Skip to content

Repository navigation

chore(deps): update pnpm to v11.18.0 - #1540

Merged
renovate[bot] merged 1 commit into
mainfrom
renovate/pnpm-11.x
Aug 2, 2026
Merged

chore(deps): update pnpm to v11.18.0#1540
renovate[bot] merged 1 commit into
mainfrom
renovate/pnpm-11.x

Conversation

@renovate

@renovate renovate Bot commented Jul 29, 2026 •
edited
Loading

Copy link
Copy Markdown
Contributor

ℹ️ Note

This PR body was truncated due to platform limits.

This PR contains the following updates:

Package Change Age Confidence
pnpm (source) 11.6.0 → 11.18.0 age confidence

Release Notes

pnpm/pnpm (pnpm)

v11.18.0: pnpm 11.18

Compare Source

Minor Changes

  • Fixed an installed optional dependency being left without one of its own required dependencies. When a package reached through optionalDependencies is installable on the current system but one of its regular dependencies is not, a lockfile-based install skipped that dependency and installed the parent anyway, so importing the parent failed with MODULE_NOT_FOUND. The dependency is now installed, and an install-check warning reports the incompatibility. A dependency is still only skipped when every path to it is optional, or when the package that pulls it in was itself skipped #​13286.

  • pnpm setup now appends PNPM_HOME and the global bin directory to the GitHub Actions environment files (GITHUB_ENV and GITHUB_PATH), so later steps in the same job can run pnpm add --global and other global commands #​9191.

  • Added support for publishConfig.name, which publishes a package under a different name than the one its manifest carries in the workspace. It is for a project whose published name is already taken by a sibling project, which otherwise has to be renamed by a build step just before publishing. Only the published artifact is renamed — dependents, pnpm-lock.yaml, and release tooling keep addressing the project by its manifest name — and the new name reaches the packed manifest, the tarball filename, and everything that addresses the package at the registry: the already-published check of pnpm publish -r, its registry selection, and the release-planning probes of pnpm change status and pnpm version -r #​13345.

  • pnpm self-update no longer takes any instruction from the project it is run in:

    • pnpm is fetched through the same trusted registry and auth configuration used when switching pnpm versions, so a project .npmrc or pnpm-workspace.yaml can no longer redirect the download or attach credentials to it, and the project's default .pnpmfile.(c|m)js is no longer loaded. Pnpmfiles from trusted sources (the pnpmfile setting, the global pnpmfile, config dependencies) still apply.
    • The minimumReleaseAge settings in pnpm-workspace.yaml no longer affect self-update. They still govern the project's own dependencies; for self-update the cooldown now comes from the built-in default, your global config, a PNPM_CONFIG_* environment variable, or a command-line flag. This fixes self-update failing inside a workspace that raises the cutoff while succeeding everywhere else, and stops a repository from either waiving the cooldown or keeping you on an outdated pnpm by raising it.
    • The same applies to the trustPolicy settings and to ci: a project can no longer weaken the trust check that guards the pnpm download, nor re-enable the confirmation prompt that a CI run suppresses.

    When self-update refuses a version that is younger than the cutoff, an interactive run now offers to update anyway; non-interactive runs still fail. CI never prompts, even on a runner that attaches a TTY.

Patch Changes

  • Fixed pnpm licenses list to report every version when the same package is installed under multiple aliases pnpm/pnpm#13438.

  • Sort pnpm dedupe --check snapshot changes for stable output across pnpm implementations.

  • Strip Unicode formatting characters from registry- and manifest-derived terminal output.

  • Speed up installs after compatible catalog or direct dependency range changes by retaining the locked version without resolving the dependency graph again.

  • Speed up installs after safe override changes by reusing unambiguous compatible dependency resolutions, pruning obsolete dependencies, applying independent replacements and removals together, and handling parent-scoped "-" overrides without full lockfile resolution.

  • Installing a local file: directory dependency with the global virtual store enabled no longer fails with TypeError: Cannot read properties of undefined (reading 'split') #​13335.

    Local directory dependencies — file: directories and injected workspace packages — now get a global-virtual-store slot of their own per project. They used to share one slot across every project that depended on a directory of the same name, so a project could end up linked to another project's copy of the dependency.

  • The Workspace column of pnpm update --interactive now falls back to the project's path when its name is only whitespace, as it already did for a missing or empty one — all three render an equally blank label otherwise.

  • Checking GitHub Actions dependencies for updates is now opt-in for every command. Neither pnpm outdated nor pnpm update reads the workflow files unless --include-github-actions is passed or update.githubActions is set to true in pnpm-workspace.yaml. Reading them runs git ls-remote against every referenced repository, which fails in environments where GitHub is not reachable the way pnpm assumes (a GitHub Enterprise Server, a custom certificate authority, or an offline network) #​13254.

    pnpm outdated accepts the --include-github-actions option too.

  • pnpm update --interactive now measures its table in terminal columns rather than in characters. A package name, workspace name, or version containing wide characters (CJK, most emoji) no longer knocks its row's columns out of line with the rest of the group, and a wide character in a version no longer aborts the command with Subject parameter value width cannot be greater than the container width #​13357.

  • The Workspace column of pnpm update --interactive is more informative in two cases. A dependency outdated at the same version in several workspace projects is offered as one choice, since selecting it updates every project — that choice now names all of them instead of only the first. And a workspace project without a name is now labelled with its path rather than left blank, so several unnamed projects can be told apart.

  • An auto-installed optional peer is no longer hoisted at a version the workspace root's own dependency on that package excludes. resolvePeersFromWorkspaceRoot already made the workspace root's specifier decide which version a missing required peer is installed at; the optional-peer picker ignored it and always took the highest version present anywhere in the graph. In a workspace whose root pins postcss: 8.5.10, an importer that depends on webpack and declares no postcss of its own got [email protected] hoisted for terser-webpack-plugin's optional postcss peer, leaving two [email protected] instances in the graph #​13320.

  • overrides now also govern peers that pnpm auto-installs. Previously an override only rewrote dependencies declared in a manifest, so a peer nobody declares — installed because autoInstallPeers is on — resolved against its declared peer range and could bring in a second copy of the very package the override pinned. For example, with overrides: { react: npm:[email protected] } and a lone lucide-react dependency, pnpm installed [email protected]; it now installs the pinned [email protected] #​13320.

  • Under resolvePeersFromWorkspaceRoot, a workspace root dependency declared with link: or file: (or the path form of workspace:, such as workspace:../pkg) now satisfies another project's missing peer dependency at the linked package's own version, instead of being hoisted as a path. Those specifiers are relative to the project that declares them, so the same specifier reached a different directory — or none — from the project the peer was hoisted into, leaving a broken link. The root now has the same authority over the peer as it has when it declares the package with a version range #​13373.

  • Installs through a pnpr server now apply the project's whole verification policy. minimumReleaseAgeExclude, minimumReleaseAgeIgnoreMissingTime, trustPolicy, trustPolicyExclude, trustPolicyIgnoreAfter, and trustLockfile were ignored, so excluded packages were still held back and a lockfile containing them could be rejected.

    trustPolicy: no-downgrade no longer fails with TRUST_POLICY_INCOMPATIBLE_WITH_PNPR when a pnpr server is configured.

    --frozen-lockfile and --no-prefer-frozen-lockfile are now honored on the pnpr path, instead of resolving and rewriting the lockfile anyway. Since frozenLockfile defaults to true on CI, a CI install through a pnpr server now fails on an out-of-date lockfile rather than updating it.

  • Workspace installs through a pnpr server no longer crash with Cannot read properties of undefined (reading 'filter') after linking, when minimumReleaseAge is active #​13275.

  • Fixed pnpm dedupe updating valid catalog resolutions when another matching version exists in the lockfile.

  • pnpm -r run "/pattern/" --no-bail no longer exits zero when one of a project's matched scripts fails and a later one passes. The run summary carries a single status per project, and the passing script overwrote the recorded failure.

