Ard Schrijvers wrote:
Now, Marcel correctly pointed about the problem when referring to the node
scope search:
//*[jcr:contains(., 'hägar')]
As I have given it another thought, it does make sense to me, that when
searching in the node scope, you use the global/default analyzer, and when
you are searching in a property, you might use a specific analyzer you used
for that property. In this way, we can actually do stemming and stopwords,
and implement synonym analyzer for any language you want to support per
property. IMHO, configurable analyzers in IndexingConfiguration are a
valuable add. Also, people who don't want it, because they do not like the
idea that you might find a hit for some word in a property, while it is
possible that there is no hit on the same word for the node scope (due to
analyzers), just do not configure it. WDOT?
sounds good. We just have to document the limitations properly.
regards
marcel
