David M. Lloyd
David M. Lloyd
@HannuLahtinen commented https://github.com/HannuLahtinen/jboss-modules/commit/246204425eb6696dca05471333e20922bc3d9da7 In our tests this fix stopped the blooming effect described in the bug. We tested with the spring-demo, I created to reproduce the issue for Arcadiy's tests....
@arcivanov commented [Hannu Lahtinen](https://issues.redhat.com/secure/ViewProfile.jspa?name=hannu1) I'm not sure the transformation in the PR is equivalent. For any module dependency a->[x->[m,n], y->[m,n]], the addExportedPaths will never descend into dependency paths [m,n] for...
@HannuLahtinen commented You are correct that the addExportedPaths() will never descend into dependency paths [m,n] for "y" -root. This was intentional. Does it cause a problem? I think the only...
@HannuLahtinen commented Here is where the attempted fix breaks the behaviour: ``` Module.java from lines 1335 and 1382: if (accept) for (Object filter : filterStack.getRawArray()) { if (filter != null...
@dmlloyd commented The reason that a visited set is not enough is that to truly track it correctly, you'd need to track not only the visited dependency but also the...
@arcivanov commented [Hannu Lahtinen](https://issues.redhat.com/secure/ViewProfile.jspa?name=hannu1) The addExportedPath assembles Import/Export Path Filters, Class Filters and Resource Filters. Failing to descend into the dependencies M, N of Y may result in improper loader...
Maybe we could set the defaults at a `ClassOutput` level like this: ```java co.defaultAccess(ALL_TYPES, PUBLIC); // all types would be public by default co.defaultAccess(CLASSES, PACKAGE_PRIVATE); // ...except classes which would...
This should wait until we have an answer for #304 and for #335.
Right, we'd want to have a `nonFinal()` in this case I guess?
I don't really like the configuration. It seems over the top. I would instead rather be able to configure a handler for logstash or splunk, by name, and point it...