lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Uwe Schindler (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (LUCENE-7114) analyzers-common tests fail with JDK9 EA 110 build
Date Sun, 20 Mar 2016 10:53:34 GMT

    [ https://issues.apache.org/jira/browse/LUCENE-7114?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15203239#comment-15203239
] 

Uwe Schindler edited comment on LUCENE-7114 at 3/20/16 10:52 AM:
-----------------------------------------------------------------

bq. Bugs get fixed by the developers quickly, but then take weeks to find their way into a
build, yet the builds are still horrendously unstable, so whats the point of all the delay?

I think this should again be brought up on the Hotspot mailing list. I already did this on
the core-dev list, but the people there are different and they don't have the problems. Non-Hotspot
fixes get it into the EA builds quite fast.

The problem are Hotspot fixes. As noted in my article about FOSDEM (which was titled "How
free is OpenJDK" in German version: see https://jaxenter.com/java-9-steals-spotlight-open-jdk-project-takes-back-123721.html),
the problems in Hotspot are more complicated. Also other non-Oracle developers, e.g. from
SAP, were complaining about the processes inside Hotspot and how patches get accepted. There
is also a separate test suite that is not public! Before this one not passes, it does not
get into EA builds and so on. To me it looks like there is a lot of additional bureaucratic
"quality" control behind the scenes that delay inclusion into the EA builds. It looks like
the stuff is merged several times to other branches (check out JIRA about that). The frightening
fact is: these "quality" checks don't ensure quality, as we all see. So it would be better
to remove that bureaucracy and let the community test as soon as the patch is committed to
repository. We should mention that over an over on conferences, discussions, and the mailing
lists.


was (Author: thetaphi):
bq. Bugs get fixed by the developers quickly, but then take weeks to find their way into a
build, yet the builds are still horrendously unstable, so whats the point of all the delay?

I think this should again be brought up on the Hotspot mailing list. I already did this on
the core-dev list, but the people there are different and they don't have the problems. Non-Hotspot
fixes get it into the EA builds quite fast.

The problem are Hotspot fixes. As noted in my article about FOSDEM (which was titled "How
free is OpenJDK in German"; see https://jaxenter.com/java-9-steals-spotlight-open-jdk-project-takes-back-123721.html),
the problems in Hotspot are more complicated. Also other non-Oracle developers, e.g. from
SAP, were complaining about the processes inside Hotspot and how patches get accepted. There
is also a separate test suite that is not public! Before this one not passes, it does not
get into EA builds and so on. To me it looks like there is a lot of additional bureaucratic
"quality" control behind the scenes that delay inclusion into the EA builds. It looks like
the stuff is merged several times to other branches (check out JIRA about that). The frightening
fact is: these "quality" checks don't ensure quality, as we all see. So it would be better
to remove that bureaucracy and let the community test as soon as the patch is committed to
repository. We should mention that over an over on conferences, discussions, and the mailing
lists.

> analyzers-common tests fail with JDK9 EA 110 build
> --------------------------------------------------
>
>                 Key: LUCENE-7114
>                 URL: https://issues.apache.org/jira/browse/LUCENE-7114
>             Project: Lucene - Core
>          Issue Type: Bug
>            Reporter: Robert Muir
>
> Looks like this:
> {noformat}
>    [junit4] Suite: org.apache.lucene.analysis.fr.TestFrenchLightStemFilter
>    [junit4]   2> NOTE: reproduce with: ant test  -Dtestcase=TestFrenchLightStemFilter
-Dtests.method=testVocabulary -Dtests.seed=4044297F9BFA5E32 -Dtests.locale=az-Cyrl-AZ -Dtests.timezone=ACT
-Dtests.asserts=true -Dtests.file.encoding=UTF-8
>    [junit4] FAILURE 0.44s J0 | TestFrenchLightStemFilter.testVocabulary <<<
>    [junit4]    > Throwable #1: org.junit.ComparisonFailure: term 0 expected:<grosi[]>
but was:<grosi[er]>
> {noformat}
> So far i see these failing with french and portuguese. It may be a hotspot issue, as
these tests stem more than 10,000 words.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org


Mime
View raw message