Message4600
I've committed a fix for this in r6266. Rather than trying to build up a set of method signatures just for the
respectJavaAccessibility portion of the code, the commit grabs all the methods, stuffs them into a list, and lets
PyReflectedFunction sort out the argument conflicts.
Both the patch and the committed version will lead to a circumstance where if a superclass declares a private or
package protected method with more restrictive types than a subclass, the subclass method won't be callable from Jython
with those types. The alternative is doing more type analysis for respectJavaAccessibility and not including
superclass methods if they're more specific than subclass methods, but that means the superclass method isn't directly
accessible on the subclass. I believe this is how respectJavaAccessibility has always worked, so I'm going this way to
keep from rocking the boat.
In any case, I advise getting away from respectJavaAccessibility if possible. Its effects on code can be non-obvious,
it's far less solid than the regular Java integration, and it may go away in a future version. |
|
Date |
User |
Action |
Args |
2009-04-27 05:07:59 | cgroves | set | messageid: <1240808879.99.0.0483035579571.issue1285@psf.upfronthosting.co.za> |
2009-04-27 05:07:59 | cgroves | set | recipients:
+ cgroves, fwierzbicki, zyasoft, marcdownie, kfitch42 |
2009-04-27 05:07:57 | cgroves | link | issue1285 messages |
2009-04-27 05:07:53 | cgroves | create | |
|