Caching readline completion
I'm thinking about caching the results of aql completers (except filename completer) for aql readline completion.
Questions
- Is it worth it? We are not daeling with huge amount of data (right now i think there about ~100 strings at most - list of algorithms/datatypes). I dont think that there will be more than (lets say) 4 calls of this function while writing a single line (
au<TAB>::co<TAB>::ToRTE<TAB>St<TAB>
) - If I implemented it, when to invalidate cache? Is the dynamic (un)loading of libraries already implemented? Or are there any other events that should invalidate the cache?