On 02/19/2013 01:25 PM, monarch_dodra wrote:
If it keeps a copy, how it is passed (value or ref) actually becomes irrelevant.
Yes, I didn't explain myself well there. The real problem is that copying really duplicates the RNG rather than just copying a reference to the RNG engine.
This is also a strong argument for the reference semantic PRNG approach.
Yes. Are you having any luck in discussions on how to map out a random2 where this will be implemented? I remember your original code was rejected as a plain update to Phobos because of fears over code breakage (though TBH I think this is a case where if you break someone's code you're doing them a favour, as they were probably doing something unwise).
