[node-modules] Improve deduplication efficiency by tracking ancestor dependencies #955
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
…ons the packages were not hoisted
larixer
force-pushed
the
efficient-dedupe
branch
from
791302b to
5bcfa92
Compare
larixer
force-pushed
the
efficient-dedupe
branch
from
5399b0a to
ecaa3a2
Compare
larixer
force-pushed
the
efficient-dedupe
branch
from
ecaa3a2 to
97ac7b1
Compare
|
@larixer In babel/babel#11167, the size of all the Let me know when I can test this PR! |
|
@nicolo-ribaudo Sure, I will The difference in size is certainly above what I have expected, I will test on babel repo the deduplication efficiency, since on my repo it was only 18% worser (10% due to native package installation by v2 for the cpu and os variants that are different from current OS and CPU and 8%, due to v1 and npm have bug when they quite often violate peer dependency promise in certain situations) |
|
I'm computing the With yarn 1 + lerna I'm getting Detailed sizes
EDIT: I have opened #996, which is the cause of this big size difference. |
1 task
|
@nicolo-ribaudo Currently on this PR I have: Could you give it a shot please? |
|
Somehow I get way less (644720), but all the tests are passing 👍 Also, this feels faster (I don't know if it's true), because the loading bar appears immediately instead of after a few seconds. |
|
@nicolo-ribaudo Yes, it is much faster, now the hoisting takes 1 second for babel. And I'm still going to improve that in next PRs :) |
|
@nicolo-ribaudo Oh, also the whole thing is much faster thanks to @arcanis PR: |
|
OMG I can't believe my eyes 😍 This is with an empty cache:
I guess that the |
|
🚀 |
Merged
|
@nicolo-ribaudo Sorry, this PR happened to be broken. I have run |
|
No worries, I will test the other PR when it doesn't have the "in progress" label 👍 |
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.
What's the problem this PR addresses?
The deduplication efficiency is still poor both size-wise and performance-wise in
masterHow did you fix it?
I have rewritten hoisting algorithm to hoist dependencies from the whole tree into current root node to improve performance and added ancestor dependencies tracking to improve deduplication efficiency. This has simplified algorithm in some areas (checks) and sophisticated in another areas (node cloning), but generally it is now far superior than the one used in
masterboth size-wise and performance-wise. It is not yet as perfect as it can be, but it is still a good checkpoint IMO.