monitor()

monitor() is the low-level (but exposed) API responsible for monitoring the use of flows and reactive variables.

Syntax

monitor(types, callback);

Parameters

types
An array of the types to monitor the callback for. By default, the only valid types are 'undo' and 'live', though custom types may be extended using monitor.register().
callback
A function to monitor. This may be an asynchronous function, though for monitorability await must be used in conjunction with until().

Return value

A call object; it is guaranteed a .result key containing the return value of the function. For each type monitored, there is a same-named key on the call object with the aggregated monitoring data. The exact type or shape of these values differ on a per-type basis; see monitoring undo and monitoring live for details on their respective interfaces.

Methods

monitor.ignore()
Inside a monitored context, prevents the monitoring of its callback.
monitor.add()
Manually notifies the monitored context of an item for a sepecific type to be monitored.
monitor.register()
Register custom monitorable types.
Monitoring async functions with until()
Asynchronous functions are not out-of-the-box monitorable. await expressions break the synchronous flow of functions, and as such, until() has to be used with await in order for asynchronous functions to remain monitorable.
Monitoring the "undo" type
Used internally in flows, is intended to define cleanup functions for contexts that benefit from it.
Monitoring the "live" type
For internal use only. This type is used for monitoring the use of live variables.

Details

The monitoring system is used internally to monitor for live variables as well as cleanup callbacks. It monitors a callback by opening a monitored context right before it executes the callback to be monitored. Only one monitored context can be open at any given moment; when opening a new monitored context inside another, it pauses the outer context for the duration of the inner one. After the callback has returned, the monitored context is closed, and the aggregated monitoring data is returned from the monitor() call in the form of a so-called "call object". In the case of an asynchronous callback, it is possible that this call object implicitly changes over time. Once this call object has been obtained, it is possible to hook into when a live variable that was used in the callback changes, or we could run the cleanup callbacks that were added to the monitored context. In practice, this is rarely something that needs to be handled manually, and even if so, it's good practice to write a wrapper function for the specific use-case. It is also good to understand where Yozo uses this mechanism; specifically,

There are not actually many different places where things are added to a monitoring context. For the "live" type, only accessing live variables does so, and for the "undo" type, only the creation of flows does so. Note that the latter happens quite a bit; when() creates and returns a Flow, hooks such as connected() do, effect() does, timing-related functions such as interval() do, and even register() does.

Examples

While monitoring manually is generally not needed, here's an example of monitoring for cleanup callbacks.

Monitoring cleanup

See the following snippet of code:

const call = monitor(['undo'], () => { when(button).clicks().then(() => { console.log('button clicked!'); }); }); button.click(); // "button clicked!" call.undo(); button.click(); // * crickets *

In this example, we monitor a hard-coded callback (which isn't particularly useful for anything other than educational purposes). We monitor only for "undo" through the first argument, ['undo'], and provide a callback that sets up an event listener using when(). Since when() returns a flow, and flows add a cleanup callback to the monitored context, the returned call object can undo this event listener using the .undo key (which is a function - see "undo" for more details).

Usage notes

As mentioned, this API is very low-level and generally not to be used willy-nilly. If monitoring is needed, and no function exists for it yet in Yozo, write a wrapper function for more readable and streamlined monitoring.