  • Restored the store block a first install prints, naming how packages were materialized and where the stores live #​13315:

    Packages are hard linked from the content-addressable store to the virtual store.
  Content-addressable store is at: ~/.local/share/pnpm/store/v11
  Virtual store is at:             node_modules/.pnpm

  • The root project's pnpm:devPreinstall script now runs before resolution and linking, as it does in pnpm 11. It is skipped under --ignore-scripts, --lockfile-only and --dry-run, by pnpm fetch and pnpm rebuild, and by a repeat install that is already up to date. Workspaces that use the hook to prepare state the install depends on — such as next.js, which generates a placeholder next bin with it — were left with dependents linked against files that were never created #​13313.

  • Prevented pnpm dedupe --check from removing an incompatible node_modules directory.

  • pnpm update --workspace no longer links dependencies the user never named:

    • Running it with updateConfig.ignoreDependencies configured no longer fails with ERR_PNPM_WORKSPACE_PACKAGE_NOT_FOUND for a dependency that is only published to the registry. Such dependencies keep their specifiers, as they already did when no dependencies were ignored.
    • Passing package selectors that match no direct dependency no longer falls back to linking every workspace dependency.

Platinum Sponsors

Bit
OpenAI

Gold Sponsors

Sanity Discord Vite
SerpApi CodeRabbit Stackblitz
Workleap Nx

v11.17.0: pnpm 11.17

Compare Source

Minor Changes

  • Added a new setting, update.githubActionsServer, for specifying the base URL of the GitHub server that hosts the repositories of the GitHub Actions referenced by the workflow files (for example, a GitHub Enterprise Server). When the setting is not defined, the URL is read from the GITHUB_SERVER_URL environment variable, falling back to https://github.com. The URL must use the https:// or http:// protocol #​13220.

    pnpm outdated and pnpm update no longer fail when the refs of a GitHub Action's repository cannot be read (for example, when the action's repository is private or hosted on a different GitHub server). Such actions are now skipped with a warning.

    Setting update.githubActions to false now makes pnpm outdated and the interactive pnpm update skip GitHub Actions dependencies.

Patch Changes

  • The token poll for web-based authentication no longer reads the body of non-OK or still-pending (HTTP 202) responses, and caps the token response body it does read at 64 KiB, so a malicious or compromised registry cannot exhaust memory through the poll pnpm/pnpm#12721.

  • Fixed catalog: references in dependencies and overrides failing to resolve when installing through a pnpr server, which errored with "No catalog entry '' was found for catalog 'default'." even though the catalog entry existed. Also fixed a crash on Windows when installing a nested workspace member (e.g. packages/foo) through a pnpr server #​13232.

  • Republished every package: the tarballs published by the v11.13.1 through v11.16.0 releases were missing most of their compiled files due to a packing bug #​13164.

  • Revert script ordering change for pnpm run --sequential /regex/

  • Support the from-git argument in the pnpm version command.

  • When the authentication URL cannot be rendered as a QR code (for example when it exceeds the maximum QR data capacity), web-based login now displays the URL alone with a warning instead of aborting authentication pnpm/pnpm#12721.

Platinum Sponsors
Bit
OpenAI
Gold Sponsors
Sanity Discord Vite
SerpApi CodeRabbit Stackblitz
Workleap Nx

v11.16.0: pnpm 11.16

Compare Source

Minor Changes

  • The first release of a package now publishes the version written in its manifest verbatim, instead of bumping off it. pnpm version -r and pnpm change status check the registry for each release's current version; when that version is not yet published, the package debuts at it and its pending changesets apply only from the next release. A newly added package seeded at 1100.0.0 with a minor changeset is therefore published as 1100.0.0 rather than skipping straight to 1100.1.0.

  • Added a --changeset flag to pnpm update. Set update.changeset to true in pnpm-workspace.yaml to enable this behavior by default, and use --no-changeset to override the setting for one update. After the update completes, pnpm writes a .changeset/pnpm-update-<suffix>.md file declaring a patch bump for every workspace package whose dependencies or optionalDependencies were changed by the update and a major bump when peerDependencies changed, including packages that consume an updated catalog entry via the catalog: protocol. Private packages, packages without a name, and packages listed in the ignore array of .changeset/config.json are skipped. If .changeset/config.json does not exist, a warning is printed and no changeset is generated.

  • Added GitHub Actions dependencies to pnpm outdated and interactive pnpm update. Non-interactive updates can include them with --include-github-actions or by setting update.githubActions to true in pnpm-workspace.yaml. Updated actions are pinned to exact commit hashes with their release tags preserved in comments.

  • Added update and audit settings sections to pnpm-workspace.yaml, superseding the awkwardly named updateConfig, auditConfig, and top-level auditLevel settings:

    update:
  ignoreDeps: # was updateConfig.ignoreDependencies
    - webpack
    - "@&#8203;babel/*"

audit:
  level: high # was auditLevel
  ignore: # was auditConfig.ignoreGhsas
    - GHSA-xxxx-yyyy-zzzz

    update.ignoreDeps lists dependency name patterns that pnpm update and pnpm outdated should skip. audit.level and audit.ignore tune pnpm audit.

    The deprecated updateConfig, auditConfig, and auditLevel settings keep working until the next major version. When both a new section value and its deprecated counterpart are set, the new section takes precedence and a warning is printed. Both the TypeScript CLI and the Rust config surface (pacquet) recognize the new sections.

Patch Changes

  • Fixed pnpm add --save-exact/--save-prefix and pnpm update writing a package's version with the peerDependencies range's prefix (e.g. ^19.2.7 instead of the requested 19.2.7) whenever the same package also appeared in peerDependencies. A real dependencies/devDependencies/optionalDependencies entry now takes precedence over a same-named peerDependencies entry when computing the current specifiers #​13108.

Platinum Sponsors

Bit
OpenAI

Gold Sponsors

Sanity Discord Vite
SerpApi CodeRabbit Stackblitz
Workleap Nx

v11.15.1

Compare Source

v11.15.0: pnpm 11.15

Compare Source

Minor Changes
  • Optional peer dependencies declared only via peerDependenciesMeta (for example debug's supports-color peer) are now resolved from a satisfying version already present in the dependency graph, the same way explicitly declared optional peer dependencies are. Previously such peers were only resolved this way when the package's metadata was read back from the lockfile, so an unrelated dependency change could rewrite peer resolutions across the whole lockfile.
Patch Changes

  • Updated adm-zip to prevent crafted ZIP archives from causing excessive memory allocation.

  • pnpm version -r no longer writes a versioning-ledger entry with no consumed intents as a bare intents: key, which the next run failed to read with ERR_PNPM_INVALID_VERSIONING_LEDGER. Empty intent lists are now written as intents: [], and the ledger reader accepts the bare form left by earlier releases.

  • Fixed pnpr workspace resolution to preserve project names and versions for workspace: dependencies.

Platinum Sponsors

Bit
OpenAI

Gold Sponsors

Sanity Discord Vite
SerpApi CodeRabbit Stackblitz
Workleap Nx

v11.14.0: pnpm 11.14

Compare Source

Minor Changes

  • peerDependencies now accept dependency specifiers that carry a scheme — a named-registry spec (<registry>:<version>), an npm: alias, or a file:/git/URL spec — instead of rejecting them with ERR_PNPM_INVALID_PEER_DEPENDENCY_SPECIFICATION #​13095. Such a peer is matched against the semver range carried by the specifier (work:5.x.x is checked as 5.x.x, npm:bar@^5 as ^5), or against * when it carries no version, while the original specifier still selects the package to auto-install. Bare name@version values, which are almost always a mistake, are still rejected.

  • Added pnpm doctor, which diagnoses the pnpm installation and the environment it runs in: the versions and install method, whether the global bin directory is on PATH, whether the store and cache are writable, which link strategies (reflink, hardlink, symlink) the store's filesystem supports, registry connectivity, and an offline file: install that exercises the resolve/store/link path end to end. Each check reports how to fix what it finds, and the command exits non-zero when any check fails.

    Use --offline to skip the checks that need network access, --json for machine-readable output, and --benchmark to time the filesystem and install checks.

