You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@daffodil.apache.org by "Michael Beckerle (JIRA)" <ji...@apache.org> on 2018/12/12 15:57:00 UTC

[jira] [Resolved] (DAFFODIL-2034) Pool leaks if encodingErrorPolicy="error" globally

     [ https://issues.apache.org/jira/browse/DAFFODIL-2034?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Michael Beckerle resolved DAFFODIL-2034.
----------------------------------------
    Resolution: Fixed
      Assignee: Dave Thompson  (was: Michael Beckerle)

Fixed in 3868989577be2e9b85194cc866de9e29cf4c574e

There's no functional test for this. It's reducing a test interaction when a test fails with an assertion failure. 

> Pool leaks if encodingErrorPolicy="error" globally
> --------------------------------------------------
>
>                 Key: DAFFODIL-2034
>                 URL: https://issues.apache.org/jira/browse/DAFFODIL-2034
>             Project: Daffodil
>          Issue Type: Bug
>          Components: Back End
>    Affects Versions: 2.2.0
>            Reporter: Michael Beckerle
>            Assignee: Dave Thompson
>            Priority: Major
>             Fix For: 2.3.0
>
>
> I've discovered many 
>             TLRegistersPool.returnToPool(delimReg)
> Unprotected by try/catch logic. You can search and find them.
> If a test case fails with say, Assert.nyi (such as for encodingErrorPolicy="error") such as test_hexBinary_unparse_14, then the pool is not properly cleaned up, and a subsequent test that just so happens to run on the same thread, then gets an error. E.g., test test_hexBinary_unparse_15 fails with a pool error, as does test_hexBinary_unparse_19. Even though if you run each of these tests individually, it passes.
> This creates a situation where tests that should be isolated from each other are in fact interacting with eachother. 



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)