Custom faster ESM to CommonJS transformer #4366
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Benchmark ResultsKitchen Sink ✅
Timings
Cold Bundles
Cached Bundles
React HackerNews ✅
Timings
Cold Bundles
Cached Bundles
AtlasKit Editor ✅
Timings
Cold Bundles
Cached Bundles
Three.js ✅
Timings
Cold BundlesNo bundle changes detected. Cached Bundles
|
wbinnssmith
reviewed
Mar 23, 2020
wbinnssmith
approved these changes
Mar 23, 2020
|
You made casts to any and used And |
Oh, what? Why is that? |
|
For no particular reason, it's just not implemented: facebook/flow#7735 |
# Conflicts: # packages/transformers/js/src/JSTransformer.js
|
@devongovett Do you still want to merge this? Should I implement the flow changes I suggested? |
|
Yeah I just haven’t had time yet, sorry! |
|
In case we still want to do this:
|
# Conflicts: # packages/core/integration-tests/test/sourcemaps.js # packages/transformers/js/package.json # yarn.lock
|
Finally had time to update this again. Made it even faster by completely removing |
mischnic
reviewed
Dec 6, 2020
mischnic
reviewed
Dec 6, 2020
mischnic
reviewed
Dec 6, 2020
|
We should also make sure that duplicate reexports works correctly (which we fixed previously in Babel's transformer): #4399 (comment) (So the snippet in that comment should work) |
devongovett
force-pushed
the
custom-esm
branch
from
52c0a7c
to
8bd84e2
Compare
mischnic
reviewed
Dec 11, 2020
|
Should we remove the parser plugin as well then? At the moment it parses only if the babel transformer doesn't run and then you get an unhelpful assertion error from the babel cjs-esm plugin |
|
Ah. Well maybe it should be the other way then? Disable the parser plugin by default, but do support it in the transformer. That way people can enable it by adding the syntax plugin to their babel config. If we didn't support transforming it, then they'd also have to use babel's modules transform but that might break some scope hoisting stuff... |
|
Ok I actually removed all parser plugins from both babel transformer and js transformer. Both dynamic import and export namespace declarations are supported by default now. Export namespace is stage 4 and supported natively in chrome, so I've added support for that in the modules transformer. I added default export support as well, but it must be enabled by adding |
# Conflicts: # yarn.lock
mischnic
reviewed
Dec 11, 2020
mischnic
approved these changes
Dec 12, 2020
There was a problem hiding this comment.
Maybe @wbinnssmith could give this a spin in case we forgot some obvious case?
wbinnssmith
suggested changes
Dec 17, 2020
# Conflicts: # packages/core/integration-tests/test/javascript.js
This was referenced Dec 21, 2020
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently, when not scope hoisting, we compile ES modules to CommonJS using
@babel/plugin-transform-modules-commonjs. I noticed while benchmarking that this appeared to be quite slow - by far the slowest thing we do in Parcel. Looking through the source code, it appears to do multiple traversals over the entire AST. I'm sure there's a reason for that, but I don't think we need to do it for our use in Parcel. In addition, using@babel/corehas some overhead over just using@babel/traversedirectly.This PR implements a new AST visitor that transforms ESM to CommonJS in a single pass over only the top-level statements in the program. Using scope information collected by babel traverse already, we can replace references without doing multiple full program traverses. On the esbuild benchmark, this improved overall build performance by ~25%!
In addition, builds may be slightly smaller because the helpers for various interop functionality are not duplicated in every file but included once in a separate asset.