  • Added support for executing multiple scripts matching a RegExp passed to pnpm run (e.g., pnpm run "/^build:.*/"), running matched scripts in deterministic lexicographical order. Restored the --sequential (-s) CLI option for pnpm run, which forces workspaceConcurrency to 1 so that matched scripts run sequentially one by one across and within packages.

Patch Changes

  • Fixed pnpm install failing with ERR_PNPM_LOCKFILE_IS_SYMLINK when pnpm-lock.yaml is a symlink, as build sandboxes such as Bazel and Nix stage it #​13073. Reading a lockfile through a symlink is allowed again, and an install that leaves the lockfile unchanged no longer rewrites it, so --frozen-lockfile no longer needs to write at all. Writing a changed lockfile through a symlink is still refused, as that would redirect the write onto the symlink's target.

  • Fixed frozen installs incorrectly treating equivalent Git dependency specifiers as a stale lockfile. See #​13039.

  • pnpm owner ls now reports authentication and authorization failures (401/403) as dedicated errors that include the registry's response body, matching pnpm owner add/rm, instead of a generic Failed to fetch owners message.

  • Recover from a metadata cache entry that disappears (concurrent cache cleanup, antivirus) after the registry has already answered the conditional request with 304 Not Modified. The metadata is re-requested once without cache validators instead of failing the install with ERR_PNPM_CACHE_MISSING_AFTER_304.

  • A project pinned to a broken pnpm release via packageManager or devEngines.packageManager now reports which release is broken and what to do about it, instead of failing inside the installer. pnpm self-update already refused these releases; the version switch does too.

  • Prevent broken-lockfile errors from including snippets of the lockfile's contents.

  • pnpm self-update now checks that the version it installed can run before making it the active pnpm. A release that installs but cannot execute is discarded with an error instead of replacing a working installation.

  • Fixed an out-of-memory regression when workspace projects concurrently resolve a package with large registry metadata pnpm/pnpm#13077.

  • Fixed pnpm update rewriting exact version pins that use the = operator (for example =3.5.1) to a caret range (^3.5.1). Exact pins are now preserved and written back as the bare version. See #​12745.

Platinum Sponsors

Bit
OpenAI

Gold Sponsors

Sanity Discord Vite
SerpApi CodeRabbit Stackblitz
Workleap Nx

v11.13.1: pnpm 11.13.1

Compare Source

Patch Changes

  • Fixed pnpm pack applying workspace-root ignore rules when a workspace package has its own .npmignore file.
  • Keep the interactive minimumReleaseAge approval prompt visible during pnpm install. The progress reporter now pauses its redraws while a prompt is waiting for input instead of overwriting it, so the install no longer hangs on a question the user cannot see #​13019.
  • Fixed pnpm self-update failing to link native platform binaries stored in sibling global virtual store slots.

v11.13.0: pnpm 11.13

Compare Source

Minor Changes

  • Added versioning.epics to pnpm-workspace.yaml. An epic ties a group of member packages to a lead package, constraining every member's major version to a band derived from the lead's major: while the lead is on major M, members live in M*100 … M*100+99. Members move independently inside the band (patch, minor, and a major intent that stays in-band); a bump that would carry a member past the band ceiling is rejected until the lead advances its own major. When a release plan takes the lead to a new stable major, every member re-bases to the band floor in the same plan. Membership is matched with pnpm's package selectors — name globs, ./-prefixed directory globs, and !-prefixed negations.

  • Added the team command for managing organization teams and team memberships on the registry, with create, destroy, add, rm, and ls subcommands and support for --otp, --parseable, and --json flags.

  • Added native workspace release management #​12952: the new pnpm change command records change intents as changesets-compatible .changeset/*.md files (pnpm change status shows the pending release plan), and the bare pnpm version -r consumes them — bumping versions across the workspace with dependent propagation through workspace: ranges, fixed groups, a maxBump cap, --filter narrowing, and --dry-run — writing changelogs, and recording consumed intents in a committed ledger that keeps cherry-picks and merge-backs between release branches safe. Packages can be moved onto per-package release lanes with the new pnpm lane <name> --filter <pkg> command and back with pnpm lane main --filter <pkg> (pnpm lane shows the membership), releasing X.Y.Z-lane.N prereleases from the same runs that release stable versions of the packages on the main lane. Configuration lives under the new versioning key of pnpm-workspace.yaml (fixed, ignore, maxBump, lanes, changelog). When two workspace projects publish the same name, intent files, versioning.lanes, and versioning.fixed/ignore may reference a project by its workspace-relative directory path (e.g. "./pnpm/npm/pnpm") — the one additive extension to the changesets format, applied automatically by pnpm change.

    Release changelogs default to registry storage (versioning.changelog.storage): no CHANGELOG.md is committed. Each release's section is composed at publish time and packed into the published tarball on top of the previously published version's changelog, and the consumed change intents are garbage-collected by a later pnpm version -r only once the registry confirms the version is published with its section. Set versioning.changelog.storage: repository to keep committed CHANGELOG.md files instead.

  • Added a new override selector form with an empty range — "pkg@": "<version>" — called a convergence override. It rewrites a dependency edge only when its exact version satisfies the edge's declared range, so compatible consumers converge on one version while incompatible consumers keep their own resolution — now and for any dependent added in the future #​12794.

    overrides:
  "form-data@": 4.0.6

    The value must be an exact version. When a full resolution detects that every declared range also admits a newer version, pnpm warns that the override is stale and names the version to converge on. Previously an empty range in an override selector was undocumented and behaved like a bare (unscoped) override.

Patch Changes

  • A tokenHelper set in the global pnpm auth.ini is no longer rejected as project-level configuration. The guard that blocks tokenHelper from a project .npmrc only treated ~/.npmrc as a trusted source, so a helper written to auth.ini (for example by pnpm config set) failed on every command and could not even be removed with pnpm config delete. A tokenHelper in a workspace or project .npmrc is still rejected.

  • pnpm cache delete now removes a package's metadata from every metadata cache directory (metadata, metadata-full, and metadata-full-filtered), instead of only the one the current resolution mode reads. Previously a package cached under a different mode (e.g. metadata-full-filtered) was left behind. Closes #​12753.

  • Fixed an injected workspace dependency (injectWorkspacePackages: true) incorrectly staying as file: instead of deduping back to link: when an unrelated, ordinary shared dependency resolved to a peer-suffixed variant for the target project's own copy but not for the injected occurrence. See #​10433.

  • pnpm deploy now supports workspaces that use catalogs.

  • Fixed pnpm deploy with a shared lockfile so local file: tarball dependencies keep their package name in the generated deploy lockfile. This prevents warm-store deploys from failing with ERR_PNPM_UNEXPECTED_PKG_CONTENT_IN_STORE when the tarball filename includes the version.

  • Options that follow create, exec, or test appearing as a subcommand of another command are now parsed instead of being silently treated as positional parameters. For example, pnpm team create @&#8203;org:team --registry <url> previously ignored the --registry option and sent the request to the default registry.

  • pnpm add -g, pnpm update -g, pnpm setup, and the self-updater no longer fail with ERR_PNPM_MISSING_TIME when trustPolicy: no-downgrade or resolutionMode: time-based is set in the global config #​12883. The decision to fetch full registry metadata now lives in one place, and the no-downgrade trust policy always requests full metadata (matching the self-updater), since the trust evidence it checks is missing from abbreviated metadata even on registries that include the time field.

  • pnpm list and pnpm why no longer crash with EMFILE: too many open files when a project has a large number of unsaved dependencies (packages present in node_modules but not in the lockfile). The reads of those packages are now concurrency-limited.

  • The published pnpm package no longer declares dependencies or devDependencies. Because the CLI bundles its runtime dependencies into dist/node_modules, those fields are dropped when packing, so npm install of the tarball no longer tries to resolve internal-only packages such as @pnpm/test-ipc-server. Closes #​12955.

  • Fixed pnpm publish --otp and pnpm publish --batch --otp to send the configured OTP to the registry.

