Adrien Grand created LUCENE-4438:
------------------------------------
Summary: Lucene40StoredFieldsReader's constructor calls close()
instead of IOUtils.closeWhileHandlingException in its finally block
Key: LUCENE-4438
URL: https://issues.apache.org/jira/browse/LUCENE-4438
Project: Lucene - Core
Issue Type: Bug
Affects Versions: 4.0
Reporter: Adrien Grand
Priority: Minor
Fix For: 4.1, 5.0
It would be nice to have automated tests for this kind of things (similarly to
the check-forbidden-api task).
Here is the discussion I just had with Uwe on IRC:
{noformat}
15:32 < jpountz> ThetaPh1: I just saw that Lucene40StoredFieldsReader's
constructor calls close in its finally block. I think it is wrong since close
might throw an IOE (should be catched), am I
correct? If yes, then is it something we could test with ASM
(similarly to the forbidden API checks)?
15:32 <@ThetaPh1> it does not use IOUtils?
15:33 <@ThetaPh1> we cannot check this with asm easily
15:33 <@ThetaPh1> we could only forbid calling Closeable.close() and exclude
IOUtils from that
15:34 <@ThetaPh1> so the correct fix is to use IOUtils when clsoing
15:34 <@ThetaPh1> it also checks for null and suppresses exceptions
15:36 < jpountz> ThetaPh1: no, it does no
15:36 <@ThetaPh1> its a bug :-)
15:36 <@ThetaPh1> but not serious
15:36 <@ThetaPh1> for local files it never throws exceptions
15:36 < jpountz> ok
15:37 < jpountz> I think catching calls to Closeable.close would already be nice
15:37 <@ThetaPh1> but we might think about disallowing Closeable.close()
15:37 < jpountz> I'll open an issue
15:37 <@ThetaPh1> we can add that as a separate check-forbidden-apis file
15:37 <@ThetaPh1> but exclude IOUtils from the fileset
15:37 <@ThetaPh1> unfortunately all tests do this
15:38 <@ThetaPh1> so i would restriuct this to non-tests, too
{noformat}
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]