Update documentation throughout the code that references RequireJS

[jessegreenberg]

Documentation about RequireJS no longer applies since we have switched to ES6 modules. For instance from phet-software-design-patterns.md

Since requireJS does not do well with circular dependencies, two types can refer to each other through the namespace.

And this comment in GrabDragInteraction

// wrap in a function because phet.joist.sim doesn't exist at RequireJS time

We should do an inspection to either update comments to reflect "compilation free mode" or remove entirely if they no longer apply.

@jonathanolson @samreid how should we do this, is this a potential chip away issue to be divvied up like phetsims/tasks#1017?

[jessegreenberg]

kite, mobius, scenery namespaces have comments like

The main 'kite' namespace object for the exported (non-Require.js) API. Used internally since it prevents Require.js issues with circular dependencies.

[samreid]

I'm also seeing that chipper/package.json still refers to requirejs.

[jonathanolson]

Sounds like we'll want to do a project-wide search and try to replace things with improved terminology.

[samreid]

That sounds like something a primary developer could take the lead on, and request help if they come across parts where it is unclear how to proceed. @ariel-phet would you like to recommend someone to take the lead on this? It could be any developer.

[ariel-phet]

@jessegreenberg is willing to take this on, and will call on other devs as needed

[zepumph]

I also think that there is a bigger discussion to be had here about the vestigial, but widely used naming of REQUIREJS_NAMESPACE formatting. You can search for usages of requirejsNamespace including that field in every package.json.

[jessegreenberg]

Right, and it is still used throughout chipper @jonathanolson do you have a recommendation for a rename of this field?

[jonathanolson]

do you have a recommendation for a rename of this field?

It accurately describes its legacy purpose, but I'm also not against other people wanting to rename it. Not attached either way.

[jessegreenberg]

I realized I am not knowledgeable enough about requireJS and how various things have changed with ES6 modules to do this. I have assembled a list of about 2 dozen questions that are of the form "should the documentation change or does the implementation need to change", and I am still not done going through everything. So I would recommend that a responsible developer for each repository go through first and review references. Here is the main list of places I am seeing them.

• chipper generally (@jonathanolson)
• chipper - can we remove requirejs from chipper package.json? (@jonathanolson)
• chipper (@samreid)
• chipper no-extend.js (@zepumph)
• aqua (@jonathanolson)
• scenery-phet (@jessegreenberg)
• phet-info (@jonathanolson)
• sherpa (specifically, usage of almond, text-2.0.12 and require-2.3.6, which look specific to requirejs) (@jonathanolson )
• installer-builder (@jbphet)
• joist (documentation, and a usage still of window.requirejs) (@jonathanolson )
• tambo (@jbphet)
• phet-io (@samreid)
• phet-io-wrappers (@chrisklus @zepumph)
• dot (todo files) (@jonathanolson)
• yotta (@jonathanolson)
• scenery (@jonathanolson)

Adding to developer meeting to check in, make sure responsible devs are correct, and assign to the few repos that are missing names.

[jonathanolson]

Should be complete, a lot of work done in 0e01ac5 removes require.js compatibility in a lot of our build files, and gets rid of a lot of vestigial stuff (e.g. global.phet.chipper). @ariel-phet A brief review by another dev would probably be good.

[chrisklus]

from 06/18/20 dev meeting: @samreid volunteered to review. thanks!

[samreid]

In Slack, @jonathanolson said:

Main review thing is basically that require/define were removed from our browser lint globals + I stripped out ways where files could be used from require.js or node.js (we only needed them in node.js now) + chipperGlobals went away

To review, I took the following steps:

• Examined the change set
• Looked for other occurrences of typeof define !== 'undefined' and found none
• Looked for other occurrences of global.define and found none
• Tested natural selection in unbuilt mode briefly
• Tested phet and phet-brand builds of natural selection briefly

No problems noted, nice work @jonathanolson, closing.