Uh oh!
There was an error while loading. Please reload this page.
errorbase: optimize errors.Is - #153
Merged
Merged
Conversation
|
This is a benchmark created by @rafi. It demonstrates `errors.Is` is very inefficient when the reference error does not match the input error.
Previously, `errors.Is` was very inefficient if the reference error did not match any errors in the chain. There were two significant sources of inefficiency: 1. The code would pessimistically construct the error mark for every error in the input error chain. This is O(chain_length^2). This was a lot of unnecessary allocations and caused the runtime to be O(n^2) in the average case instead of O(n). 2. The code compared the `Error()` message before comparing the types. It is possible to compare error types with zero allocations whereas computing the message often requires an allocation.
jeffswensonforce-pushed
the
jeffswenson-optimize-errors-is
branch
from
October 7, 2025 15:35
3ff6ee5 to
3a21e3dComparejeffswenson
commented
Oct 7, 2025
| if m1.msg != m2.msg { | ||
| return false | ||
| } | ||
| if len(m1.types) != len(m2.types) { |
ContributorAuthor
There was a problem hiding this comment.
This is a bug that is pretty unlikely to occur in the original implementation that checks error message before types, but its somewhat likely now that we check the types first.
jeffswenson
commented
Oct 7, 2025
| return true | ||
| } | ||
| for _, reference := range references { | ||
| if Is(err, reference) { |
ContributorAuthor
There was a problem hiding this comment.
I think the reason we didn't use this implementation originally is Is was pretty slow. So it was better to use the quick check before the slow mark check. Now that Is is efficient, we don't need to duplicate the implementations.
Contributor
There was a problem hiding this comment.
Thanks for the consolidation.
ContributorAuthor
|
jeffswenson
marked this pull request as ready for review
October 7, 2025 15:59
dhartunian
approved these changes
Oct 10, 2025
dhartunian
left a comment
Contributor
There was a problem hiding this comment.
@dhartunian reviewed 1 of 3 files at r1, all commit messages.
Reviewable status: 1 of 3 files reviewed, 2 unresolved discussions (waiting on @jeffswenson)
| return true | ||
| } | ||
| for _, reference := range references { | ||
| if Is(err, reference) { |
Contributor
There was a problem hiding this comment.
Thanks for the consolidation.
jeffswenson
commented
Oct 10, 2025
ContributorAuthor
Thanks for the review! |
Uh oh!
There was an error while loading. Please reload this page.
craigBot
pushed a commit
to cockroachdb/cockroach
that referenced
this pull request
Oct 10, 2025
![craig[bot]](https://github.laxmiang.work.gd/r/github.githubassets.com/images/gravatars/gravatar-user-420.png){kind=avatar}
155254: go.mod: update cockroachdb/errors package r=jeffswenson a=jeffswenson Update the errors package to get the performance optimization from cockroachdb/errors#153. Release note: none Informs: #154125 Co-authored-by: Jeff Swenson <jeffswenson@betterthannull.com>
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
Sign up for freeto 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.
benchmark: add a benchmark for the errors package
This is a benchmark created by @rafiss. It demonstrates
errors.Isis very inefficient when the reference error does not match the
input error.
errbase: optimize errors.Is
Previously,
errors.Iswas very inefficient if the reference error didnot match any errors in the chain.
There were two significant sources of inefficiency:
error in the input error chain. This is O(chain_length^2). This was
a lot of unnecessary allocations and caused the runtime to be O(n^2)
in the average case instead of O(n).
Error()message before comparing the types.It is possible to compare error types with zero allocations whereas
computing the message often requires an allocation.
refs: cockroachdb/cockroach#154555
This change is