Repository navigation
Bring the Breaking-IT fork up to date and run lean CI - #2
Merged
Merged
Conversation
The Go versions used across the CI workflows were hardcoded in each file. With the Renovate manager added in #5981, the deliberate N-1 matrix entry in test.yml/test-browser.yml looked identical to the current-version entry, so Renovate tried to bump it to latest (#6085), collapsing the current + previous coverage into testing one version twice. Introduce .github/go-versions.env as the single source of truth for GO_VERSION_CURRENT and GO_VERSION_PREV, read via a new .github/actions/go-versions composite action. Non-matrix workflows (lint, tc39, wpt, xk6) and the lint action resolve the version in a step; the matrix workflows (test, test-browser) gain a small configure job that feeds the versions into their matrices. Both versions are now bumped manually with each Go release, so the Renovate custom manager for Go is removed. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* Skip code CI for docs-only PRs * Skip code CI for release-notes-only PRs
* Add feature brief template * Add feature briefs README * Add feature brief CI size check * Add CODEOWNERS entry for feature briefs
The v2 of grafana/shared-workflows/actions/get-vault-secrets will no
longer auto-export retrieved secrets as environment variables; consumers
must read them from the action's `secrets` output via
`${{ fromJSON(steps.<id>.outputs.secrets).NAME }}` instead of
`${{ env.NAME }}`.
This change opts every existing usage into the new pattern by:
- giving each get-vault-secrets step an explicit `id`
- setting `export_env: false` to match the v2 behavior
- replacing downstream `env.NAME` references with the `fromJSON(...)` form,
and threading bash consumers via per-step `env:` blocks where needed
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
Closes #2694 Signed-off-by: Yordis Prieto <yordis.prieto@gmail.com> Co-authored-by: Théo Crevon <oleiade@users.noreply.github.com>
* fix: Don't freeze JS primitives * Fix linter errors
…#6147) * fix potential data race and TOCTOU inconsistency in route operation * added code review comments: better code comments and errcheck
…curity] (#6134) | datasource | package | from | to | | -------------- | ------- | ------- | ------- | | golang-version | go | 1.25.11 | 1.25.12 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
…080) Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
…tion to v1.0.3 (#6124) | datasource | package | from | to | | ------------- | ------------------------ | ---------------------------- | ---------------------------- | | github-digest | grafana/shared-workflows | azure-trusted-signing/v1.0.2 | azure-trusted-signing/v1.0.3 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
| datasource | package | from | to | | ---------- | ------------------------ | ------- | ------- | | go | github.com/evanw/esbuild | v0.28.0 | v0.28.1 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
…1 (#6083) | datasource | package | from | to | | ---------- | ---------------------------------- | ------- | ------- | | go | github.com/mccutchen/go-httpbin/v2 | v2.23.0 | v2.23.1 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
… ID widening (#6152) Upstream regenerated the client from a newer API spec that widens essentially every resource ID (projects, load tests, test runs, stack/default-project IDs, etc.) from int32 to int64. Follow suit in internal/cloudapi/v6: widen Project.ID, LoadTest.ID/ProjectID, and the Client method parameters/return values that carry these IDs, and adapt the ValidateOptions/CreateLoadTest/StartTest/etc. request builders and the v6test fixture server to the new SDK types. This also removes the int64->int32 "overflows int32" truncation dance in internal/cmd/cloud.go and cloud_loadtest_list.go for project IDs, since the v6 client now accepts the same int64 range as the rest of k6's cloud config. Stack IDs are unaffected (X-Stack-Id stays int32 upstream) and keep their existing truncation check. Pinned to the unreleased latest commit on k6-cloud-openapi-client-go's main (v0.0.3-0.20260625132134-c48eae4f9a18); no tagged release exists past v0.0.2 yet. SetStackID took int32, forcing every caller to duplicate an int64->int32 overflow check before calling it (cloud.go, cloud_loadtest_list.go, cloud_project_list.go all had their own copy). Move that check into SetStackID itself: it now takes int64 (matching how stack IDs are stored everywhere else, e.g. cloudapi.Config) and returns an error if the value doesn't fit in the int32 the underlying SDK's X-Stack-Id header setter still requires. That int32 ceiling is a real upstream limitation, not something fixable client-side: X-Stack-Id is generated from a single shared OpenAPI parameter component that never got the format: int64 bump the other ID fields received during the same regeneration passes - confirmed by reading the spec directly, not templates or generator config. Filing that upstream separately; until it's fixed, values overflowing 32 bits are rejected up front instead of being silently truncated.
Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
| datasource | package | from | to | | ---------- | ------- | ------ | ------ | | docker | alpine | 3.24.0 | 3.24.1 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
| datasource | package | from | to | | ---------- | ------------------- | ------- | ------- | | go | golang.org/x/crypto | v0.53.0 | v0.54.0 | | go | golang.org/x/net | v0.56.0 | v0.57.0 | | go | golang.org/x/sync | v0.21.0 | v0.22.0 | | go | golang.org/x/term | v0.44.0 | v0.45.0 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
…4 (#6123) | datasource | package | from | to | | ------------- | ------------------------ | --------------- | --------------- | | github-digest | grafana/shared-workflows | aws-auth/v1.0.3 | aws-auth/v1.0.4 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
…6122) | datasource | package | from | to | | ----------- | ----------------------------- | ------ | ------ | | github-tags | golangci/golangci-lint-action | v9.2.0 | v9.3.0 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
* feat(streams): implement the WritableStream Web API Adds a spec-compliant WritableStream to the experimental streams module, alongside its WritableStreamDefaultWriter and WritableStreamDefaultController, following the same structure and conventions as the existing ReadableStream implementation (one method per WHATWG Streams abstract operation, with spec-linked doc comments). The stream's JavaScript object is built as a plain, extensible object so that consumers (and the WPT recording helpers) can attach their own properties; a hidden reference to the Go value is kept for brand checks and the writer constructor. Because settling a promise from Go runs its reactions synchronously in the k6/Sobek event loop, promise settlements are batched and flushed once the surrounding synchronous state changes complete, so reactions observe up-to-date state and settle in specification order. AbortSignal (controller.signal), ByteLengthQueuingStrategy and piping are not supported yet and are intentionally left out. * test(streams): run the WritableStream Web Platform Tests Runs the WHATWG Streams WPT 'writable-streams' suite against the module, the same way the readable-streams suite is run: the tests are checked out and patched by tests/checkout.sh, then executed by TestWritableStream. The shared harness setup now also loads resources/recording-streams.js, which the writable-streams tests rely on (recordingWritableStream). Two patches adapt the suite to k6: - aborting.any.js: comments out the AbortSignal/DOMException tests (unsupported) and two promise-ordering tests that only fail under the fully concurrent suite due to the event loop's synchronous promise settlement (they pass in isolation). - reentrant-strategy.any.js: comments out one pathological test that recursively calls write() from strategy.size() and depends on exact microtask interleaving. * docs(streams): add a WritableStream example Adds examples/experimental/writable-streams.js, mirroring the existing ReadableStream example, to demonstrate creating a WritableStream, acquiring a writer, and writing chunks while respecting backpressure via writer.ready. * fix(streams): reject controller constructor calls * fix(streams): handle undefined sink members * fix(streams): reuse writable controller wrapper * fix(streams): handle abrupt size conversion * fix(streams): track queued promise settlements * fix(streams): put writable writers on prototype * refactor(streams): address review cleanup * refactor(streams): simplify writer prototype setup * fix(streams): protect writable stream prototypes * test(streams): restore writable stream regressions * fix(streams): align writable Web IDL bindings * refactor(streams): centralize promise helpers Move the stream-local promise wrapper next to the existing Sobek helpers and rename goja.go to reflect the runtime used by k6. This keeps promise creation, settlement, and handled-state behavior in one place instead of maintaining parallel readable and writable implementations. Represent ReadableStream reader closed promises with the same wrapper used by WritableStream. These promises settle synchronously on the runtime thread, so the event-loop-oriented js/promises helper is intentionally not used. WritableStream transaction scheduling remains local because it implements writable-specific settlement ordering. * fix(streams): keep promise accessors private ReadableStream's closed-promise accessors return the package-private promise wrapper and are only used inside the streams package. Keep the methods private as well so the API reflects their scope and satisfies revive's unexported-return check.
