Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Set up experimental builds #17071

Merged
merged 2 commits into from
Oct 14, 2019

Conversation

acdlite
Copy link
Collaborator

@acdlite acdlite commented Oct 11, 2019

The experimental builds are packaged exactly like builds in the stable release channel: same file structure, entry points, and npm package names. The goal is to match what will eventually be released in stable as closely as possible, but with additional features turned on.

Versioning and Releasing

The experimental builds will be published to the same registry and package names as the stable ones. However, they will be versioned using a separate scheme. Instead of semver versions, experimental releases will receive arbitrary version strings based on their content hashes. The motivation is to thwart attempts to use a version range to match against future experimental releases. The only way to install or depend on an experimental release is to refer to the specific version number.

Building

I did not use the existing feature flag infra to configure the experimental builds. The reason is because feature flags are designed to configure a single package. They're not designed to generate multiple forks of the same package; for each set of feature flags, you must create a separate package configuration.

Instead, I've added a new build dimension called the release channel. By default, builds use the stable channel. There's also an experimental release channel. We have the option to add more in the future.

There are now two dimensions per artifact: build type (production, development, or profiling), and release channel (stable or experimental). These are separate dimensions because they are combinatorial: there are stable and experimental production builds, stable and experimental development builds, and so on.

You can add something to an experimental build by gating on __EXPERIMENTAL__, similar to how we use __DEV__. Anything inside these branches will be excluded from the stable builds. This gives us a low effort way to add experimental behavior in any package without setting up feature flags or configuring a new package.

@sizebot
Copy link

sizebot commented Oct 11, 2019

ReactDOM: size: 0.0%, gzip: -0.0%

Details of bundled changes.

Comparing: 75955bf...9977c2c

react-dom

File Filesize Diff Gzip Diff Prev Size Current Size Prev Gzip Current Gzip ENV
react-dom.profiling.min.js -0.0% -0.0% 119.46 KB 119.41 KB 37.72 KB 37.71 KB NODE_PROFILING
react-dom-server.browser.development.js 0.0% 0.0% 140.06 KB 140.06 KB 36.8 KB 36.8 KB UMD_DEV
ReactDOM-dev.js -0.1% -0.0% 968.85 KB 968.24 KB 215.29 KB 215.23 KB FB_WWW_DEV
react-dom-server.browser.production.min.js 0.0% -0.0% 19.84 KB 19.84 KB 7.37 KB 7.37 KB UMD_PROD
react-dom-unstable-fizz.browser.development.js 0.0% -0.1% 3.78 KB 3.78 KB 1.53 KB 1.53 KB UMD_DEV
react-dom-test-utils.production.min.js 0.0% 0.0% 11.17 KB 11.17 KB 4.14 KB 4.14 KB UMD_PROD
react-dom-unstable-fizz.browser.development.js 0.0% -0.1% 3.61 KB 3.61 KB 1.48 KB 1.48 KB NODE_DEV
react-dom-test-utils.production.min.js 0.0% 0.0% 10.94 KB 10.94 KB 4.08 KB 4.08 KB NODE_PROD
react-dom.development.js -0.0% -0.0% 946.68 KB 946.25 KB 214.41 KB 214.37 KB UMD_DEV
react-dom.production.min.js -0.0% -0.0% 115.69 KB 115.65 KB 37.24 KB 37.23 KB UMD_PROD
react-dom.profiling.min.js -0.0% -0.0% 119.22 KB 119.17 KB 38.35 KB 38.35 KB UMD_PROFILING
react-dom.development.js -0.0% -0.0% 940.75 KB 940.31 KB 212.84 KB 212.8 KB NODE_DEV
react-dom-server.node.development.js 0.0% -0.0% 137.1 KB 137.1 KB 36.02 KB 36.02 KB NODE_DEV
react-dom.production.min.js -0.0% -0.0% 115.81 KB 115.77 KB 36.7 KB 36.69 KB NODE_PROD
react-dom-server.node.production.min.js 0.0% -0.0% 20.17 KB 20.17 KB 7.5 KB 7.5 KB NODE_PROD
ReactDOM-prod.js -0.0% -0.0% 398.09 KB 398.04 KB 72.46 KB 72.45 KB FB_WWW_PROD
ReactDOM-profiling.js -0.0% -0.0% 398.88 KB 398.83 KB 72.97 KB 72.96 KB FB_WWW_PROFILING
react-dom-server.browser.development.js 0.0% -0.0% 135.99 KB 135.99 KB 35.8 KB 35.79 KB NODE_DEV
react-dom-server.browser.production.min.js 0.0% -0.0% 19.76 KB 19.76 KB 7.35 KB 7.35 KB NODE_PROD
ReactDOMServer-prod.js 0.0% 0.0% 48.5 KB 48.5 KB 11.08 KB 11.08 KB FB_WWW_PROD
react-dom-unstable-fizz.node.development.js 0.0% -0.1% 3.87 KB 3.87 KB 1.51 KB 1.51 KB NODE_DEV
react-dom-unstable-native-dependencies.production.min.js 0.0% -0.0% 10.46 KB 10.46 KB 3.57 KB 3.57 KB NODE_PROD
react-dom-unstable-fizz.node.production.min.js 0.0% -0.1% 1.1 KB 1.1 KB 668 B 667 B NODE_PROD

