If a client violates these restrictions, I prefer that their code fail
at compile time instead of run time. Changing a few key types from
`Class` to `Class<? extends AbstractAndroidModel>` gives us precisely
the static enforcement we need and lets us remove an
`AbstractAndroidModel.class.isAssignableFrom` run-time check.
However, this does change the public API of `AndroidEntryPointManager`
in two ways. The `getModelBehavior` and `setModelBehavior` methods now
respectively accept and return `Class<? extends AbstractAndroidModel>`
instead of `Class`. Is tightening up a public API in this manner
considered OK?
This fixes one Eclipse "no valid properties files exist in the
localization directory specified" warning.
Ordinarily I would also change this project's configuration to treat
this warning as an error in the future. That's a good way to
discourage regressions. Unfortunately in this particular case I
cannot find a setting that has the desired effect, even after hunting
around in Eclipse PDE sources. It seemed that setting
"compilers.p.unknown-resource=0" in
"com.ibm.wala.util/.settings/org.eclipse.pde.prefs" should do the
trick, but it does not. I don't know why.
This resolves one Eclipse "'...' build entry is missing" warning.
Also update the project configuration to treat this warning as an
error. This should discourage commits that create new instances of
this sort of problem in the future.
This resolves one "'...' is not a source folder" Eclipse warning.
Also update the project configuration to treat this warning as an
error. This should discourage commits that create new instances of
this sort of problem in the future.
Other subdirectories' "build.properties" generally seem to include this
already, so who am I to argue?
This resolves one "An entry for META-INF/ is required in bin.includes"
Eclipse warning.
Also update the project configuration to treat this warning as an
error. This should discourage commits that create new instances of
this sort of problem in the future.
In general, the WALA code base is not really ready for nullness
checking. It would be nice if we got there some day, but I'm not
planning to take that on now or any time soon. Until then, it's not
useful to warn about missing @NonNullByDefault declarations on WALA
packages.
See also older commit 7b6811b.
A subclass of TabulationSolver can now override the methods
newNormalExplodedEdge(), newCallExplodedEdge(), and
newReturnExplodedEdge() to take some action whenever (logically)
some edge in the exploded supergraph is "discovered" during
tabulation.
These methods were constructing an IR based on some default
AnalysisOptions, which may not match the options used when constructing
the underlying CallGraph. This mismatch can lead to bad bugs.
Instead of these methods, analyses should get IR directory from the
CGNodes via CGNode.getIR().
Ideally we would fix the methods and not change the interface, but
that would require knowing the right AnalysisOptions, which itself
would necessitate an interface change.
Ben Liblit
Manu Sridharan
Julian Dolby
Brian Pfretzschner
Michael Herzberg
Manu Sridharan
Julian Dolby
Julian Dolby