fix: repair all article solutions that deterministically fail on the NC judge - #6051
Open
neetcode-gh wants to merge 5 commits into
Open
fix: repair all article solutions that deterministically fail on the NC judge#6051neetcode-gh wants to merge 5 commits into
neetcode-gh wants to merge 5 commits into
Conversation
Article Go snippets are spliced into a judge driver that already declares
`package main` and its own import block, so a snippet can neither add an
import nor pull in a third-party module. Every Go solution that used
github.com/emirpasic/gods therefore failed to compile on the judge.
Rewrites all 28 affected tabs to the standard library only:
- priorityqueue.Queue -> container/heap with a local sort.Interface type
- linkedliststack -> plain slice used as a stack
- arrayqueue /
linkedlistqueue -> plain slice used as a queue
- redblacktree /
treemap -> count map + sorted slice, binary searched with
sort.Search / sort.SearchInts
The drivers already provide container/heap, sort, math and friends
ambiently, so none of the new snippets declare imports. Published
structure and naming are kept as close to the originals as possible.
Also removes a stray `import "container/heap"` from the
design-a-leaderboard "Heap for top-K" tab, which failed to compile for the
same reason (redeclared / imported and not used against the driver block).
Behaviour is preserved throughout. Two latent bugs disappear as a side
effect: hand-of-straights now bounds-checks the heap before peeking, and
design-a-leaderboard Reset no longer decrements the count for score 0 when
the player is absent.
Every rewritten tab was run against the production judge
(test-problem-article-solutions.ts --lang go); all 29 pass.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The Binary Exponentiation (Iterative) JavaScript snippet was the only one of the article's nine languages that ran the bit loop on a value that does not fit in a signed 32-bit integer. `Math.abs(-2147483648)` correctly produces `2147483648` (JS numbers are doubles), but `&` and `>>=` apply ToInt32 to their operands first. So `power & 1` evaluates `-2147483648 & 1` -> `0` (the multiply is skipped) and `power >>= 1` evaluates `-2147483648 >> 1` -> `-1073741824`, which fails the `power > 0` guard. The loop exits after a single iteration with `res` still 1. On LeetCode's own published test case `x = 2.00000, n = -2147483648` (expected `0.00000`) the snippet returned `1`. Every other language already widens to 64 bits before the loop -- `Math.abs((long)n)` in Java/C#, `abs((long)n)` in C++, `(n as i64).abs()` in Rust, `n.toLong()` in Kotlin, 64-bit `Int`/`int` in Swift/Go, and arbitrary precision in Python -- so none of them were affected. The article's own "Integer Overflow When Negating n" pitfall section documents exactly this trap. Fixed by using arithmetic instead of bitwise operations, which keeps `power` a full-precision double: `power % 2 === 1` and `Math.floor(power / 2)`. The recursive JavaScript solution already did this and was correct. Also extended the pitfall section with the JavaScript-specific form of the trap. Verified against the NeetCode judge: all 9 languages pass both binary exponentiation approaches, including an `x = 2.00000, n = -2147483648` case that the previous JavaScript code failed with Wrong Answer. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…/judge-article-failures
…/judge-article-failures
- time-based-key-value-store Brute Force (all 9 languages): a value set at
timestamp 0 was unreachable — the 'seen' sentinel collided with a legal
timestamp. Use -1 (timestamps are >= 0 on the site). Also make the C# tab
take the max like every other tab instead of relying on insertion order.
- palindrome-partitioning rust Recursion + Backtracking (DP): 'i + l - 2'
underflows usize when l = 1, i = 0, panicking at runtime on every input.
Compute j = i + l - 1 and guard with l <= 2.
- copy-linked-list-with-random-pointer rust (all 5 tabs): replace the
index-based placeholder stubs ('LeetCode does not support Rust') with real
implementations against Option<Rc<RefCell<Node>>>, including both O(1)
extra-space variants.
- kth-smallest-integer-in-bst: kotlin Morris assigned through nullable
receivers and did not compile — rewritten null-safely. The go and kotlin
Brute Force tabs did an inorder collect (no sort), a different algorithm
from the python/java tabs under the same heading — now collect + sort.
- longest-substring-without-duplicates: the swift Brute Force snippet was
fenced as kotlin, so kotlin appeared twice and swift had no coverage.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Owner
Author
|
@coderabbitai review |
Owner
Author
|
@greptileai review |
Owner
Author
|
bugbot run |
|
Skipping Bugbot: Bugbot is disabled for this repository. Visit the Bugbot dashboard to update your settings. |
Owner
Author
|
@coderabbitai review |
Owner
Author
|
@greptileai review |
Owner
Author
|
Review loop tally: CodeRabbit and Greptile were triggered twice (~18:48Z and ~18:36Z cycles) and stayed silent; neither app has ever reviewed a PR in this repo (checked #6039/#6044 history) and Bugbot is disabled here, so bot review is not available on this repo. Verification instead ran against the live judge: 20 problems × 2 runs each via test-problem-article-solutions.ts — every previously-failing deterministic approach×language combo now passes; remaining failures are the pre-existing Brute Force time-limit tier only. Ready for review/merge — merging this PR also marks #6039 and #6044 as merged (their branches are included). |
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 free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Summary
A 4-runs-per-combo soak of every article approach × language across 72 problems against the live judge surfaced a set of deterministic (4/4) failures. This PR fixes all of them on the article side. It includes #6039 (Go stdlib heap/stack rewrites) and #6044 (pow-x-n JS INT_MIN fix) as merges — merging this PR marks both as merged — and adds one commit on top:
seensentinel of0collided with a legal timestamp (site contract is0 <= timestamp <= 10^7). Sentinel is now-1. The C# tab also now takes the max qualifying timestamp like every other tab instead of relying on dictionary insertion order.RecursionandBacktracking (DP): the palindrome-table fill computedi + l - 2, which underflowsusizeatl = 1, i = 0and panics at runtime on every input (the judge saw empty output). Now computesj = i + l - 1guarded byl <= 2.Option<Rc<RefCell<Node>>>type, including both O(1) extra-space interleaving variants.Morris Traversal: assigned through nullable receivers and did not compile; rewritten null-safely. Go/KotlinBrute Force: these tabs did an inorder collect with no sort — a different algorithm from the python/java tabs under the same heading; now collect + sort like the rest.```kotlin, so kotlin appeared twice and swift had zero coverage.Verification
Every touched problem was run twice (three times for time-based-key-value-store) through
test-problem-article-solutions.tsagainst the live judge — 20 problems, 41 runs total. All previously-failing deterministic combos now pass. The only remaining failures are the known pre-existing time-limit tier (Brute Force TLEs) and the kth-largest Quick Select python stress-case issue tracked separately on the site side.Companion PR on the site repo enriches four judge drivers (swift
Node: Hashable, kotlin/swift 3-argTreeNode, swift 2-argListNode) that three of these fixes rely on.🤖 Generated with Claude Code