@hackage lazy-async1.0.0.1
Asynchronous actions that don't start right away
Categories
License
MIT
Maintainer
Chris Martin, Julie Moronuki
Links
Versions
Installation
Dependencies (8)
- base ^>=4.14 || ^>=4.15 || ^>=4.16
- exceptions ^>=0.10.4
- lifted-async ^>=0.10.0.6
- monad-control ^>=1.0.2.3
- rank2classes ^>=1.4.0.1
- stm ^>=2.5 Show all…
Dependents (0)
Sometimes we have a bunch of IO actions that do things like
read files, make HTTP requests, or query a database. Some of the
information that these actions produce might not end up being
needed, depending on the breaks. In the interest of avoiding
unnecessary effort, we don't want to simply run all the actions
and collect their results upfront. We also don't want to simply
run an action right before its result is needed, because it might
be needed in more than one place, which opens the possibility of
unnecessarily running the same action more than once. In
situations like these, we use LazyAsync.
Under the hood, an IO action is turned into a LazyAsync by
constructing two things: An Async (from the async package),
and a TVar Bool (from the stm package). The TVar, initialized
to False, indicates whether the action is wanted yet. The async
thread waits until the TVar turns True and then runs the action.