>We can't just replace all the existing interfaces with 64-bit versions, of course - we need to continue suporting the existing interfaces for existing code. T

>We can't just replace all the existing interfaces with 64-bit versions, of course - we need to continue suporting the existing interfaces for existing code. The backwards compatibility problem in a nutshell. You can put off breaking stuff today and kick the can down the road some more, but where does it end? January 19, 2038? I think it boils down to a philosophical difference, not really a practical difference, despite the inevitable arguments that "we can't just go around breaking all the shit that's already deployed." Even though the pain would be greater short-term, the benefits of breaking stuff today almost always outweighs the hazards of putting it off indefinitely, since there's almost never any motivation for going back and removing the workarounds once a stable solution is implemented. As evidence, OpenSSL is Exhibit A.