On 11/15/2012 5:01 PM, David DeHaven wrote:
L428-430: is this fallback needed? Would it be better
if LauncherHelper.getApplicationClass() always returns
a non-null class if the mainClass has been loaded successfully.
Looks like this is the case in your implementation.
Good point, the helper would have aborted by that point. How about I change to
NULL_CHECK(appClass) just for safety's sake?
Sounds good to me.
The change looks okay to me and I can't spot anything wrong there.
L496-517 somewhat duplicates the logic added for FX in the
getMainClassFromJar method. Have you considered some refactoring
work you could do to simplify the fix since I think once you get
the classname of the entry point (either from a JAR or command-line
and with and without the static void main method), the logic is
essentially the same. To elaborate, I see that FXHelper.launchName
L707-725 is needed mainly to give a useful error message. When
you find the classname of the entry point, perhaps you can load
the class and catch any linkage error and determine if it's caused
by the absence of FX runtime and output an appropriate error.
If the main class is successfully loaded, then proceed with
L496-517 (or something like that). Just an idea you can explore.
Yes, I do feel that especially in the -jar case there is some repetition. The
trouble is the ambiguity of ClassNotFoundException.
I'll poke at it and see what I can come up with.
That's good.
Mandy
