perf: use libxml2 patch for faster xpath context init #3389
+329
−148
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
Revert the pieces of 969bea9 related to re-use of xpath context objects. Leave in these pieces of that commit: - ability to de-register variables and namespaces - XPathContext#node= and backfill explicit test coverage for these features.
See extended discussion at #3378 Benchmark comparing this commit against v1.17.x ("main"): Comparison: large: main: 3910.6 i/s large: patched: 3759.6 i/s - same-ish: difference falls within error Comparison: small: patched: 242901.7 i/s small: main: 127486.0 i/s - 1.91x slower I think we could get greater performance gains by re-using XPathContext objects, but only at the cost of a significant amount of additional complexity, since in order to properly support recursive XPath evaluation, Nokogiri would have to push and pop "stack frames" containing: - internal state contextSize and proximityPosition - registered namespaces - registered variables - function lookup handler That feels like a lot of code for a small win. Comparatively, pulling in this upstream patch is still a 2x speedup for zero additional complexity.
flavorjones
force-pushed
the
flavorjones-use-libxml2-patch-for-faster-xpath-context-init
branch
from
fcc4298 to
3f20dd9
Compare
flavorjones
changed the title
flavorjones
deleted the
flavorjones-use-libxml2-patch-for-faster-xpath-context-init
branch
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Outdated suggestions cannot be applied.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 problem is this PR intended to solve?
After chatting with @nwellnhof in #3378, he went ahead and made an performance-focused change upstream in libxml2/gnome@bf5fcf6e that I'm pulling in here to the vendored library.
Benchmark comparing this PR against v1.17.x ("main"):
We could get greater performance gains by re-using
XPathContextobjects, but only at the cost of a significant amount of additional complexity, since in order to properly support recursive XPath evaluation, Nokogiri would have to push and pop "stack frames" containing:That feels like a lot of code for a small win.
Comparatively, pulling in this upstream patch is still a ~2x speedup with zero additional complexity. That's a win.
Have you included adequate test coverage?
N/A
Does this change affect the behavior of either the C or the Java implementations?
Performance improvement impacting CRuby only.