  • pnpm publish again sends the package's README to the registry as metadata, so registries can render it on the package page. The readme is always included in the published metadata (matching the npm CLI), while the embed-readme setting continues to control only whether the readme is written into the package.json inside the tarball. This restores the behavior that was lost when publishing became fully native. Closes #​12966.

  • Fixed the d

✂ Note

PR body was truncated to here.

Configuration

📅 Schedule: (UTC)

  • Branch creation
    • At any time (no schedule defined)
  • Automerge
    • At any time (no schedule defined)

🚦 Automerge: Enabled.

♻ Rebasing: Whenever PR is behind base branch, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

  • If you want to rebase/retry this PR, check this box

This PR was generated by Mend Renovate. View the repository job log.

@cloudflare-workers-and-pages

cloudflare-workers-and-pages Bot commented Jul 29, 2026 •
edited
Loading

Copy link
Copy Markdown

Deploying with  Cloudflare Workers  Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status Name Latest Commit Preview URL Updated (UTC)
✅ Deployment successful!
View logs		 ccusage-guide 6bc395e Commit Preview URL

Branch Preview URL		 Aug 02 2026, 01:26 AM

@pkg-pr-new

pkg-pr-new Bot commented Jul 29, 2026 •
edited
Loading

Copy link
Copy Markdown

Open in StackBlitz

ccusage

npx https://pkg.pr.new/ccusage@1540

@ccusage/ccusage-darwin-arm64

npx https://pkg.pr.new/@ccusage/ccusage-darwin-arm64@1540

@ccusage/ccusage-darwin-x64

npx https://pkg.pr.new/@ccusage/ccusage-darwin-x64@1540

@ccusage/ccusage-linux-arm64

npx https://pkg.pr.new/@ccusage/ccusage-linux-arm64@1540

@ccusage/ccusage-linux-x64

npx https://pkg.pr.new/@ccusage/ccusage-linux-x64@1540

@ccusage/ccusage-win32-x64

npx https://pkg.pr.new/@ccusage/ccusage-win32-x64@1540

commit: 6bc395e

@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 687ece5588bc
Base SHA: c4be7c150fd1

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 353.0ms 2.85 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 314.1ms 3.21 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 122.7ms 8.21 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 114.9ms 8.77 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 27.9ms 5.4ms 5.11x 55.00 MiB 14.45 MiB 0.26x 0.06 MiB/s 0.28 MiB/s
claude session --offline --json 0.00 MiB 27.1ms 4.0ms 6.70x 55.25 MiB 12.44 MiB 0.23x 0.06 MiB/s 0.38 MiB/s
codex daily --offline --json 0.00 MiB 25.9ms 2.2ms 11.53x 55.25 MiB 10.68 MiB 0.19x 0.03 MiB/s 0.38 MiB/s
codex session --offline --json 0.00 MiB 25.8ms 2.4ms 10.62x 55.00 MiB 10.69 MiB 0.19x 0.03 MiB/s 0.35 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 376.6ms 325.8ms 1.16x 950.58 MiB 970.83 MiB 1.02x 2.67 GiB/s 3.09 GiB/s
codex --offline --json 1.01 GiB 116.4ms 95.4ms 1.22x 422.64 MiB 398.65 MiB 0.94x 8.65 GiB/s 10.55 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 18.78 KiB 18.78 KiB +0.00 KiB 1.00x
installed native package binary 4158.97 KiB 4158.97 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 687ece5588bc
Base SHA: c4be7c150fd1

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 354.1ms 2.84 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 301.6ms 3.34 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 123.6ms 8.15 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 100.8ms 9.99 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 33.5ms 30.5ms 1.10x 55.00 MiB 55.00 MiB 1.00x 0.05 MiB/s 0.05 MiB/s
claude session --offline --json 0.00 MiB 29.8ms 31.5ms 0.94x 55.00 MiB 55.00 MiB 1.00x 0.05 MiB/s 0.05 MiB/s
codex daily --offline --json 0.00 MiB 27.8ms 27.5ms 1.01x 55.00 MiB 54.50 MiB 0.99x 0.03 MiB/s 0.03 MiB/s
codex session --offline --json 0.00 MiB 26.1ms 27.1ms 0.96x 55.25 MiB 55.00 MiB 1.00x 0.03 MiB/s 0.03 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 349.6ms 338.9ms 1.03x 958.82 MiB 966.82 MiB 1.01x 2.88 GiB/s 2.97 GiB/s
codex --offline --json 1.01 GiB 120.3ms 135.4ms 0.89x 402.64 MiB 406.65 MiB 1.01x 8.37 GiB/s 7.44 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 18.78 KiB 18.78 KiB +0.00 KiB 1.00x
installed native package binary 4158.97 KiB 4158.97 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from 687ece5 to 4b2c0d9 Compare July 29, 2026 15:35
@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 4b2c0d9f8c90
Base SHA: 63fcdd9f4f6e

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 350.7ms 2.87 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 324.7ms 3.10 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 120.5ms 8.36 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 95.8ms 10.51 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 28.0ms 5.3ms 5.31x 55.00 MiB 12.69 MiB 0.23x 0.06 MiB/s 0.29 MiB/s
claude session --offline --json 0.00 MiB 29.8ms 3.0ms 9.80x 55.00 MiB 12.44 MiB 0.23x 0.05 MiB/s 0.51 MiB/s
codex daily --offline --json 0.00 MiB 28.4ms 2.5ms 11.24x 55.00 MiB 10.69 MiB 0.19x 0.03 MiB/s 0.34 MiB/s
codex session --offline --json 0.00 MiB 23.8ms 2.3ms 10.41x 55.25 MiB 10.68 MiB 0.19x 0.04 MiB/s 0.38 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 389.0ms 353.3ms 1.10x 934.82 MiB 964.82 MiB 1.03x 2.59 GiB/s 2.85 GiB/s
codex --offline --json 1.01 GiB 124.6ms 121.8ms 1.02x 414.64 MiB 418.65 MiB 1.01x 8.08 GiB/s 8.27 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 18.78 KiB 18.78 KiB +0.00 KiB 1.00x
installed native package binary 4158.97 KiB 4158.97 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 4b2c0d9f8c90
Base SHA: 63fcdd9f4f6e

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 341.0ms 2.95 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 309.4ms 3.25 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 116.7ms 8.63 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 94.9ms 10.61 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 29.5ms 26.3ms 1.12x 55.25 MiB 55.00 MiB 1.00x 0.05 MiB/s 0.06 MiB/s
claude session --offline --json 0.00 MiB 25.9ms 23.8ms 1.09x 55.00 MiB 55.00 MiB 1.00x 0.06 MiB/s 0.06 MiB/s
codex daily --offline --json 0.00 MiB 23.9ms 25.0ms 0.95x 55.25 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.03 MiB/s
codex session --offline --json 0.00 MiB 23.3ms 24.1ms 0.97x 55.00 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 367.8ms 352.9ms 1.04x 970.59 MiB 950.83 MiB 0.98x 2.74 GiB/s 2.85 GiB/s
codex --offline --json 1.01 GiB 122.7ms 118.3ms 1.04x 428.66 MiB 414.64 MiB 0.97x 8.21 GiB/s 8.51 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 18.78 KiB 18.78 KiB +0.00 KiB 1.00x
installed native package binary 4158.97 KiB 4158.97 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from 4b2c0d9 to ebf641c Compare July 29, 2026 16:50
@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: ebf641cc287c
Base SHA: 28d54f968f14

