On 24.11.15 10:09 , Chetan Mehrotra wrote:
On Tue, Nov 24, 2015 at 2:35 PM, Michael Dürig <[email protected]> wrote:
I'd rather remove the method and break backward compatibility at compile
time instead of runtime.
Can do that but then would need to bump the major version of the stats
package. Would that be fine?
I'd say yes, even when leaving the method in. While the signature stays
backward compatible, the semantic is quite different.
I missed marking this one as deprecated so would that needs to be done.
@deprecate would be an alternative. But then I'd just log a warning
should the method be called instead of throwing.
Michael
