Page MenuHomePhabricator

Regression Test OpenSearch Language Analysis
Closed, ResolvedPublic5 Estimated Story Points

Description

As we have for other major Elasticsearch upgrades, we should do a regression test between our current version of Elastic (7.10) and our target version of OpenSearch (1.x), to see if there are any problems that can and should be addressed.

Event Timeline

Gehel set the point value for this task to 5.Mar 3 2025, 4:47 PM

After I got my OpenSearch environment running and all the new versions of the plugins installed, there were almost no changes! Korean (using the Nori analyzer) had miniscule changes (5/260K tokens = ~0.002%). All three Japanese options (CJK, Kuromoji, Sudachi) had no changes. The old nnbsp problem persists, and our patch to fix it remains. Overall, everything looks great!

More details on MediaWiki.

EBernhardson subscribed.

Everything here looks reasonable, lets ship it

                   v  ~.      v
          v           /|
                     / |          v
              v     /__|__
                  \--------/
~~~~~~~~~~~~~~~~~~~`~~~~~~'~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~