LPP » INSTRU » QLop » SciQLOP

After investigation, it seems that the lack in performance is due to :

• a huge number of calls to Regexp's match function
• the number of field tested for Regexp is too important (19 fields)

Performances increased a lot by pre-filtering on the data.
The proxy now only checks a data attribute, which takes a lot less time than the previous implementation. (check on parents and children for every node)

Still needs improvements to have a letter-by-letter filtering though.