id needed in script tag

[TheThirdOne]

Why exactly does the script tag need to have an id?

This seems unnecessary and it would allow more diversity if jsrepl was not dependent on DOM . This is a particularly large problem when working with Chrome extensions, where you have no control over the background DOM.

[amasad]

it's needed to get the url for the sandbox to load[1]. The sandbox being the iframe on old browser, or the Web Worker in newer browsers. Which will contain the interpreter for security, isolation, and performance reasons.

[amasad]

If you have better ideas, please would love to hear them

[TheThirdOne]

I see now why it needs an id. One idea I had as a fallback at least:

function endsWith(str, suffix) {
    return str.indexOf(suffix, str.length - suffix.length) !== -1;
}
var script = document.getElementsByTagName('script');
for(var i = 0; i < script.length;i++){
if(endsWith(script[i].src, 'jsrepl.js')){
return script.src
}
}

It's not ideal in the least, but it would remove (or negate the need of) the id system. It would simply need the name of the file to be jsrepl.js

[amasad]

I feel that the name of the script is the most variable here. But that wouldn't hurt I guess. Maybe a regexp to look for jsrepl would be best.

[TheThirdOne]

Sorry, I accidentally posted early. Why do you feel the name is the most variable?

[TheThirdOne]

Another suggestion, what if when a user initializes jsrepl, they supply a src attribute if they need to deviate from the normal pattern? That should be unobtrusive, and flexible for any projects needs.

[amasad]

Because people load the scripts from different folders and sometimes append version numbers to scripts.
I know that ACE the code editor, does a similar thing because it uses workers too. Take a look at how they do it. They match using regexp but you can also pass in options using data- attributes.
https://github.com/ajaxorg/ace/blob/86e7a496667f18913ff9bf4019f802578d6524b5/lib/ace/config.js#L155-L185

[TheThirdOne]

At least loading from folders is handled by my approach, but regex would be fine for testing src to see if it is jsrepl.js or some variant. I think a combination of manual override, searching script elements, and special id would provide coverage for almost all use cases and not be hard not implement.

[amasad]

ah sorry, missed the endsWith function. Sounds great, feel free to send a pull request to implement this. Let me know if you need any help

[TheThirdOne]

Is there any way to move the loader script detection inside the loader class, or any easy way a user could manually initiate it?

I will be glad to make a pull request implementing this, but it may have to wait for a bit because I have finals and don't know coffee-script at all.

[amasad]

It seems possible but would take a little bit of refactoring.

CoffeeScript is not that different from JS, try this book http://arcturo.github.io/library/coffeescript/

[TheThirdOne]

I realize that CoffeeScript is not very different, but it has different enough syntax that I can't whip up a pull request in a very short time.

I will probably implement searching script tags by next week, but manual override will have to wait for a while.