Why am I writing a Swing app in 2007? That's what I'm asking myself. It's a conspiracy, I tell you, involving the dreaded secret society
the FuglyPLAF Posse. Anyway, onward to more pissing and moaning...
Take the humble
MouseEvent (alias
java.awt.event.MouseEvent). The integer constants on
MouseEvent include:
- BUTTON1
- BUTTON1_MASK
- BUTTON1_DOWN_MASK
BUTTON1 is for use with the
getButton() method, which (you'd think) might return the button being pressed when the event was generated. It actually does that on the Apple Java 5
JVM. On the Sun
JVMs, apparently,
getButton() returns the button "whose state has changed". If you're holding down a mouse button and dragging,
getButton() lamely returns zero. Thanks for nothing.
BUTTON1_MASK is for use with getModifiers(). According to the JavaDocs, it's now recommended to use BUTTON1_DOWN_MASK which goes with getModifiersEx().
And, before you can say "Type-safe Enumeration", stop to ponder how many constants are needed to represent mouse buttons. How about none? Wouldn't a plain integer do just fine? How about a boolean function: event.buttonDown(int button)? OK, maybe it's useful to expose the bitmasking of the modifiers to the user, but the resulting mess argues that it would have been more useful not to.
Three different constants for a concept with all the complexity of mouse buttons? I sincerely hope someone was taken out and given an over-the-head turbo-wedgie for that one.