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 331.7ms 3.04 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 303.6ms 3.32 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 114.7ms 8.78 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 93.7ms 10.74 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 29.1ms 5.5ms 5.29x 55.25 MiB 12.70 MiB 0.23x 0.05 MiB/s 0.28 MiB/s
claude session --offline --json 0.00 MiB 26.3ms 2.7ms 9.87x 55.00 MiB 12.43 MiB 0.23x 0.06 MiB/s 0.58 MiB/s
codex daily --offline --json 0.00 MiB 24.3ms 3.0ms 8.17x 55.00 MiB 10.69 MiB 0.19x 0.04 MiB/s 0.29 MiB/s
codex session --offline --json 0.00 MiB 25.2ms 2.5ms 10.27x 55.25 MiB 10.69 MiB 0.19x 0.03 MiB/s 0.35 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 370.8ms 336.1ms 1.10x 948.83 MiB 948.58 MiB 1.00x 2.72 GiB/s 3.00 GiB/s
codex --offline --json 1.01 GiB 119.0ms 92.2ms 1.29x 418.64 MiB 400.64 MiB 0.96x 8.46 GiB/s 10.92 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 18.78 KiB 18.78 KiB -0.00 KiB 1.00x
installed native package binary 4158.97 KiB 4158.97 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: ebf641cc287c
Base SHA: 28d54f968f14

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 358.8ms 2.81 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 306.5ms 3.28 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 115.0ms 8.76 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 111.0ms 9.07 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 27.2ms 26.4ms 1.03x 54.75 MiB 55.25 MiB 1.01x 0.06 MiB/s 0.06 MiB/s
claude session --offline --json 0.00 MiB 25.2ms 24.3ms 1.04x 55.00 MiB 55.00 MiB 1.00x 0.06 MiB/s 0.06 MiB/s
codex daily --offline --json 0.00 MiB 23.5ms 22.5ms 1.04x 55.00 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s
codex session --offline --json 0.00 MiB 23.0ms 22.8ms 1.01x 55.00 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 371.6ms 345.6ms 1.08x 950.32 MiB 958.83 MiB 1.01x 2.71 GiB/s 2.91 GiB/s
codex --offline --json 1.01 GiB 120.2ms 119.9ms 1.00x 416.65 MiB 432.65 MiB 1.04x 8.38 GiB/s 8.40 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 18.78 KiB 18.78 KiB -0.00 KiB 1.00x
installed native package binary 4158.97 KiB 4158.97 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from ebf641c to 26b6c7f Compare July 29, 2026 17:29
@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 26b6c7f577a9
Base SHA: 3988dde441e5

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 354.1ms 2.84 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 306.7ms 3.28 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 115.5ms 8.72 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 93.8ms 10.73 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 28.6ms 5.6ms 5.14x 55.25 MiB 12.68 MiB 0.23x 0.05 MiB/s 0.28 MiB/s
claude session --offline --json 0.00 MiB 27.1ms 3.0ms 8.91x 54.75 MiB 12.44 MiB 0.23x 0.06 MiB/s 0.51 MiB/s
codex daily --offline --json 0.00 MiB 23.7ms 2.4ms 9.93x 55.00 MiB 10.69 MiB 0.19x 0.04 MiB/s 0.36 MiB/s
codex session --offline --json 0.00 MiB 23.1ms 2.4ms 9.48x 55.00 MiB 10.69 MiB 0.19x 0.04 MiB/s 0.35 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 374.3ms 325.1ms 1.15x 952.57 MiB 932.57 MiB 0.98x 2.69 GiB/s 3.10 GiB/s
codex --offline --json 1.01 GiB 119.4ms 93.7ms 1.27x 414.65 MiB 392.64 MiB 0.95x 8.43 GiB/s 10.74 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 18.78 KiB 18.78 KiB +0.00 KiB 1.00x
installed native package binary 4158.97 KiB 4158.97 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 26b6c7f577a9
Base SHA: 3988dde441e5

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 332.9ms 3.02 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 323.6ms 3.11 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 114.6ms 8.78 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 93.1ms 10.82 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 28.1ms 26.7ms 1.05x 55.00 MiB 55.00 MiB 1.00x 0.06 MiB/s 0.06 MiB/s
claude session --offline --json 0.00 MiB 25.9ms 24.3ms 1.07x 55.00 MiB 55.00 MiB 1.00x 0.06 MiB/s 0.06 MiB/s
codex daily --offline --json 0.00 MiB 25.4ms 22.2ms 1.14x 55.00 MiB 55.00 MiB 1.00x 0.03 MiB/s 0.04 MiB/s
codex session --offline --json 0.00 MiB 23.9ms 22.6ms 1.06x 55.00 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 364.2ms 346.8ms 1.05x 956.82 MiB 952.83 MiB 1.00x 2.76 GiB/s 2.90 GiB/s
codex --offline --json 1.01 GiB 117.6ms 117.7ms 1.00x 436.64 MiB 420.89 MiB 0.96x 8.56 GiB/s 8.55 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 18.78 KiB 18.78 KiB +0.00 KiB 1.00x
installed native package binary 4158.97 KiB 4158.97 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from 26b6c7f to 690bfba Compare July 29, 2026 20:30
@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 690bfbae9126
Base SHA: 615f460400f3

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 353.6ms 2.85 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 314.7ms 3.20 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 147.1ms 6.84 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 96.6ms 10.42 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 30.3ms 25.6ms 1.18x 55.25 MiB 55.00 MiB 1.00x 0.05 MiB/s 0.06 MiB/s
claude session --offline --json 0.00 MiB 28.3ms 24.9ms 1.14x 55.00 MiB 55.00 MiB 1.00x 0.05 MiB/s 0.06 MiB/s
codex daily --offline --json 0.00 MiB 23.6ms 22.0ms 1.07x 55.00 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s
codex session --offline --json 0.00 MiB 23.2ms 22.9ms 1.01x 55.00 MiB 55.25 MiB 1.00x 0.04 MiB/s 0.04 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 376.0ms 330.3ms 1.14x 948.58 MiB 950.82 MiB 1.00x 2.68 GiB/s 3.05 GiB/s
codex --offline --json 1.01 GiB 120.4ms 115.3ms 1.04x 422.64 MiB 424.65 MiB 1.00x 8.36 GiB/s 8.73 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 18.78 KiB 18.78 KiB -0.00 KiB 1.00x
installed native package binary 4158.97 KiB 4158.97 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 690bfbae9126
Base SHA: 615f460400f3

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 350.4ms 2.87 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 300.6ms 3.35 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 116.4ms 8.65 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 92.4ms 10.89 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 28.6ms 5.5ms 5.25x 55.25 MiB 12.69 MiB 0.23x 0.05 MiB/s 0.28 MiB/s
claude session --offline --json 0.00 MiB 24.2ms 4.6ms 5.31x 54.75 MiB 12.44 MiB 0.23x 0.06 MiB/s 0.34 MiB/s
codex daily --offline --json 0.00 MiB 25.4ms 2.2ms 11.37x 55.00 MiB 10.69 MiB 0.19x 0.03 MiB/s 0.38 MiB/s
codex session --offline --json 0.00 MiB 23.9ms 2.3ms 10.45x 55.00 MiB 10.68 MiB 0.19x 0.04 MiB/s 0.37 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 369.2ms 314.6ms 1.17x 944.82 MiB 944.83 MiB 1.00x 2.73 GiB/s 3.20 GiB/s
codex --offline --json 1.01 GiB 116.2ms 92.7ms 1.25x 416.65 MiB 412.65 MiB 0.99x 8.67 GiB/s 10.86 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 18.78 KiB 18.78 KiB -0.00 KiB 1.00x
installed native package binary 4158.97 KiB 4158.97 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from 690bfba to ffa50da Compare July 29, 2026 21:21
@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: ffa50da1297a
Base SHA: 50614f1373ea

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 356.1ms 2.83 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 317.9ms 3.17 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 121.6ms 8.28 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 110.5ms 9.11 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 30.1ms 5.5ms 5.45x 55.00 MiB 12.45 MiB 0.23x 0.05 MiB/s 0.28 MiB/s
claude session --offline --json 0.00 MiB 24.1ms 3.4ms 7.16x 55.00 MiB 12.46 MiB 0.23x 0.06 MiB/s 0.46 MiB/s
codex daily --offline --json 0.00 MiB 23.4ms 2.5ms 9.51x 55.25 MiB 10.70 MiB 0.19x 0.04 MiB/s 0.35 MiB/s
codex session --offline --json 0.00 MiB 23.6ms 2.4ms 10.05x 55.00 MiB 10.70 MiB 0.19x 0.04 MiB/s 0.36 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 383.3ms 308.7ms 1.24x 956.59 MiB 944.59 MiB 0.99x 2.63 GiB/s 3.26 GiB/s
codex --offline --json 1.01 GiB 121.4ms 93.0ms 1.31x 400.90 MiB 408.91 MiB 1.02x 8.29 GiB/s 10.82 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB -0.00 KiB 1.00x
installed native package binary 4190.28 KiB 4190.28 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Jul 29, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: ffa50da1297a
Base SHA: 50614f1373ea

