JavaScript chat summary for Sept. 26th

Below is a summary of the discussion from this week’s JavaScript chat (agenda, Slack transcript):

This week’s meeting was focused around the role of a JavaScript framework in WordPress. The meeting was attended by many guests to the conversation, speaking on behalf of their respective frameworks or web standards:

Google (Web Components / Chrome / Polymer / AMP) – Justin Fagnani (Chrome / Polymer), Alex Russell (Chrome), Wendy Ginsberg (Polymer), Paul Bakaus (AMP), Alberto Medina (AMP)
Facebook (React) – Sophie Alpert, Dan Abramov, Dominic Gannaway
Vue – Evan You

 

What role should a framework play in a WordPress developer’s workflow, and in which contexts? For what purposes do we rely on a framework?

@youknowriad started off the discussion by offering three things as important for the role of the framework

  • Build UI and handle dom updates
  • Build reusable “components”. Reused in Core and potentially for plugins
  • Ability to extend the UI (using the same framework or not)

@dmsnell added it should provide a well-defined interface for isolated components to interact with so that people can have the freedom to build pieces how they want.
@mrahmadawais added that A framework should have a great development community, great documentation
@herregroen added [it]… should make it as easy as possible for components to communicate through more than simply DOM updates.
@peterbooker raised the importance of ease of access, for  … the [WordPress] development community… used to PHP, HTML, CSS, etc.
@mrmaniac (Paul Bakaus) added the best level of separation is if the framework is used to build the core, but isn’t exposed as API to block builders. This gives one the choice to replace the underlying foundation whenever necessary.
@mcsf said that for a project of the nature of WordPress, a framework should be relatively self-effacing, and its interface surface should be minimized and as agnostic as possible, mimicking the evolving JS specs.
@westonruter pointed out that If the underlying framework is being loaded on the page anyway, it should be available for plugin authors to utilize since otherwise they’d have to each include their own bundled frameworks… if core is shipping with one framework, then it would become the de facto framework by virtue of it being available out of the box.
@matias disagreed a bit with the idea that whatever core uses… is going to be the de facto standard for plugin development. The actual framework here, in general terms, is going to be what WordPress exposes and the APIs.
@flixos90 said I think core shipping a default framework and thus somewhat defining a standard to use is necessary to prevent compatibility issues by people using whatever framework (version) they like.
@mrmaniac (Paul Bakaus) suggested maybe there’s a middle ground, where, if you happen to use the same framework as the core, you benefit from the bundling, but if the core ever switches to another one, the framework gets lazy loaded in, and we create a mechanism that does this reliably.
@evanyou added I believe it’s important (and technically feasible) to separate “which framework to use for core” and “which framework community devs use for extensions”

 

The discussion moved on to interoperability and specifically ideas for providing a generic interface which can adapt to future change.

As background, @aduth put together two pull requests in Gutenberg on interoperability, one exploring combination of generic data structure / DOM access, and another using web components as the common interface:
https://github.com/WordPress/gutenberg/pull/2463 – Framework-agnostic block interoperability (Vanilla, Vue)
https://github.com/WordPress/gutenberg/pull/2791 – Support block `edit` defined as tag name for web components interoperability

@slightlyoff sees two major classes of interop: ease of development of leaf components and ease of integration of leaves into a larger whole.
@Justin Asked does anyone think that the custom elements lifecycle is not sufficient? At least for a foundation?

 

Discussion continued, largely focused around web components – how they could be leveraged for interoperability and how well they are supported in each framework.

@sophiebits (Sophie Alpert) added that in React, we have some web component support but haven’t made it a large priority since use cases have seemed slim in the past… but we do have some support for them nonetheless and I’m happy to entertain adding more, either now or in the future
@evanyou commented on the high level I think frameworks like React/Vue provides what is not really addressed in web components: efficient and declarative DOM updates reacting to state changes.. this is also why Polymer exists on top of WC
@Justin said that is ok: Web Components don’t need to answer _every_ question about components: they’re a foundation and interop layer
@trueadm added One major advantage of abstract components over web components is the fact they don’t live in the DOM. This gives several advantages – such as not being a global register custom element and being able to easily change/alter to an underlying layer as times change
@gaearon (Dan Abramov) added I’m not convinced custom elements are best interop layer. I would prefer plain JS hooks. On top of that, you can always add everything including them.

 

Discussion continued to patterns for pluggable interfaces, including the current use of react-slot-fill in Gutenberg and comparing it to <slot>. The meeting wrapped up with some discussion of how the choice of a framework for core impacts the wider community.

@gaearon (Dan Abramov) said I don’t really know WordPress well, so it’s hard for me to say whether [React is] a great fit for the use case or not… I think in general people have strong opinions about, for example, templating vs expressiveness, and I don’t feel like forcing React upon everyone is the best way. 

To which @evanyou responded I also feel the same way – forcing a single framework on everyone, regardless of which one, is IMO not a good idea because it is bound to alienate the group of devs who are not into that framework, and imposes a bigger long term stability risk.

 

Thanks to everyone who participated – especially to our guests from outside of the WordPress community: we welcome further collaboration with you in the future.

It is very clear (and refreshing to see!) from the conversation that the developers working on React, Vue and Polymer all care deeply about pushing the open web forward in a collaborative fashion. While we may not always agree on approaches, syntax or terminology, we do all agree that developers should be free to choose the platform and tools they find to be the best fit for their requirements. For WordPress core, this means that regardless of the framework we choose to use internally on projects like Gutenberg, we should ensure that the API we expose externally remains framework agnostic.

 

Add your voice and assistance!
We can use help especially testing @aduth‘s pull requests exploring interoperability: https://github.com/WordPress/gutenberg/pull/2463 and https://github.com/WordPress/gutenberg/pull/2791. Please read thru these PRs and test them out – see what works and what is missing, and give us your feedback.

Please join us next week at our regularly scheduled time Tuesday, October 3, 2017 at 13:00 UTC for the next JavaScript chat.

#core-js, #javascript, #summaries