|
Tasks are often planned and performed repeatedly, very often recurring in regular intervals. This includes annual events on calendar dates.
Hence what we need is an iterator that is able to iterate over time and date.
The Boost Date/Time library is a very powerful library for this. In fact,
synkronize utilises the Boost Date/Time library.
The problem is that time/date iterators relate to scheduling like C++ iterators to their respective container/range.
A programmer usually cares less about the iteration process and the necessary iterator - rather the wish is to express when and how often to schedule an action. Ultimately, one needs the timepoints yielded by the iteration process.
It's all about simplicity and semantically meaningful usage.
Obviously things are getting quickly more complicated if you wanted to deal with multiple calendar dates and certain actions to be triggered - consider e.g. a U.S. Federal Holiday schedule that should suspend a program during these known days of observance.
This is where the huge benefit of synkronize providing a
timepoint generator comes into play. It's a timepiece device that you simply
configure how you want it to generate timepoints, and off you go getting sequential
timepoints. That's also the reason why the library focuses on the concept of
'scheduling' rather than timepoint iteration.
The synkronize library offers a rich set of features for
date/time scheduling.
Yet, there are basically only 4 things to do:
// 1) prepare schedule definition fd::interval_schedule_data sd{ fd::day_interval }; // 2) create and initialize timepoint generator from schedule definition fd::schedule_timepiece scheduler{ sd }; // 3) extract generated timepoint as unix timestamp std::chrono::seconds next = currently_scheduled_at__s(scheduler); // 4a) advance to next timepoint and extract generated timepoint next = advance__s(scheduler, fd::timepiece_move::to_adjacent) // 4b) work with the generated timepoint next = currently_scheduled_at__s(scheduler);