Ticket #194 (closed task: fixed)
Analyzing role-method visibility should be separated out
| Reported by: | stephan | Owned by: | stephan |
|---|---|---|---|
| Priority: | minor | Milestone: | OTDT_1.2.6 |
| Component: | compiler | Version: | 1.2.5 |
| Keywords: | Cc: |
Description
During recent changes we repeatable broke the analysis whether a role method is visible in a given context. This shows that the current implementation is fragile, because it is tangled with the original implementation of std. Java visibility rules.
For role fields we already use a separate method Protections.canFieldBeSeenBy. This method should be generalized to cover methods, too.
Here is an almost random list of test cases that recently should some effect while trying out different implementations:
- 0.a.8-otjld-role-field-accessed-by-team-6f
- 0.a.8-otjld-role-field-accessed-by-team-7
- 1.1.2-otjld-team-method-invocation-4a
- 1.1.2-otjld-team-method-invocation-5a
- 1.5.18-otjld-implicitly-inherit-role-file-1 and -1f
- 2.3.14-otjld-inaccessible-tsuper-constructor-2
- 2.4.1-otjld-visible-base-class-18e
- B.1.1-otjld-sh-70
JUnit tests:
- ModifierCorrectionsQuickFixTest (OT-variant)
Furthermore, compiling the OTDT's OT-plugins also gives a pretty good test case ;-)
Change History
Note: See
TracTickets for help on using
tickets.
all news
RSS feed