Are all C/C++ APIs for example spell()
, suggest()
; analyse()
in the Hunspell library thread-unsafe?
When I use the suggest()
API with a lock, I see on an average 50-100 requests for suggestion processed in a second? Did any one try to do benchmarks on this?
Any suggestion for improvement is welcome (including any commercial options for Spell Check and corrections, our web-application is based on C++.)