* Add parse header function
This function will be used to parse the incoming headers from extraInfo
* Add extraInfo event registration
Registers our interest to receive extraInfo data async from chromium.
* Add event handling in select
* Add initial handler and wire in header parsing
* Add extraHeader field in request and response
* Add methods on request and response to add...
...extra headers from extraInfo asynchronously.
* Add extraInfoTracker
This is the key to being able to track all the extraInfo headers that
arrive for each of the redirect requests/responses. The initial request
will have async extraInfo headers under the same requestID. Subsequent
redirects will also have the same requestID. So extraInfoTracker helps
to keep track of each redirect's extraInfo header.
* Add extraInfoTracker at the network manager level
Since it already tracks the requests, this feels like a natural place
for the extraInfoTracker to live.
* Add tracking of extraInfo headers for the ...
...request and response. This will be matched up with the corresponding
request or response. When there's a match the request/response will
have its extra headers set.
* Add request/response processing for new requests
This covers two cases:
1. A non-redirecting request, so only processing of a request.
2. A redirecting request which will process a request and a response.
* Add response processing onResponseReceived
* Add a cleanup of extra header request tracking
* Update request headersSize
This now takes into account whether there are extra headers before
calculating the headers size.
* Update AllHeaders to take into account the...
...extra headers in request.
* Update request headersArray to account for...
...extra headers.
* Update response to account for extra headers
* Refactor existing tests so that we disregard...
...extra headers which may or may not be present.
* Add TestPageOnRequestAllHeadersExtraInfo
* Add TestPageOnResponseAllHeadersExtraInfo
* Add TestPageOnResponseAllHeadersRedirectChain
* Remove frameUrl assertion
Since the frame URL is updated asynchronously, we cannot rely on a
deterministic way of asserting on it. It is being ignored for now.
* Update internal/js/modules/k6/browser/tests/page_test.go
Co-authored-by: Inanc Gumus <inanc.gumus@grafana.com>
* Fix HeaderValues for newline-joined headers
HeaderValues looked up the name without lowercasing and split it on a
comma, but AllHeaders lowercases keys and joins multiple values with a
newline. For Set-Cookie this returned a single newline-glued string,
or an empty value when the name was not already lowercase. Lowercase
the name and split on the newline to match AllHeaders.
* Fix extra-info pairing across redirect chains
The tracker paired the i-th ExtraInfo event with the i-th response,
but enqueued every response unconditionally. CDP only emits ExtraInfo
when hasExtraInfo is set, and may report it erroneously for cached
responses (crbug.com/1340398). A redirect chain with a no-extra-info
hop therefore slipped the index and applied a later hop's headers to
the wrong response.
Track only responses that expect ExtraInfo, gating on hasExtraInfo,
RedirectHasExtraInfo and servedFromCache, and attach the raw request
headers via the response, mirroring Playwright. Defer cleanup until
every tracked response is paired so a late ExtraInfo event is not
dropped. This drops processRequest, so internal data:/blob: requests
no longer leak a tracker entry.
* Add wait for raw headers to header accessors
allHeaders, headerValue(s) and headersArray returned whatever had
arrived, so reading them before the ExtraInfo event (e.g. in a
page.on('response') handler) missed raw headers such as Set-Cookie.
Make the JS accessors wait for the raw headers to be resolved first,
matching Playwright.
A request/response resolves its wait once the raw headers are applied,
or once it is known none will arrive (no ExtraInfo, served from cache,
or the terminal loadingFinished/loadingFailed event), so a reader
never blocks past the request's lifetime.