Generated by 🚫 dangerJS against 9977c2c

The method names don't get stripped out of the production bundles
because they are passed as arguments to the error decoder.

Let's just always use the unprefixed APIs in the messages.
@acdlite acdlite force-pushed the experimental-release-channel branch from 7687b9e to 8c4854d Compare October 12, 2019 00:34
The experimental builds are packaged exactly like builds in the stable
release channel: same file structure, entry points, and npm package
names. The goal is to match what will eventually be released in stable
as closely as possible, but with additional features turned on.

Versioning and Releasing
------------------------

The experimental builds will be published to the same registry and
package names as the stable ones. However, they will be versioned using
a separate scheme. Instead of semver versions, experimental releases
will receive arbitrary version strings based on their content hashes.
The motivation is to thwart attempts to use a version range to match
against future experimental releases. The only way to install or depend
on an experimental release is to refer to the specific version number.

Building
--------

I did not use the existing feature flag infra to configure the
experimental builds. The reason is because feature flags are designed
to configure a single package. They're not designed to generate multiple
forks of the same package; for each set of feature flags, you must
create a separate package configuration.

Instead, I've added a new build dimension called the **release
channel**. By default, builds use the **stable** channel. There's
also an **experimental** release channel. We have the option to add more
in the future.

There are now two dimensions per artifact: build type (production,
development, or profiling), and release channel (stable or
experimental). These are separate dimensions because they are
combinatorial: there are stable and experimental production builds,
stable and experimental developmenet builds, and so on.

You can add something to an experimental build by gating on
`__EXPERIMENTAL__`, similar to how we use `__DEV__`. Anything inside
these branches will be excluded from the stable builds.
This gives us a low effort way to add experimental behavior in any
package without setting up feature flags or configuring a new package.
@acdlite acdlite force-pushed the experimental-release-channel branch 8 times, most recently from b1b697e to f17702d Compare October 13, 2019 01:18
@acdlite acdlite marked this pull request as ready for review October 13, 2019 01:26
@acdlite acdlite force-pushed the experimental-release-channel branch from f17702d to 9977c2c Compare October 13, 2019 01:36
Copy link
Contributor

@bvaughn bvaughn left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No release script changes. Will those be follow up PRs?

@@ -74,6 +100,18 @@ jobs:
- *run_yarn
- run: yarn test --maxWorkers=2

test_source_experimental:
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I assume that since we're using parallel containers, this won't actually increase the duration of every CI job? That would not be great.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The workflows run in parallel, yeah

@@ -682,9 +671,8 @@ function warnIfReactDOMContainerInDEV(container) {
if (__DEV__) {
warningWithoutStack(
!container._reactRootContainer,
'You are calling ReactDOM.%s() on a container that was previously ' +
'You are calling ReactDOM.createRoot() on a container that was previously ' +
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I would not expect all fo these conditional names to change, given that we run this test against not-experimental builds as well. Why did they?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

See the first commit.

The method names don't get stripped out of the production bundles
because they are passed as arguments to the error decoder.

Slightly unrelated but I did it this way instead of checking __EXPERIMENTAL__ in the tests. We're probably going to remove the prefixed versions from the stable releases, anyway.

@acdlite
Copy link
Collaborator Author

acdlite commented Oct 14, 2019

@bvaughn I'll do release script changes in a follow-up. The existing publish script should actually work if you pass the build number in manually. The main thing I want to add is a check to the prepare-stable script that prevents you from publishing an experimental build as stable.

@acdlite acdlite merged commit d364d85 into facebook:master Oct 14, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants