Martin Kellogg comments

Results 61 comments of


                                            Martin Kellogg

Cannot implement checkerframework by using task avoidance

Sorry you're running into some build performance degradation while using the plugin. > Is there any future plans to make the plugin task avoidance compliant? While I would love to...

Inconsistent inferred annotated file generation during WPI iterations causing non-termination

Here is a minimized test case: ``` // test case for https://github.com/typetools/checker-framework/issues/6570 // A class is technically permitted to end with a semicolon (though it doesn't do // anything). public...

Inconsistent inferred annotated file generation during WPI iterations causing non-termination

> @kelloggm I ran on the original project url7a69e77641_benmouh_MySgbd_tgz-pJ8-Demo_CheckRowNumJ8 with the fix from https://github.com/typetools/checker-framework/pull/6588, but still getting the non-termination. Seems like the compier erorr still exists: @iamsanjaymalakar I can't reproduce...

Non-termination of WPI on NJR Benchmarks for Nullness Checker Annotations

I still don't know what's causing this, but I'm writing down some notes before I give up on this for the moment in case someone else has some insight here...

Should the default stub of JUnit asserts like assertNotNull really be NotNull?

I find @agentgt's argument that `assertNonNull` _in JUnit specifically_ is not for defensive programming to be convincing. Personally, I would use `assertNonNull` in a test if and only if I:...

Don't output top annotations to .ajava files

While this change looks sensible, I can't see a difference between the output of this PR and `master` on either `./gradlew ainferResourceLeakAjavaTest` or `./gradlew ainferIndexAjavaTest`, both of which seem to...

Resource Leak Checker: false positives when a resource is allocated and closed in the same expression

If we have a `ResourceLeakStore` or create one, we might be able to override `CFAbstractStore#insertOrRefine` (which the postcondition code calls indirectly) so that it checks for temporary variables and also...

Resource Leak Checker: false positives when a resource is allocated and closed in the same expression

With some more investigation, I'm not sure that our current implementation can handle this case without either a major refactoring, a new CFG API, or both. In particular, the core...

Resource Leak Checker: false positives when a resource is allocated and closed in the same expression

> At this step the logic must have a handle on the Tree or Node for the method invocation, correct? Is there hope of doing something like making a copy...

Resource Leak Checker: false positives when a resource is allocated and closed in the same expression

Sorry to come back to this much later. > I guess what I don't understand from suggestion (1) is why it would take a JavaExpression as an input? Shouldn't it...