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 345.4ms 2.91 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 326.9ms 3.08 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 123.9ms 8.13 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 91.9ms 10.95 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 25.2ms 25.3ms 1.00x 55.00 MiB 55.00 MiB 1.00x 0.06 MiB/s 0.06 MiB/s
claude session --offline --json 0.00 MiB 27.2ms 25.8ms 1.05x 55.00 MiB 55.00 MiB 1.00x 0.06 MiB/s 0.06 MiB/s
codex daily --offline --json 0.00 MiB 24.4ms 22.8ms 1.07x 54.75 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s
codex session --offline --json 0.00 MiB 23.5ms 22.5ms 1.05x 55.00 MiB 55.25 MiB 1.00x 0.04 MiB/s 0.04 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 381.2ms 346.7ms 1.10x 958.59 MiB 954.59 MiB 1.00x 2.64 GiB/s 2.90 GiB/s
codex --offline --json 1.01 GiB 118.3ms 116.3ms 1.02x 402.91 MiB 416.91 MiB 1.03x 8.51 GiB/s 8.65 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB -0.00 KiB 1.00x
installed native package binary 4190.28 KiB 4190.28 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from ffa50da to 68527b8 Compare July 30, 2026 02:19
@github-actions

github-actions Bot commented Jul 30, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 68527b8cdaf2
Base SHA: a71d92eb2fc9

Performance comparison skipped.

Base package URL was not ready before 300.000s. Fixture performance comparison requires a base package when --base-dir is not provided.

Base package: a71d92eb2fc9

@github-actions

github-actions Bot commented Jul 30, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 68527b8cdaf2
Base SHA: a71d92eb2fc9

Performance comparison skipped.

Base package URL was not ready before 300.000s. Fixture performance comparison requires a base package when --base-dir is not provided.

Base package: a71d92eb2fc9

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from 68527b8 to bb1215c Compare July 30, 2026 06:04
@github-actions

github-actions Bot commented Jul 30, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: bb1215cb8a2b
Base SHA: 8b7b2aa86559

Performance comparison skipped.

Base package URL was not ready before 300.000s. Fixture performance comparison requires a base package when --base-dir is not provided.

Base package: 8b7b2aa86559

@github-actions

github-actions Bot commented Jul 30, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: bb1215cb8a2b
Base SHA: 8b7b2aa86559

Performance comparison skipped.

Base package URL was not ready before 300.000s. Fixture performance comparison requires a base package when --base-dir is not provided.

Base package: 8b7b2aa86559

@github-actions

github-actions Bot commented Jul 31, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 2fbab42e036d
Base SHA: 5ee72780d8c1

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 348.8ms 2.89 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 303.3ms 3.32 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 117.8ms 8.55 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 94.4ms 10.67 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 30.6ms 5.5ms 5.54x 55.25 MiB 12.46 MiB 0.23x 0.05 MiB/s 0.28 MiB/s
claude session --offline --json 0.00 MiB 26.5ms 2.7ms 9.91x 54.75 MiB 12.45 MiB 0.23x 0.06 MiB/s 0.58 MiB/s
codex daily --offline --json 0.00 MiB 25.0ms 2.4ms 10.61x 55.00 MiB 10.45 MiB 0.19x 0.03 MiB/s 0.36 MiB/s
codex session --offline --json 0.00 MiB 25.7ms 2.4ms 10.75x 55.00 MiB 10.45 MiB 0.19x 0.03 MiB/s 0.36 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 357.3ms 309.8ms 1.15x 956.59 MiB 962.59 MiB 1.01x 2.82 GiB/s 3.25 GiB/s
codex --offline --json 1.01 GiB 125.9ms 106.3ms 1.18x 406.91 MiB 434.91 MiB 1.07x 8.00 GiB/s 9.47 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB -0.00 KiB 1.00x
installed native package binary 4194.28 KiB 4194.28 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from 2fbab42 to 214b95a Compare July 31, 2026 23:28
@github-actions

github-actions Bot commented Jul 31, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 214b95ac5919
Base SHA: 56f347cb3321

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 337.1ms 2.99 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 321.6ms 3.13 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 121.3ms 8.30 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 90.7ms 11.10 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 25.8ms 5.2ms 4.96x 55.00 MiB 12.46 MiB 0.23x 0.06 MiB/s 0.30 MiB/s
claude session --offline --json 0.00 MiB 24.9ms 4.6ms 5.35x 55.00 MiB 12.45 MiB 0.23x 0.06 MiB/s 0.33 MiB/s
codex daily --offline --json 0.00 MiB 24.6ms 2.4ms 10.28x 54.75 MiB 10.45 MiB 0.19x 0.03 MiB/s 0.36 MiB/s
codex session --offline --json 0.00 MiB 25.0ms 2.3ms 10.89x 55.00 MiB 10.46 MiB 0.19x 0.03 MiB/s 0.37 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 373.6ms 328.9ms 1.14x 944.59 MiB 946.59 MiB 1.00x 2.69 GiB/s 3.06 GiB/s
codex --offline --json 1.01 GiB 125.9ms 112.4ms 1.12x 412.91 MiB 412.91 MiB 1.00x 7.99 GiB/s 8.96 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.28 KiB 4194.28 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Jul 31, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 214b95ac5919
Base SHA: 56f347cb3321

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 357.3ms 2.82 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 320.9ms 3.14 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 116.3ms 8.66 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 91.8ms 10.97 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 29.3ms 26.7ms 1.09x 55.00 MiB 55.25 MiB 1.00x 0.05 MiB/s 0.06 MiB/s
claude session --offline --json 0.00 MiB 23.8ms 24.5ms 0.97x 55.25 MiB 55.00 MiB 1.00x 0.06 MiB/s 0.06 MiB/s
codex daily --offline --json 0.00 MiB 22.7ms 23.8ms 0.95x 55.00 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s
codex session --offline --json 0.00 MiB 23.1ms 22.9ms 1.01x 55.25 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 385.5ms 335.2ms 1.15x 966.59 MiB 964.59 MiB 1.00x 2.61 GiB/s 3.00 GiB/s
codex --offline --json 1.01 GiB 143.7ms 121.2ms 1.19x 430.91 MiB 412.91 MiB 0.96x 7.01 GiB/s 8.31 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.28 KiB 4194.28 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from 214b95a to 8489d70 Compare July 31, 2026 23:43
@github-actions

github-actions Bot commented Jul 31, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 8489d70f8454
Base SHA: e93a1faf0dc1

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 338.5ms 2.97 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 302.5ms 3.33 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 116.7ms 8.63 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 90.9ms 11.08 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 27.9ms 5.0ms 5.53x 55.00 MiB 12.46 MiB 0.23x 0.06 MiB/s 0.31 MiB/s
claude session --offline --json 0.00 MiB 27.3ms 4.5ms 6.05x 55.00 MiB 12.46 MiB 0.23x 0.06 MiB/s 0.34 MiB/s
codex daily --offline --json 0.00 MiB 23.7ms 2.3ms 10.37x 55.00 MiB 10.45 MiB 0.19x 0.04 MiB/s 0.38 MiB/s
codex session --offline --json 0.00 MiB 24.8ms 2.4ms 10.51x 55.00 MiB 10.46 MiB 0.19x 0.03 MiB/s 0.36 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 358.7ms 334.3ms 1.07x 944.59 MiB 954.59 MiB 1.01x 2.81 GiB/s 3.01 GiB/s
codex --offline --json 1.01 GiB 117.5ms 110.8ms 1.06x 408.91 MiB 406.91 MiB 1.00x 8.57 GiB/s 9.09 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.28 KiB 4194.28 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Jul 31, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 8489d70f8454
Base SHA: e93a1faf0dc1

