[
https://issues.apache.org/jira/browse/LUCENE-2964?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13009062#comment-13009062
]
Doron Cohen commented on LUCENE-2964:
-------------------------------------
Apparently using '$' is safe in that it is always correct as this is the
standard definition for inner class separator - see [Inner Classes
Specification | http://www.retrologic.com/innerclasses.doc7.html].
The required change to the Benchmark Algorithm is to define '$' as a word part,
by adding this to the Algorithm's tokenizer intialization:
{code}
stok.wordChars('$','$');
{code}
I am not sure about this.
It would allow easier creation of tasks as inner classes.
But this is mainly for tests I assume, because for real tasks one would want
them to read nicely - without that '$' and to have their own class file.
And making '$' a regular word char just for this test need, will prevent us
from adding some semantics to it in the future, should the need arise,
So, I rather keep the additional class file in the test package.
> Allow benchmark tasks from alternative packages
> -----------------------------------------------
>
> Key: LUCENE-2964
> URL: https://issues.apache.org/jira/browse/LUCENE-2964
> Project: Lucene - Java
> Issue Type: Improvement
> Components: contrib/benchmark
> Reporter: Doron Cohen
> Assignee: Doron Cohen
> Priority: Minor
> Fix For: 3.2, 4.0
>
> Attachments: LUCENE-2964.patch, LUCENE-2964.patch
>
>
> Relax current limitation of all tasks in same package - that of PerfTask.
> Add a property "alt.tasks.packages" - its value are comma separated full
> package names.
> If the task class is not found in the default package, an attempt is made to
> load it from the alternate packages specified in that property.
--
This message is automatically generated by JIRA.
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]
