Originally Posted by crazyNut
Is it something like in case of i++
the compiler first evaluates the expression
j = i++*i++
and then increments i 2 times in a row
and opposite for ++i
means there it first increments i 2 times in a row and then evaluates the exp k = ++i*++i
No. Look again at the definition of a sequence point. It is a point after which all side effects are guaranteed to have been completed. It is *NOT* a point before which all side effects are guaranteed NOT to have been completed. Before a sequence point, some side effects may or may not have been completed and you have no way of knowing which have and which haven't.
So i=3; j=i++ * i++;
could mean j=i(3, post incremented to 4) * i(4, post incremented to 5)=12,
OR it could mean j=i*i(post incremented twice to 5)=9.
The result is compiler dependent. In Visual Studio j=i++*i++; is equivalent to j=i*i; i++; i++; but you cannot guarantee this for all compilers (or even all versions of Visual Studio).
The end result of this is that if you only need a variable once in an expression, it's safe to use modifiers, but if you need it more than once, use it WITHOUT pre/post increment/decrement within the expression, THEN modify it.
By the way, this is a very frequently asked question. Check out stuff like the C++ FAQ before asking anything else. It will almost certainly answer your next 20 questions, plus there will be a load of other interesting stuff in there, so it's well worth a read.