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 345.6ms 2.91 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 339.3ms 2.97 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 114.4ms 8.80 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 91.2ms 11.04 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 27.3ms 28.2ms 0.97x 55.00 MiB 55.25 MiB 1.00x 0.06 MiB/s 0.05 MiB/s
claude session --offline --json 0.00 MiB 27.5ms 27.3ms 1.01x 55.00 MiB 55.00 MiB 1.00x 0.06 MiB/s 0.06 MiB/s
codex daily --offline --json 0.00 MiB 26.0ms 24.4ms 1.07x 55.00 MiB 55.25 MiB 1.00x 0.03 MiB/s 0.04 MiB/s
codex session --offline --json 0.00 MiB 23.7ms 24.5ms 0.97x 55.25 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.03 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 386.1ms 338.4ms 1.14x 952.59 MiB 1016.59 MiB 1.07x 2.61 GiB/s 2.97 GiB/s
codex --offline --json 1.01 GiB 117.1ms 115.8ms 1.01x 408.91 MiB 420.91 MiB 1.03x 8.60 GiB/s 8.69 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.28 KiB 4194.28 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from 8489d70 to 622c0c3 Compare August 1, 2026 01:16
@github-actions

github-actions Bot commented Aug 1, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 622c0c364a58
Base SHA: 735d55f9d792

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 360.0ms 2.80 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 310.5ms 3.24 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 119.6ms 8.41 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 91.4ms 11.01 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 26.0ms 5.3ms 4.89x 55.00 MiB 12.46 MiB 0.23x 0.06 MiB/s 0.29 MiB/s
claude session --offline --json 0.00 MiB 27.2ms 2.9ms 9.52x 55.00 MiB 12.45 MiB 0.23x 0.06 MiB/s 0.54 MiB/s
codex daily --offline --json 0.00 MiB 22.3ms 2.5ms 8.75x 55.00 MiB 10.45 MiB 0.19x 0.04 MiB/s 0.34 MiB/s
codex session --offline --json 0.00 MiB 23.2ms 2.4ms 9.79x 55.00 MiB 10.45 MiB 0.19x 0.04 MiB/s 0.36 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 370.3ms 333.5ms 1.11x 954.59 MiB 978.59 MiB 1.03x 2.72 GiB/s 3.02 GiB/s
codex --offline --json 1.01 GiB 115.8ms 94.7ms 1.22x 412.91 MiB 398.91 MiB 0.97x 8.69 GiB/s 10.63 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.34 KiB 4194.34 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Aug 1, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 622c0c364a58
Base SHA: 735d55f9d792

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 339.9ms 2.96 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 320.2ms 3.14 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 116.7ms 8.63 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 93.1ms 10.82 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 31.2ms 27.5ms 1.13x 55.00 MiB 55.25 MiB 1.00x 0.05 MiB/s 0.06 MiB/s
claude session --offline --json 0.00 MiB 25.9ms 25.1ms 1.03x 54.75 MiB 54.75 MiB 1.00x 0.06 MiB/s 0.06 MiB/s
codex daily --offline --json 0.00 MiB 22.6ms 24.8ms 0.91x 55.00 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.03 MiB/s
codex session --offline --json 0.00 MiB 22.5ms 23.6ms 0.95x 55.25 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 357.0ms 342.6ms 1.04x 956.59 MiB 950.59 MiB 0.99x 2.82 GiB/s 2.94 GiB/s
codex --offline --json 1.01 GiB 120.2ms 119.0ms 1.01x 420.91 MiB 410.90 MiB 0.98x 8.37 GiB/s 8.46 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.34 KiB 4194.34 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from 622c0c3 to e9991c5 Compare August 1, 2026 09:13
@renovate renovate Bot changed the title chore(deps): update pnpm to v11.17.0 chore(deps): update pnpm to v11.18.0 Aug 1, 2026
@github-actions

github-actions Bot commented Aug 1, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: e9991c54c793
Base SHA: da32a706a362

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 343.8ms 2.93 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 309.5ms 3.25 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 118.8ms 8.47 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 91.6ms 10.99 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 28.7ms 5.4ms 5.32x 55.25 MiB 12.46 MiB 0.23x 0.05 MiB/s 0.29 MiB/s
claude session --offline --json 0.00 MiB 26.7ms 2.8ms 9.58x 55.00 MiB 12.45 MiB 0.23x 0.06 MiB/s 0.55 MiB/s
codex daily --offline --json 0.00 MiB 26.5ms 2.4ms 11.12x 55.00 MiB 10.45 MiB 0.19x 0.03 MiB/s 0.36 MiB/s
codex session --offline --json 0.00 MiB 25.4ms 2.3ms 10.87x 55.00 MiB 10.45 MiB 0.19x 0.03 MiB/s 0.37 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 354.9ms 318.2ms 1.12x 924.59 MiB 942.59 MiB 1.02x 2.84 GiB/s 3.16 GiB/s
codex --offline --json 1.01 GiB 113.1ms 95.5ms 1.18x 408.91 MiB 428.91 MiB 1.05x 8.90 GiB/s 10.54 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.34 KiB 4194.34 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Aug 1, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: e9991c54c793
Base SHA: da32a706a362

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 348.1ms 2.89 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 298.0ms 3.38 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 120.0ms 8.39 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 94.1ms 10.70 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 30.1ms 28.2ms 1.07x 55.00 MiB 55.00 MiB 1.00x 0.05 MiB/s 0.05 MiB/s
claude session --offline --json 0.00 MiB 30.0ms 27.6ms 1.08x 55.25 MiB 55.25 MiB 1.00x 0.05 MiB/s 0.06 MiB/s
codex daily --offline --json 0.00 MiB 25.4ms 24.0ms 1.06x 55.00 MiB 55.00 MiB 1.00x 0.03 MiB/s 0.04 MiB/s
codex session --offline --json 0.00 MiB 22.1ms 23.3ms 0.95x 54.75 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 367.0ms 332.6ms 1.10x 942.59 MiB 962.59 MiB 1.02x 2.74 GiB/s 3.03 GiB/s
codex --offline --json 1.01 GiB 119.6ms 136.6ms 0.88x 418.91 MiB 412.91 MiB 0.99x 8.42 GiB/s 7.37 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.34 KiB 4194.34 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from e9991c5 to 9618a33 Compare August 1, 2026 11:28
@github-actions

github-actions Bot commented Aug 1, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 9618a3307a12
Base SHA: d426a9de9edf

Performance comparison skipped.

Base package URL was not ready before 300.000s. Fixture performance comparison requires a base package when --base-dir is not provided.

Base package: d426a9de9edf

@github-actions

github-actions Bot commented Aug 1, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 9618a3307a12
Base SHA: d426a9de9edf

Performance comparison skipped.

Base package URL was not ready before 300.000s. Fixture performance comparison requires a base package when --base-dir is not provided.

Base package: d426a9de9edf

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from 9618a33 to 3a65e8c Compare August 1, 2026 18:37
@github-actions

github-actions Bot commented Aug 1, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 3a65e8c9a475
Base SHA: 7fcbda3f1a91

Performance comparison skipped.

Base package URL was not ready before 300.000s. Fixture performance comparison requires a base package when --base-dir is not provided.

Base package: 7fcbda3f1a91

@github-actions

github-actions Bot commented Aug 1, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 3a65e8c9a475
Base SHA: 7fcbda3f1a91

Performance comparison skipped.

Base package URL was not ready before 300.000s. Fixture performance comparison requires a base package when --base-dir is not provided.

Base package: 7fcbda3f1a91

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from 3a65e8c to f0c6740 Compare August 1, 2026 22:31
@github-actions

github-actions Bot commented Aug 1, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: f0c6740dcd11
Base SHA: 11b68625ef8d

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 342.5ms 2.94 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 318.4ms 3.16 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 119.0ms 8.46 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 101.0ms 9.97 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 25.5ms 25.6ms 0.99x 55.00 MiB 55.00 MiB 1.00x 0.06 MiB/s 0.06 MiB/s
claude session --offline --json 0.00 MiB 26.8ms 23.6ms 1.14x 55.25 MiB 55.00 MiB 1.00x 0.06 MiB/s 0.07 MiB/s
codex daily --offline --json 0.00 MiB 27.0ms 22.4ms 1.20x 54.75 MiB 55.00 MiB 1.00x 0.03 MiB/s 0.04 MiB/s
codex session --offline --json 0.00 MiB 24.4ms 23.4ms 1.04x 55.00 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 361.6ms 346.9ms 1.04x 958.59 MiB 950.59 MiB 0.99x 2.78 GiB/s 2.90 GiB/s
codex --offline --json 1.01 GiB 114.8ms 117.4ms 0.98x 414.91 MiB 404.91 MiB 0.98x 8.77 GiB/s 8.58 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.34 KiB 4194.34 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Aug 1, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: f0c6740dcd11
Base SHA: 11b68625ef8d

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 351.3ms 2.87 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 297.3ms 3.39 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 142.9ms 7.05 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 97.0ms 10.38 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 29.4ms 5.5ms 5.39x 55.00 MiB 12.45 MiB 0.23x 0.05 MiB/s 0.28 MiB/s
claude session --offline --json 0.00 MiB 24.9ms 3.4ms 7.43x 55.00 MiB 12.45 MiB 0.23x 0.06 MiB/s 0.46 MiB/s
codex daily --offline --json 0.00 MiB 26.4ms 2.6ms 10.17x 55.00 MiB 10.46 MiB 0.19x 0.03 MiB/s 0.33 MiB/s
codex session --offline --json 0.00 MiB 22.6ms 2.4ms 9.54x 55.00 MiB 10.45 MiB 0.19x 0.04 MiB/s 0.36 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 364.1ms 311.8ms 1.17x 952.59 MiB 952.59 MiB 1.00x 2.77 GiB/s 3.23 GiB/s
codex --offline --json 1.01 GiB 136.7ms 104.5ms 1.31x 416.91 MiB 398.90 MiB 0.96x 7.36 GiB/s 9.63 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.34 KiB 4194.34 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot force-pushed the renovate/pnpm-11.x branch from f0c6740 to 6bc395e Compare August 2, 2026 01:25
@github-actions

github-actions Bot commented Aug 2, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 6bc395e9814b
Base SHA: 11bb17f23d27

This compares the PR package against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 339.0ms 2.97 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 301.6ms 3.34 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 119.9ms 8.39 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 115.7ms 8.70 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 28.4ms 26.7ms 1.07x 55.25 MiB 55.00 MiB 1.00x 0.05 MiB/s 0.06 MiB/s
claude session --offline --json 0.00 MiB 25.9ms 25.2ms 1.03x 55.00 MiB 55.25 MiB 1.00x 0.06 MiB/s 0.06 MiB/s
codex daily --offline --json 0.00 MiB 23.6ms 24.1ms 0.98x 54.75 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.04 MiB/s
codex session --offline --json 0.00 MiB 24.3ms 25.7ms 0.94x 55.00 MiB 55.00 MiB 1.00x 0.04 MiB/s 0.03 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published ccusage package from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 357.0ms 352.9ms 1.01x 948.59 MiB 954.59 MiB 1.01x 2.82 GiB/s 2.85 GiB/s
codex --offline --json 1.01 GiB 111.7ms 125.0ms 0.89x 416.91 MiB 408.91 MiB 0.98x 9.01 GiB/s 8.06 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.34 KiB 4194.34 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@github-actions

github-actions Bot commented Aug 2, 2026

Copy link
Copy Markdown
Contributor

ccusage performance comparison

PR SHA: 6bc395e9814b
Base SHA: 11bb17f23d27

This compares the Rust PR release binary against the configured base package on the same CI runner.

Package runtime diagnostics

Compares the PR package wrapper, the installed native optional dependency binary, and the workspace release binary on the same large fixture. This identifies whether slow package results come from JavaScript wrapper overhead, the published native binary build, or the Rust core itself.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
All rows run --offline --json, measured by hyperfine with 0 warmups and 1 runs. This isolates wrapper overhead from the installed native optional dependency and the workspace release binary built on the runner.

Command Runtime Input Median Throughput Samples
claude --offline --json Package wrapper 1.01 GiB 354.6ms 2.84 GiB/s 1
claude --offline --json Installed native binary 1.01 GiB 316.1ms 3.18 GiB/s 1
codex --offline --json Package wrapper 1.01 GiB 128.7ms 7.82 GiB/s 1
codex --offline --json Installed native binary 1.01 GiB 109.3ms 9.21 GiB/s 1

Committed fixture performance

Committed small fixtures for stable PR-to-PR feedback and explicit Claude/Codex command coverage.

Fixtures: Claude apps/ccusage/test/fixtures/claude (0.00 MiB, 2 files), Codex apps/ccusage/test/fixtures/codex (0.00 MiB, 1 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 2 warmups and 7 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude daily --offline --json 0.00 MiB 30.2ms 5.7ms 5.31x 55.00 MiB 12.45 MiB 0.23x 0.05 MiB/s 0.27 MiB/s
claude session --offline --json 0.00 MiB 25.3ms 2.9ms 8.78x 55.00 MiB 12.46 MiB 0.23x 0.06 MiB/s 0.54 MiB/s
codex daily --offline --json 0.00 MiB 25.7ms 2.3ms 11.33x 55.00 MiB 10.45 MiB 0.19x 0.03 MiB/s 0.38 MiB/s
codex session --offline --json 0.00 MiB 29.1ms 2.3ms 12.41x 55.00 MiB 10.45 MiB 0.19x 0.03 MiB/s 0.36 MiB/s

Large real-world-shaped fixture performance

Generated fixtures shaped from aggregate local log statistics: thousands of JSONL files, many small sessions, and a long tail of larger sessions. No real prompts, paths, or outputs are stored in the fixtures.

Fixtures: Claude /home/runner/_work/_temp/ccusage-large-fixture (1.01 GiB, 2597 files), Codex /home/runner/_work/_temp/ccusage-large-codex-fixture (1.01 GiB, 2597 files)
Base runs the published ccusage package from pkg.pr.new, installed before measurement; PR runs the published native ccusage binary from pkg.pr.new, installed before measurement. Both run --offline --json, measured by hyperfine with 0 warmups and 1 runs.
Peak RSS is measured separately with /usr/bin/time using 1 runs. Lower RSS ratios are better.

Command Input Base median PR median PR vs base Base peak RSS PR peak RSS PR/base RSS Base throughput PR throughput
claude --offline --json 1.01 GiB 372.8ms 324.1ms 1.15x 950.59 MiB 948.59 MiB 1.00x 2.70 GiB/s 3.11 GiB/s
codex --offline --json 1.01 GiB 117.2ms 94.3ms 1.24x 406.91 MiB 412.91 MiB 1.01x 8.59 GiB/s 10.68 GiB/s

Artifact size

Artifact Base PR Delta Ratio
packed ccusage-*.tgz 19.09 KiB 19.09 KiB +0.00 KiB 1.00x
installed native package binary 4194.34 KiB 4194.34 KiB +0.00 KiB 1.00x

Lower medians and smaller artifacts are better. CI runner noise still applies; use same-run ratios as directional PR feedback, not release guarantees.

@renovate
renovate Bot merged commit eda5e50 into main Aug 2, 2026
36 checks passed
@renovate
renovate Bot deleted the renovate/pnpm-11.x branch August 2, 2026 06:15
@github-actions github-actions Bot mentioned this pull request Aug 2, 2026
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@renovate-approve renovate-approve[bot] renovate-approve[bot] approved these changes

Assignees

No one assigned

Labels

dependencies renovate

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@aprlbrnl8-cyber