probably there are two possible ways to interact with build server process and the IDEA IDE process:

1) build server performs (parallel) instrumentation in memory and at the end of the whole build writes or updates clover.db (it has write access); next it sends notification to IDE which refreshes views and editors; potential problems: deletion of database during build, locking files

2) build server performs (parallel) instrumentation sending information about code structure back to IDEA IDE, which manages the database and sends back indexes of elements; potential problems: high messaging overhead -> need to send at most one message per file -> need to introduce local offset constant per each file;