Sounds like it's worth a JIRA to me. While not at all familiar with that class, I suspect that changing the signature at this point would break back-compat, so perhaps create a new c'tor and deprecate the old one if you decide to change the arg?
I'll defer of course to people who know the code.... Best, Erick On Tue, Jun 27, 2017 at 1:22 AM, Rob Audenaerde <rob.audenae...@gmail.com> wrote: > Hi all, > > We bumped into a hash collision in the LruTaxonomyWriterCache, which caused > a wrong facet to be indexed for a document. (It took some time to figure out > what was going wrong..) > > The javadoc of this constructor is not explicitly telling it is unsafe (you > need to check the enum itself to see this) > > Maybe to prevent others from running into this problem, the single argument > constructor can be switched to LRU_STRING? Or some extra javadoc to warn the > developer? > > WDYT? > > -Rob > > > > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
