Make bindings generation dramatically faster #292
Conversation
This file contains hidden or 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
It's always set to "bindings"
This simplifies the code flow a good bit, and I hope to make this fast enough that it doesn't matter.
Just call it processChildren now.
Pass tu through instead of doing an expensive re-parse a bunch.
It's expensive!
I want to add more precomputed stuff here, let's introduce a helper object for it first so we don't add a zillion args to every function.
This doesn't need to be created every time, and the search doesn't need to be linear every time.
Much, much faster.
|
|
This is amazing. Tried it out locally 🤯 @donalffons plz merge |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
This series of changes makes generation of the bindings dramatically faster, by fixing a number of issues around re-parsing things, re-doing expensive work, doing O(n) scans when O(1) lookups are possible, etc. It also slightly simplifies some of the code by removing the multiprocessing layer for bindings generation. Despite now being single-threaded, the total time for bindings generation on my 16-core machine drops from 8 minutes to nearly instant. (I am not sure what percent speedup that is, but it's huge!)