[PEPr] Comment on Encryption::Crypt_DiffieHellman

Philippe Jausions (http://pear.php.net/user/jausions) has commented on the proposal for Encryption::Crypt_DiffieHellman. Comment: Couple of nitpicking and PEAR CS comments: - For speed use ctype_digit() instead of preg_match(/^\d$/...) - Constructor don't actually "return" so remove the "@return void" in docblock - Use PEAR::loadExtension() instead of loaded_extension(), dl() and so on... - For Crypt_DiffieHellman_Math_BigInteger add an actual factory() method to allow overriding the automatic detection of which backend to use (sometimes needed if an extension gets buggy.) - the "class Crypt_DiffieHellman_Math extends Crypt_DiffieHellman_Math_BigInteger" is a bit an unconventional class name order. We can see the negaive side effect of this in the "if instanceof Crypt_<snip>_gmp" tests. Organize the backend classes so they are self-contained (i.e. not try to guess which type they are at runtime; if you already loaded a backend no need to recheck whether a feature is available or not.) - Always put spaces between "if" and opening ( - same for "elseif" & co. - Use ++$i instead of $i++ whenever possible ("for" loops) Proposal information: http://pear.php.net/pepr/pepr-proposal-show.php?id=496 -- Sent by PEPr, the automatic proposal system at http://pear.php.net