[
https://issues.apache.org/jira/browse/SOLR-13952?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16979772#comment-16979772
]
Erick Erickson commented on SOLR-13952:
---------------------------------------
Here are some things I'm seeing I could use some opinions on. This has been
pushed to a new branch in the repo, jira/SOLR-13952
--------------
Mark added the first two classes for ignoring thread leaks
SolrIgnoredThreadsFilter and QuickPathThreadsFilter to quite a number of Hdfs
tests: BadHdfsThreadsFilterClass was already there:
{code}
@ThreadLeakFilters(defaultFilters = true, filters = {
SolrIgnoredThreadsFilter.class,
QuickPatchThreadsFilter.class,
BadHdfsThreadsFilter.class // hdfs currently leaks thread(s)
})
public class HdfsThreadLeakTest extends SolrTestCaseJ4 {
{code}
He also added
{code}
if (threadName.startsWith("Log4j2-TF-2-AsyncLoggerConfig")) {return true; }
{code}
to SolrIgnoredThreadsFilter.
Does anyone who knows more about Hdfs tests than I do have an opinion here? My
presumption is that this is fine, but I'd like to understand it a bit more.
There's also similar additions to SaslZkACLProviderTest,
SharedFSAutoReplicaFailoverTest, SolrTestCase, TestSolrCloudWithKerberosAlt
--------------------
HTMLCharacterEntities.jflex, line 63 has been changed from:
{code}
| "zwj" | "zwnj" )
to
(' | "zwj" | "zwnj"', ')')
{code}
Is this something that makes sense? This regenerates HtmlStripCharFilter of
course. I have no problem adding this in, just want a sanity check. For the
nonce, I've put a "nocommit" in the jflex file in a comment, I've made a note
to myself to grep _everything_ for nocommit
-------------------
QueryParser.jj has a SynonymQueryStyle added to newFieldQuery that extends up
to the base class. Why?
-----------------
I pretty much just accepted all of Mark's annotations.
-----------------
XmlOffsetCorrector.java has some changes I don't understand. //nocommitted
--------------
Anyone who has any light to shed, please do. Otherwise that's enough for one
night. So far it compiles, but certainly doesn't succeed through precommit.
I'll try running through all the tests tonight. I _believe_ I've taken all of
the gradle-related changes out.
> Separate out Gradle-specific code from other (mostly test) changes and commit
> separately
> ----------------------------------------------------------------------------------------
>
> Key: SOLR-13952
> URL: https://issues.apache.org/jira/browse/SOLR-13952
> Project: Solr
> Issue Type: Task
> Security Level: Public(Default Security Level. Issues are Public)
> Reporter: Erick Erickson
> Assignee: Erick Erickson
> Priority: Major
>
> The gradle_8 branch has many changes unrelated to gradle. It would be much
> easier to work on the gradle parts if these were separated. So here's my plan:
> - establish a branch to use for the non-gradle parts of the gradle_8 branch
> and commit separately. For a first cut, I'll make all the changes I'm
> confident of, and mark the others with nocommits so we can iterate and decide
> when to merge to master and 8x.
> - create a "gradle_9" branch that hosts only the gradle changes for us all to
> iterate on.
> I hope to have a preliminary cut at this over the weekend.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
