I am using ESM in my node application, when I tried opentelemetry to auto instrument my app using #opentelemetry/auto-instrumentations-node package I could not generate any tracing, is this normal behavior? is there a suggested way to auto-instrument with ESM? Or I am left with the option to manual instrument my app.
The auto-instrumentation only works with libraries that are currently supported.
You can take a look at the OpenTelemetry Registry to check all supported libraries.
You can also check the GitHub repository, in which you could raise a Feature Request to ask for the implementation of an auto instrumentation for ESM.
At the moment ESM is not supported by the auto instrumentation, and in order to get traces and metrics to your code, you will need to manually instrument your code.
This process is detailed in the OpenTelemetry official documentation.
Related
I'm currently working on an IPFS/Ethereum dapp in Muon.
Because i need Metamask i started with this Boilerplate: https://github.com/SwapyNetwork/electron-metamask-boilerplate
Everything is working fine so far.
However i can not use require('anything') in the renderer process or in html script tags. (See below)
There seems to be a problem with the boilerplate code but i can't find it.
Or is node code in renderer not supported in Muon?
My only change in testing is setting node-integration explicitly to true and
inserting require('fs') in index.js. (i installed fs of course).
I tried many different solutions from stackoverflow and other sites but couldn't find a solution yet.
Error Message
Thank you
As per the muon's github repo:
Some of Muons goals include:
Use the Chromium source directly (eliminating electron's copy of chrome_src) with minor patches
make integrating chrome components less painful
faster and more streamlined end-to-end build process (see browser-laptop-bootstrap).
add support for Chrome extensions
add security focused features for the renderer:
remove node completely (from the renderer process)
full sandbox
scriptable window.opener support
As you can see there, muon does not supports node code in the rendered. It is by design for security purpose. Muon may be great for certain applications, but I recommend switching on to electron if you really need to use require in the renderer.
I am getting very hard time to configure Sampling Heap Profiler node module.
I would like to set up this module and want to try deploying in chrome dev tool but unfortunately, there is no document about build and deployment.
Here is the npm link - https://www.npmjs.com/package/heap-profile
Any help would be greatly appreciated.
The sampling heap profiler is designed to be used programmatically in Node.js applications.
For frontend applications, you actually don't need this module. The functionality is baked into DevTools. You can find the 'Allocation Sampling' profiler in the 'Memory' tab in Chrome Devtools:
I would like to know whether it is possible to have third-party javascript libraries' source code available whilst debugging.
FYI, I use npm/nodejs and the angular CLI (which itself relies on Webpack).
Example libraries (together with their source languages) that I would like to have available during debugging are:
Angular 2 (typescript)
RxJS (typescript)
I guess what I want to achieve is related to configuring source maps.
Any comment or guidance welcome.
edit: Can someone please advise how to configure the angular CLI in order to have angular and RxJS typescript sources available whilst debugging?
Yes, to be able to set breakpoints in source files while debugging, you need sourcemaps. But this is not the thing that can be configured in the IDE, you need to set up your build tools accordingly. The only thing that should be configured on the IDE end is the run configuration - you might need to specify Remote URL mappings for your project directories
To complement lena's answer and as of #angular/cli version's 6.1 is it now possible to output source maps for vendor libraries using the following syntax:
ng serve --source-map --vendor-source-map
It also works for the ng build command.
See also: https://blog.ninja-squad.com/2018/07/27/angular-cli-6.1/
It is then possible to debug third-party libraries using an IDE or the browser.
When developing front-end code for the browser, I often use the es2017 preset when transpiling down to a distribution bundle, which allows me all the conveniences of the included transformers. For conventional modules, I usually stick to whatever the required node engine I've specified for that particular module supports.
I would like to start developing these "conventional" modules using babel transformers as well, but I can foresee drawbacks to this, including:
It might inhibit the debugging workflow (more specifically when working with an IDE)
The performance of the module might suffer
What's the current state on this matter - would you say it makes sense to use babel in conventional modules given the aforementioned and other trade-offs? What are the pros/cons for your preferred workflow?
Bonus question: What are some reputable modules and/or module authors out there that are already using this technique? I've seen Facebook do it for their react ecosystem but I guess that makes sense since those are mostly modules for the browser.
It is converted back to vanilla JS (babel does that part).
What you get is that you can utilize Classes which I found useful.
Hopefully with time, browsers will support ES6 and we will not need babel.
The only drawback is that when debugging, you have to produce a source map, but that is temporary, see above.
To answer your second question: I'm using React in one of the websites, and most of the modules I needed (from npm) are using ES6.
I believe that the trade-offs or drawbacks that you mention both do not apply to developing nodejs code using babel as ES7 transpiler. Personally, I find using ES7 features with node tremendously productive.
There is source map support for debugging. I use karma for testing and it comes with excellent source map support (I use IntelliJ but I believe most IDEs will do). You can checkout this REST-API repository on github. It's a nice stack for building nodejs data backend. It uses karma for testing - even comes with code coverage support. It also integrates with pm2 for scaling and availability.
Regarding performance: I think transpiled code has been shown to run faster in many scenarios than the code a developer would write when not having advanced language features available. I will post some links later.
I would like to start using advanced JS features in an pre-existing app with a NodeJS serverside, React using the Fluxible architecture, Gulp task runner and Broserify/CommonJS front end modules.
Anybody who has been down that path or a similar path before and wants to share some insight I would much appreciate it.
babel-node compiles on-the-fly. You can use the API (babel-core) to pre-compile and then run the compiled output in node. There's also a gulp-babel plugin. At the expense of extra processing overhead at build time you could hijack browserify or use module-deps to figure out the dependency graph for you, if relevant. There's a notion of adding a feature to Babel to generate a dependency graph, but it's not available currently.