Created a better thread safety test and fixed a couple thread safety issues

[ivakegg]

No description provided.

[garydgregory]

-1 failed build and:
Quick note but the PR would need a detailed review from more folks: I am really skeptical here, VFS was not designed for thread safety; hacking some code here and there is more likely to make the code base much harder to understand and maintain. In the past, we've just told users they have to handle thread safety in their applications. Slapping "synchronized" here and there is likely to create side effects like dead-locks in existing applications. While we've that've made some changes in the past to help with thread safety (see the mailing list, jira, and PRs), doing that consistently is delicate and I am concerned about unintended side effects.

[ivakegg]

Well, given how many of the test cases are showing non-thread safety (as has been stated previously), I am going to change this to only test for thread safety where it is needed wrt the VFSClassLoader (e.g. jar/zip)

[ivakegg]

-1 failed build and: Quick note but the PR would need a detailed review from more folks: I am really skeptical here, VFS was not designed for thread safety; hacking some code here and there is more likely to make the code base much harder to understand and maintain. In the past, we've just told users they have to handle thread safety in their applications. Slapping "synchronized" here and there is likely to create side effects like dead-locks in existing applications. While we've that've made some changes in the past to help with thread safety (see the mailing list, jira, and PRs), doing that consistently is delicate and I am concerned about unintended side effects.

The only build failure is on mac OS 8 and the error does not appear to have anything to do with my changes. I understand the not designed for thread safety issues, however if you are going to supply a class loader implementation, then the underlying protocols being used will have to be thread safe.