Details

    • Improvement
    • Status: Closed
    • Major
    • Resolution: Fixed
    • None
    • 3.6, 4.0-ALPHA
    • Response Writers
    • None

    Description

      This stems from issues with SOLR-2588, moving the Velocity response writer back to contrib. We still want the example app to use the VelocityResponseWriter for the /browse interface. Many of Solr's core tests use the example Solr configuration. There are other contribs that are brought into the example app (extract, clustering, DIH, for examples) but these are request handlers that lazy load. Response writers don't currently lazy load, thus causing core tests that use the example config to fail unless "ant dist" is run.

      Attachments

        1. SOLR-2718.patch
          9 kB
          Erik Hatcher
        2. SOLR-2718-velocity-test-cleanup.patch
          3 kB
          Erik Hatcher

        Issue Links

          Activity

            ehatcher Erik Hatcher added a comment -

            Another option is to adjust the tests to not use the example solr-home - there's not that many of them. I'll investigate. We still want example tested somehow, but more at an integration level than a unit level.

            ehatcher Erik Hatcher added a comment - Another option is to adjust the tests to not use the example solr-home - there's not that many of them. I'll investigate. We still want example tested somehow, but more at an integration level than a unit level.

            adding a lazy load option to response writers seems like a good idea to me ... just want to make sure it defaults to false just like request handlers

            hossman Chris M. Hostetter added a comment - adding a lazy load option to response writers seems like a good idea to me ... just want to make sure it defaults to false just like request handlers
            ehatcher Erik Hatcher added a comment -

            Here's a patch that implements lazy loaded response writers, along with a test. A little bit of copy/paste/adapt from the request handler lazy loading, so perhaps worth implementing some common infrastructure for this pattern?

            ehatcher Erik Hatcher added a comment - Here's a patch that implements lazy loaded response writers, along with a test. A little bit of copy/paste/adapt from the request handler lazy loading, so perhaps worth implementing some common infrastructure for this pattern?
            ehatcher Erik Hatcher added a comment -

            Here's a patch to clean up the hack made for SOLR-2588 due to the example configuration using the VelocityResponseWriter but it not (necessarily) built when tests are run. With lazy loaded response writers, no worries now.

            ehatcher Erik Hatcher added a comment - Here's a patch to clean up the hack made for SOLR-2588 due to the example configuration using the VelocityResponseWriter but it not (necessarily) built when tests are run. With lazy loaded response writers, no worries now.
            ehatcher Erik Hatcher added a comment -

            I've seen two reports from Solr 3.5 users today (via #solr in IRC) where they've deployed Solr into Tomcat and copied the example configuration. The lazy loading was not backported to 3.x and thus the registration of VRW causes a startup error when the libs aren't found.

            This needs to be ported/adapted to 3.x to ensure 3.6 doesn't have this issue.

            ehatcher Erik Hatcher added a comment - I've seen two reports from Solr 3.5 users today (via #solr in IRC) where they've deployed Solr into Tomcat and copied the example configuration. The lazy loading was not backported to 3.x and thus the registration of VRW causes a startup error when the libs aren't found. This needs to be ported/adapted to 3.x to ensure 3.6 doesn't have this issue.
            ehatcher Erik Hatcher added a comment -

            Fixed already on 4.0, but needs to be ported to 3.x too.

            ehatcher Erik Hatcher added a comment - Fixed already on 4.0, but needs to be ported to 3.x too.
            sarowe Steven Rowe added a comment -

            Backported to branch_3x and committed.

            sarowe Steven Rowe added a comment - Backported to branch_3x and committed.

            People

              sarowe Steven Rowe
              ehatcher Erik Hatcher
              Votes:
              0 Vote for this issue
              Watchers:
              0 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: