Can we save a map access on every historical fixing lookup? about quantlib HOT 2 CLOSED

Yes, definitely there should only be a single lookup.

It's strange that T operator[](const Date& d) const exists because it's not defined on std::map and its semantics are fundamentally not const. I think we should deprecate this overload in favor of the other one.

It's also strange that const_iterator find(const Date&); is not const. It should be split into iterator find(const Date&); and const_iterator find(const Date&) const; like std::map.

from quantlib.

I'd say it's strange that std::map doesn't have a const operator[], not the other way around. Yes, I know it's because it creates the entry if it doesn't exist. That's an accident of the implementation, though. The const version makes sense in our class and I think it should stay there.

I would agree about find but currently we only have const_iterator. I don't remember the rationale, but it seems that the class provides the bare interface to build the time series and then treat it as immutable. I'm not sure though.

This said, @tomwhoiscontrary , please go ahead with your PR. Thanks!

from quantlib.