Bad Code (was: Re: [PHP] ++ BUG??)

On Sat, 25 Nov 2000, Jason wrote: > I tested it in C and you're right, it is indeed well defined. I was slightly > off in my interpretation of when, exactly, the ++ operators increment the > variable. I thought it was after the use of the variable, not after the entire > statement is finished. So I'll concede that it's a bug if they're going for > C-style behavior. > > All I was trying to say, though, was that it's bad code and the bug is easily > avoided :) > I really do hate to continue this thread, but I don't necessarily agree with the "bad code" part. Compilers and interpreters can treat them very differently. Note: the following examples are on CPUs which do not have a pure load/store architecture and assume that there is no other code. i=(i+1)%2: 1) load memory location I into register X 2) add 1 to register X 3) take modulus 2 of register X 4) store register X into memory location I i=(i++)%2: 1) take modulus 2 of memory location I 2) increment memory location I i=(++i)%2: 1) increment memory location I 2) take modulus 2 of memory location I The memory location operations would have to be quite expensive to make the first example faster. -- Ignacio Vazquez-Abrams <ignacio@openservices.net>