I think if we can satisfy some programs just by returning false without throwing any exceptions then this probably worth doing. In other words, IMHO we can afford a single violation of our general rule about NotImplementedException if it really improves usability of our implementation. Thoughts?
Regards, 2007/7/11, Tim Ellison <[EMAIL PROTECTED]>:
Zakharov, Vasily M wrote: > I was trying to run jEdit automated tests (see HARMONY-3633), and found > that it uses a special run-time patch to classlib to overcome the > RuntimeException that is always being thrown from > java.awt.Toolkit.getLockingKeyState() method, as it's not implemented. > > As Stepan Mishura suggested in that JIRA, I think that though this > method needs to be implemented properly, a temporal workaround could > make Harmony more compatible for now. So I've filed a HARMONY-4423 issue > for that problem and created a temporal workaround patch that replaces > "throw RuntimeException" with "return false" in that method. > > I think it would be good to commit this patch right now. Any objections? You should modify the method signature with a throws org.apache.harmony.luni.util.NotImplementedException which will serve as a flag that this is a stub, and will be recognized by JAPI tools as such. Or you could just implement it ;-) Regards, Tim
-- Alexei Zakharov, Intel ESSD
