[jira] [Commented] (LUCENE-5874) Allow choosing latest default codec (not random)

["Robert Muir (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/LUCENE-5874?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14089352#comment-14089352 ] 

Robert Muir commented on LUCENE-5874:
-------------------------------------

Can you comment on the proposal on LUCENE-5858 instead?

I argue there we should remove such ancient codecs from rotation: its annoying because they dont support the full feature set and just causes a bunch of wasted time debugging and false failures.

Furthermore this doesn't really bring any value anymore, it doesnt find bugs in these old codecs, it just brings pain.
Instead we have a better testing directly against old codecs, we have the mechanism to test these guys directly (various "unit" tests like BaseXXXFormatTestCase).

We could do this before trying to split out a jar, just as a step.

> Allow choosing latest default codec (not random)
> ------------------------------------------------
>
>                 Key: LUCENE-5874
>                 URL: https://issues.apache.org/jira/browse/LUCENE-5874
>             Project: Lucene - Core
>          Issue Type: Test
>          Components: modules/test-framework
>            Reporter: David Smiley
>
> Within the Lucene/Solr tests, I can see that we want to widely test the various Lucene Codecs.  And so a test can use a black-list @SuppressCodecs when certain codecs are known not to work with certain tests.  But there is no white-list nor white-list by capabilities (e.g. saying you need term vectors that support payloads).
> In external applications that are using the test infrastructure, this is annoying.  An application generally wants to test with only the codec that ships as default with the particular Lucene version they use.  Now sure, this could be done by using "-Dtests.codec=Lucene46" , or having your \@BeforeClass explicitly do Codec.setDefault(Codec.forName("Lucene46")).  But in both these cases, you have to call out a specific version; you can't simply ask for whatever the latest is for the Lucene version you're using that's on your classpath.  So what I propose is two things: a whitelist annotation \@LuceneTestCase.Codecs and a value "CURRENT" (to align with the LUCENE_CURRENT naming convention).  In this way a test could use the latest Codec and not hit a random failure the first time around just because Lucene3x or some other old codec was chosen.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

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