Documentation really sketchy on how to export a module

[moll]

I've been going back and forth the README and code for a while now trying to understand how the damn this is supposed to be used for exposing a module in the browser.

The beep-poop example and README mention of a --require flag but that even fails in the example's build.sh as robot.js is not a module in ./node_modules, but just a file.

Then an attempt to use --require with a relative path ends up meaning you'll have to use require("./foo") in the browser as well, as opposed to just using require("foo"). Oh, hey, Optionally use a colon separator to set the target., but that's not even implemented according to #319?!

Symlink the name of your library to your index.js and relative --require that? Well, that prints out peculiar useless intermediary requires named in the style of "l5QO1E".

So, how do you package an existing index.js file so you can do require("foo") in the browser?

[ghost]

If you have a foo.js:

module.exports = 555

and want require('foo') to work in arbitrary script tags, do:

$ mkdir -p node_modules/foo
$ mv foo.js node_modules/foo/index.js
$ browserify -r foo > bundle.js

Now you can require('foo') in any script tag:

<script src="bundle.js"></script>
<script>
var foo = require('foo');
console.log(foo);
</script>

which is the same as:

$ cat bundle.js <(echo "console.log(require('foo'))") | node
555

---

However, it's more common to just put the part that does require('foo') in a main.js and browserify up the entire thing into a single bundle:

main.js:

var foo = require('foo')
console.log(foo)

$ browserify main.js > bundle.js

Then just put a single script tag into your page:

<script src="bundle.js"></script>

[moll]

Thank you for the workaround example!

I do have to ask, why is the that common case so complicated? You'll have to make a whole fake directory structure and copy your code there for just building a regular library that exports regular functions to be used both with Node and in the browser.

A related question, the --requireflag seems to assign the require functions to a global variable named require. That's expected. But it seems to do so in a way that overwrites its previous value, which I was under the impression from the README, is supposed to be re-used to work with other browserified code. Or am I reading the code wrong?

[ghost]

To make require('foo') work it is exactly the same in node. Non-relative paths get resolved in node_modules/. If you want a relative path you can just do require('./foo.js') in a main.js entry file. If you have explicit entry points everything is very simple because that environment is completely self-contained so it's much easier to support.

The global require definition uses the previous value in order to cascade across multiple bundles.

[moll]

Umm, I get your simplicity point, but that shouldn't make exporting the library in the current working directory that difficult and inelegant. What's with the target option to --require? That seems to be something that would make "make this library I'm building work in the browser" possible.

You're right, I was reading the code wrong and assumed the require inside the returned function ends up referring to itself, which it doesn't.