Anonymous user
Talk:Vigenère cipher/Cryptanalysis: Difference between revisions
m
→Faster?
m (→Faster?) |
|||
Line 38:
::::: I went to do some changes of my own yesterday, but was stymied by my not having a C++0x compiler handy enough to stick with it; codepad.org hasn't upgraded yet. I'd make the further change of using std::vector for frequency()'s buffer, but making it a class member variable. I'd have frequency() take a reference as an argument, and pass a reference as its return value. And there were some other dependent changes, but that completely avoids the reallocation issue; frequency()'s buffer gets re-written from scratch each call through, so there's no need to clear and reinitialize separately. Just pass double(26) into its constructor as part of the class's constructor. --[[User:Short Circuit|Michael Mol]] 20:42, 22 June 2011 (UTC)
:::::: Alright. That's a reasonable explanation. (Except for class to struct, which just bugs me.) I think the C++0x features used here are somewhat minor at the moment, but as I mentioned before, the array class might be appropriate. I can try and make the suggested changes. [[User:MagiMaster|MagiMaster]] 21:11, 22 June 2011 (UTC)
:::::: Ok. I made frequency() use a member variable, and changed several vectors to arrays. I also added a typedef since array<pair<char, double>, 26> is kind of long and is used several times. (That also happened to remove most instances of ">>".) The targets and sortedTargets could also be of the FreqArray type for consistency, but they wouldn't be using the extra char info, so I left them as double arrays for simplicity. [[User:MagiMaster|MagiMaster]] 21:35, 22 June 2011 (UTC)
:::: Actually, using the C++0x array class might be appropriate here. It's size is a compile-time constant, but it can still be sorted and everything. (Actually, I'm not sure, but I think you can sort a normal array too, but the array class is safer.) [[User:MagiMaster|MagiMaster]] 19:52, 21 June 2011 (UTC)
| |||