The new block API

[vlad-zhukov]

I've proposed this API almost a year ago in #125 and I believe everyone is fine with it by now. TLDR:

// Current:

function postConfig (context, util) {
  const ruleConfig = Object.assign({
    test: /\.(js|jsx)$/,
    exclude: /node_modules/,
    use: [
      { loader: 'babel-loader', options: context.babel }
    ]
  }, context.match)

  return util.addLoader(ruleConfig)
}

// This PR:

function postConfig (context) {
  const ruleConfig = Object.assign({
    test: /\.(js|jsx)$/,
    exclude: /node_modules/,
    use: [
      { loader: 'babel-loader', options: context.babel }
    ]
  }, context.match)

  return context.addLoader(ruleConfig)
}

The #259 is different because it merges context and utils into an object:

function postConfig ({context, addLoader}) {
  // ...
}

cc @andywer @zcei @dmitmel @jvanbruegge

[jvanbruegge]

Not sure what this is solving actually. AFAICT 2.0 allows old blocks to be used with more or less no change at all, so not sure if this is worth a breaking change

[zcei]

So #259 and this PR are two competing approaches?

Same problem as @jvanbruegge, the mentioned discussion is too long and I don't recall all the details. Would you mind a small explanation what the problems are with the current API & what the up & downsides of the two competing PRs are?

Otherwise we can't make a profound decision, and I think some visibility helps when we later come back here (as it's linked in the changelog)

[vlad-zhukov]

Good point! However I'm unsure v1 blocks will work with v2 out of the box w/o any changes. Webpack 4 made breaking changes for loaders and in order to use old blocks with v2 a developer would need to update dependencies at the very least. Making another single line change shouldn't be a big of a problem, is it?

The reason for the change is purely stylistic. Some people (including me) would like to be it that way. There is also a rare case when a block doesn't use the context but needs a util function: (_, util) => util.addLoader({}).

There are no hard feelings about the change. How often do people develop new blocks anyway? 🙃

[andywer]

I think there was even a third option: Don't pass the util stuff (addLoader & friends) into the block at all, but rather make the blocks import them from a separate package.

(Sorry for the late reply!)

[vlad-zhukov]

@andywer Was it even considered as an option? 🙉
Right now I'm very close to leaving the API as it is for now. Maybe next year...

[andywer]

@vlad-zhukov Not sure how serious the efforts in that direction where, but I remember that one, too. And I totally get how you are feeling right now; this API discussion is a hornet's nest... 🙈

At some point I just stopped caring about it this API issue too much. Bottom line: No matter what we do here, the average webpack-blocks user will neither notice nor care 😜

[vlad-zhukov]

@andywer I don't care much about this change as I pointed above. I am fine with the discussion and I'm glad we've got people in our little community who also care about the project and can spot potential problems I wasn't aware of! 🍪

This is almost internal API and is no even near to day-to-day usage (more like year-to-year, huh 😁). As I said, the change is purely stylistic and if there are even little objections the change should no go further.

[dmitmel]

@andywer Oh, yeah! So we can move block utils into a separate module, and then instead of passing actual utilities to the block creators, we can pass a slightly modified version of them that print deprecation warnings.

[andywer]

@dmitmel No need to be snappish. I am just trying to recall the discussion at the time. And yes, I had my reservations as well! Nevertheless it doesn't hurt to gather the whole picture.

[dmitmel]

@andywer Ok, but I'm just trying to do something because I wasn't contributing to webpack-blocks for a long time.

[andywer]

@dmitmel Trust me, your efforts are highly appreciated! It's just a tricky change that requires quite a lot of communication. If you want, we can all have a chat tomorrow on gitter.im about the open issues and maybe we can find a less controversial issue that you can help with 🙂

But enough of that now. All non-PR-related talk moves to gitter now.

[andywer]

I'm sorry, but I don't see this happening at the moment. Closing for now.

Feel free to bring it up again 😉