mod thing, is in the Go library. But then, small values of E are the norm. An encryption function might be written to run faster by taking a small integer type as E rather than a big integer type. But then, RC tasks aren't generally about optimization...
:* Subtasks may well be in order. I think key generation would be a good subtask. I found it was non-trivial to generate one just the way I wanted it. (I'm still not sure it's right, btw.) Blocking might be another good subtask. Then the task writer might feel free to require several steps of padding and hashing and so on. The exp mod function could be a subtask, but a) it's pretty simple, and b) I know I've written it for at least one RC task already, Factors of a Mersenne, and I'm not sure, I kind of remember it somewhere else too. I don't know, maybe that's more of a reason for it to be a separate task, so it could be referenced from these other places as well.
:* I barely know what I'm talking about. What is this PKCS#1 thing? Do we want to go there?
:* These are just ideas to toss around. I'll happily change my code to match whatever the task settles on.
: —[[User:Sonia|Sonia]] 05:22, 27 April 2011 (UTC)
===Split the task?===
| |||