lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Michael McCandless (JIRA)" <>
Subject [jira] Commented: (LUCENE-2449) Improve random testing
Date Thu, 06 May 2010 23:55:48 GMT


Michael McCandless commented on LUCENE-2449:

Many of the failures are caused by a devious bug in DoubleBarrelLRUCache.  The bug is when
it promotes an entry from secondary -> primary on lookup; it uses the incoming key argument
to store the entry in the primary.  The problem is this key may be volatile (eg by StandardTermsDictReader)
which then messes up the ConcurrentHashMap and can lead to invalid cache hits.

But the AIOOBE bug is caused by a JRE bug!!
 We've worked around it by not using the String ctor that takes int[] code points.

We gotta do more of this random testing!

> Improve random testing
> ----------------------
>                 Key: LUCENE-2449
>                 URL:
>             Project: Lucene - Java
>          Issue Type: Test
>          Components: Build
>            Reporter: Robert Muir
>            Priority: Minor
>             Fix For: 3.1, 4.0
>         Attachments: LUCENE-2449.patch, LUCENE-2449.patch
> We have quite a few random tests, but there's no way to "crank" them.
> The idea here is to add a multiplier which can be increased by a sysprop. For example,
we could set this to something higher than 1 for hudson.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message