On Tue, 22 Nov 2022 22:21:11 GMT, Kevin Rushforth <k...@openjdk.org> wrote:
>> good catch - I think the change is not equivalent. It should be either >> >> 4L * physicalWidth * physicalHeight; >> >> >> or >> >> >> ((long)physicalWidth) * physicalHeight * 4L; >> >> >> the proposed change might result in overflow bc the first intermediate >> result is likely to be int. > > Actually, I think the new code correct, though less clear, since `(long)` is > applied to `physicalWidth` before the multiplication. The fact that we had to > scratch our head and discuss it highlights the point I was trying to make. I > do think this is one place where reverting the code back would be best. Then > it's obvious that all of the intermediate calculations are in long. The cast has precedence and Java will not downcast mid calculation and lose precision, so all examples here are correct. ------------- PR: https://git.openjdk.org/jfx/pull/960
