package cumulus
Install
Dune Dependency
Authors
Maintainers
Sources
sha256=f8b36587ef1cd547541e16fceb7b81cc1bbed26325200e51248564d094beff25
sha512=5515aff4a852decb9e9a078aa6692eb6c2d7108d55b3385396faa7f55f5d5af9c510911608e0f5a3e24ce736fe0742dd227bcc683eb56b45890495f38da5ac0e
README.md.html
Status
This library is in a working state, but the interface is not stable yet.
Synopsis
Cumulus defines a signal-like type, which facilitates differential updates of the underlying value. When a cumulus signal is changed, it emits a patch along with the new value, allowing consumers to integrate it into an existing value rather than rebuilding its value from scratch. Cumulus is based on the React FRP library.
Documentation
Overview
The main type ('a, 'b) Cumulus.t
represents a cumulus signal over a state of type 'a
with changes represented by 'b
. The idea is that an (state, change) Cumulus.t
instance makes sense if one can formulate a function
val patch : change -> state -> state
relating the previous value of the cumulus signal with the current via the current change component. If this is the case, the cumulus signal can be constructed from an initial value and an event of changes using
val integrate : ('da -> 'a -> 'a) -> 'da event -> 'a -> ('a, 'da) t
However, it is not required that patch
is manifest and, as will be seen below, the changes need not be complete for cumulus signals to be practically useful.
As special cases
val of_event : 'a event -> (unit, 'a) t
shows how a react event is the analog of a cumulus signal where no state is retained, and
val of_signal : 'a signal -> ('a, unit) t
shows how a react signal is the analog of a cumulus signal where no information is provided about what changed from one value no the text.
The state of a cumulus signal can be pure, so that a new state is constructed on each update of the signal, e.g.
('a list, [`Remove of int | `Add of int * 'a]) Cumulus.t
or mutable, allowing re-using the state between updates when possible, e.g.
('a array, [`Set of int * 'a | `Resize of int]) Cumulus.t
An essential part of the API is a set of lifting functions of different arities, which allows deriving new cumulus signal from existing cumulus signals. The basic pattern is:
let cs =
let init state1 ... stateN =
... (* construct the initial state from input status *)
in
let patch (state1, change1) ... (stateN, changeN) state =
(* inspect input states and changes *)
if ... (* our state changes *) then
let state', change' = ... in
Cumulus.Patch (state', change')
else
Cumulus.Keep state
in
Cumulus.l2 ~init ~patch cs1 cs2
See tests/test_isecn.ml
for a complete example.