$ npm install hypertrace
Add tracing and insights to classes. One of the goals of the module is that there is close to zero overhead when tracing is not enabled. This is achieved by not enabling tracing on objects created before tracing was enabled.
Similarly if caching is enabled then there is almost zero overhead while tracing.
There is support for Prometheus/Grafana through hypertrace-prometheus to get better visual insights into an application's behavior.
$ npm i hypertrace
First create tracers in the classes where insights are needed
some-module.js
import { createTracer } from 'hypertrace'
export default class SomeModule {
constructor () {
this.tracer = createTracer(this, {
props: {
some: 'property'
}
})
}
createChild () {
const child = new Child(this.tracer)
return child
}
}
class Child {
constructor (parentTracer) {
this.tracer = createTracer(this, {
parent: parentTracer,
props: {
another: 'value'
}
})
}
foo (val) {
this.tracer.trace({ val })
}
}
Then add .setTraceFunction()
when traces are needed. It's important that this happens before classes that use Hypertrace are instantiated. Otherwise the tracer will not be enabled for those objects.
app.js
import SomeModule from 'some-module'
import { setTraceFunction } from 'hypertrace'
// Log everytime .trace() is being called.
// Important to call `setTraceFunction` BEFORE objects are instantiated and calls `createTracer`
setTraceFunction(({ id, caller, object, parentObject }) => {
console.log({
id,
caller,
object,
parentObject,
})
})
const mod = new SomeModule() // Inherently calls `createTracer`
const child = mod.createChild()
child.foo(123)
/*
Prints out:
{
id: null,
caller: {
functionName: 'foo',
filename: '/Users/.../app.js',
line: 28, // The line where .trace() is being called
column: 19,
props: { val: 123 }
},
object: {
className: 'Child',
id: 1,
props: { another: 'value' },
ctx: Child { tracer: [Hypertrace] }
},
parentObject: {
className: 'SomeModule',
id: 1,
props: { some: 'property' },
ctx: SomeModule { tracer: [Hypertrace] }
}
}
*/
Create a new Hypertrace instance inside a class. Often used in the constructor
.
If this is called before setTraceFunction
then it will return a cummy class. This means that there will be close to zero overhead when tracing is not needed.
class SomeClass {
constructor() {
this.tracer = createTracer(this)
}
}
If the trace function has been set with setTraceFunction
, then it is called.
Note: If the trace function has not been set, there is no overhead in calling this.
Note: If .trace()
is called very often then there can be some measurable overhead if the trace function has been set. To avoid this overhead pass id
, which then always calls the trace function with the same cached parameters. props
will always be passed.
class SomeClass {
constructor() {
this.tracer = new Hypertrace(this)
}
fn (some, val) {
this.tracer.trace({ some, val })
}
}
Sets the parent tracer of this one. This means that subsequent calls to .trace()
will have the new parentObject
.
If parentTracer
is null, then the current parent is removed and subsequent calls to .trace()
will not have a parentObject
.
A flag that says if the tracer is enabled for this Hypertrace. This is true
if a trace function was set before initiating, and false
if not.
The objectId
of this instance.
The className
of this instance.
The props
passed when this instance was created.
The ctx
of this instance. If createTracer(this)
then ctx = this
.
Set a global trace function that is invoked everytime .trace()
is being called.
Important: Tracing is only enabled for objects created after setTraceFunction
is called.
ctx
, className
, id
, and props
parent
then it contains ctx
, className
, id
, and props
functionName
, filename
, line
, column
, and `propsRemove the global trace function. Calls to createTracer
after this will return a dummy object to reduce runtime overhead.
© 2010 - cnpmjs.org x YWFE | Home | YWFE