On 05/04/2009, sebb <[email protected]> wrote: > On 05/04/2009, Phil Steitz <[email protected]> wrote: > > sebb wrote: > > > > > On 05/04/2009, Phil Steitz <[email protected]> wrote: > > > > > > > > > > [email protected] wrote: > > > > > > > > We need to be careful with removing all of these casts. They were > > inserted > > > > earlier to eliminate some errors resulting from int/int computations in > > > > larger expressions resulting in incorrect values. Could be later JDKs > > > > handle all of this seamlessly, but I would prefer to be safe rather > than > > > > sorry here and leave the explicit casts from int/long to double alone. > > > > > > > > > > > > > > I've just checked, and there are still some (double) casts left in the > > > code that Eclipse does not complain about, so I assume that Eclipse > > > distinguishes between casts that Java does anyway and casts that > > > affect the resulting code. > > > > > > > > I would rather not leave this decision to Eclipse. I will review and > > selectively revert. > > > OK, thanks, I will go through the commits as well. >
I've looked at both the java and test tree commits, and cannot find any incorrect cast removals. I did find one place in the test classes where Eclipse removed one cast and not the other: old: expected[i] = (double)smallSampleSize*100/(double)16; new: expected[i] = (double)smallSampleSize*100/16; which is to be expected as smallSampleSize is long. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
