How does anyone know if their customer got any JS errors/application crashes? How does the operations team know about the crash before the customer screams on email/social media? Finding an answer to this question led to the building of my first npm package.
What is the motivation behind this npm package
Every front-end app needs error reporting and monitoring.
Error reporting means, the application errors need to be sent to the developer for analysis and fixes. The next step is the monitoring for which you can run some cron job etc.
Most JS frameworks/libs do not have support for error reporting to the service. Things like Reach error boundaries can stop the app from the complete crash but how will the developer sitting elsewhere know about the crash that happened to the end-user? This is a gap which this npm package tries to solve
What about amazing tools out there?
There could be amazing tools like sentry etc but they are cloud-based. If you are working in a company, you need manager approval, and it's a paid app. The manager will hate you because of its difficult times.
Moreover, we have to analyze the data policy. Some people think open source is free which is not true, the developer's time costs money. So setting up an open-source sentry is not an option at all.
Was deploying to npm easy
Deploying to npm is easy but deploying the right is hard without npm.
-
Scaffold the npm package with
npm init -
create a JS module with
module.exports -
The functionality can depend on other packages as well.
-
Meanwhile get the login access at npmjs, set it up on the local machine &
npm deploy -
The source code can be exposed to Github. So push it to a repo on GitHub.
-
But you see the code and node versions are not tagged. so we need to use
npm
what is this npm package about
np is a CLI utility that ensures you follow all the best practices when you do an npm publish.
publish can be hard because you need to take care of server & tag the version on git.
All of this is taken care of by npm. It's free to use.
What's next?
There are useful metrics we can collect from the site of error. It could be navigator object information and much more.
The immediate step is to fixate on the information that will be collected & reported.
The collected information needs to be easily clubbed up with tools like SPLUNK for monitoring.
Of course, there is a lot of work to do.
Anything else
After a long time, I got to write some code outside of React. I wrote a basic Javascript class and instantiated the object.
References
send-js-errors-to-backend on npmjs