fix(pretty-format): Crash in ES5 environments

[eps1lon]

Summary

Fixes a crash in ES5 environments (or other environments not implementing Object.entries)

Test plan

Edit: Only test I have is using prett-format@24 vs pretty-format@25

Don't have any (because the dependency is inlined) unless you somehow build canaries that I can install. I know that ansi-styles@3 does not use Object.entries while ansi-styles@4 does.

[eps1lon]

Only test I have is using prett-format@24 vs pretty-format@25

[eps1lon]

Wanted to point out that IE 11 is just a specific environment. It crashes on any environment that only supports ES5. Since this packages has a dedicated build for those environments (build-es5) it should not use dependencies that make use of Object.entries.

[thymikee]

Shouldn't we tinker our build config, so it compiles away Object.entries instead?

[eps1lon]

Shouldn't we tinker our build config, so it compiles away Object.entries instead?

That's a possibility. Though you would need to compile node_modules/ in this case.

[SimenB]

We compile (and bundle) for browser, so I think it makes more sense to transpile than downgrade

[eps1lon]

We compile (and bundle) for browser, so I think it makes more sense to transpile than downgrade

I'll take a look at the build setup. Thanks for the feedback!

[SimenB]

Sorry, I was supposed to link this but got distracted: https://github.com/facebook/jest/blob/master/scripts/browserBuild.js. It's the browser build script, theoretically targeting ES5

[eps1lon]

No worries I found it already.

We have to keep in mind though that using transform-runtime overtranspiles by a lot. Though there are currently plans in babel to improve this issue:babel/babel#10008.

I don't know if any of the affected packages are used in user-facing bundles. If so then going from 102KB unminified to 537KB is not acceptable. I personally don't care about the size of this package since I only use it when running tests in the browser. Bundle size is no major concern for these cases.

[codecov-io]

Codecov Report

Merging #9635 into master will increase coverage by 0.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #9635      +/-   ##
==========================================
+ Coverage   65.09%   65.11%   +0.01%     
==========================================
  Files         287      287              
  Lines       12144    12144              
  Branches     3009     3009              
==========================================
+ Hits         7905     7907       +2     
+ Misses       3604     3603       -1     
+ Partials      635      634       -1     

Impacted Files	Coverage Δ
packages/expect/src/utils.ts	96.22% <0.00%> (+1.25%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e3aa106...4f1a45f. Read the comment docs.

[SimenB]

Yikes. Is it possible to target just the offending package?

[eps1lon]

I already did. Transform-runtime with core-js is just that overzealous. I can take a closer look at the added bytes and see how a prod bundle would look like.

[SimenB]

We'll be splitting out the es5 builds in Jest 26 (so you'll install pretty-format-browser or some such) - at that point I guess we don't have to care about the bundle size anymore. We'll still build and publish them from here, but they won't be installed by default, and you have to alias them in your bundler (or drop an unpkg.com link in a script tag with it, I guess)

[thymikee]

@SimenB es5 builds are now removed, but we didn't provide them as separate packages just yet.

[SimenB]

Right... We can create those packages, I guess. index.ts is just export * from 'expect' or whatever and reinstate the old browser build. Anyone up for it?

Telling people to transpile themselves sounds good, but I think being able to use these packages via a script tag is useful

[eps1lon]

No longer interested in making this work.

[github-actions]

This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.
Please note this issue tracker is not a help forum. We recommend using StackOverflow or our discord channel for questions.