* Update browser data metrics to await raw headers
headersSize prefers the raw (ExtraInfo) headers when present, but the
browser_data_sent and browser_data_received metrics were emitted as
soon as the request/response was created, before the raw headers had
necessarily arrived. The same workload could therefore report
different sizes run-to-run depending on CDP event ordering.
Emit both metrics from a goroutine that first waits for the raw
headers to resolve, so the size is always computed from a stable
source. The samples keep their original wallTime, so only the emission
is deferred, not the recorded timestamp.
* Update redirect test to assert per-hop cookies
The redirect-chain test asserted only X-Step on the final response,
which is present in the provisional headers, so it passed even with
ExtraInfo pairing disabled and never checked the intermediate hops.
Assert a distinct Set-Cookie per hop instead. Set-Cookie arrives only
via ExtraInfo, so each hop's cookie proves its events were paired with
the correct response, catching both mis-attribution and merging. Also
add a unit test covering HeadersArray for multi-valued headers.
---------
Co-authored-by: Inanc Gumus <inanc.gumus@grafana.com>
| datasource | package | from | to | | ---------- | ------- | --------------- | --------------- | | docker | debian | trixie-20260610 | trixie-20260623 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
Config, NewConfig, Apply, GetConsolidatedConfig, mergeFromCloudOption, and GetTemporaryCloudConfig had zero callers anywhere outside this file and its own test file. Every consumer of internal/cloudapi/v6 (internal/cmd/cloud.go, cloud_login.go, cloud_project_list.go, and internal/cloudapi/provisioning) only uses Client/NewClient/Project/ TestProgress/Status*/Result*/StatusEvent from this package; the real config path for the v6 API goes through the root cloudapi.Config instead, which this file duplicated ~90% of and then drifted from (e.g. missing StackID in its options.cloud merge, no Hostv6/ MetricsPushURL/TestRunToken/StackURL/DefaultProjectID fields). Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
…156) | datasource | package | from | to | | ---------- | ----------------------------- | ------- | ------- | | go | github.com/klauspost/compress | v1.18.5 | v1.19.0 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
… v6 clients (#6159) * refactor(cloudapi): share 401/403 error classification between v1 and v6 clients cloudapi.CheckResponse and internal/cloudapi/v6.CheckResponse each independently defined their own errNotAuthenticated/errNotAuthorized sentinel errors and duplicated the same "401 -> not authenticated, 403 -> not authorized" branching. Extract that single genuinely shared piece into a small internal/cloudapi/httperr package (ErrNotAuthenticated, ErrNotAuthorized, ClassifyStatus), and have both clients use it instead of maintaining separate sentinel instances with identical message text. * test: drop redundant httperr test, trim comments ClassifyStatus's 401/403 mapping is already exercised directly by cloudapi's and v6's own CheckResponse tests (both assert on httperr.ErrNotAuthenticated/ErrNotAuthorized for those status codes), so a dedicated unit test for the same three-way switch adds nothing. Also trims the package/var doc comments down to what isn't already obvious from the names.
…orkaround (#6151) Bumps github.com/grafana/k6-cloud-openapi-client-go to main (v0.0.3-0.20260716094638-49089b6e624d), which includes the upstream fix for callAPI's retry loop not resetting the request body before replaying it (grafana/k6-cloud-openapi-client-go#31) - the SDK now resets req.Body from GetBody itself on every retry attempt, the same thing k6's own bodyResetTransport worked around client-side. With that fixed upstream, drop bodyResetTransport entirely and build the client Configuration via k6cloud.NewConfiguration() instead of a hand-written struct literal, since the zero-value fields it was setting (DefaultHeader, OperationServers) are pure boilerplate the package itself never reads or writes. This SDK commit also carries the int32->int64 ID widening already consumed in #6152, and an added LICENSE file - both already reflected in/compatible with the current vendor tree, so this change is scoped to just the retry-fix consumption.
| datasource | package | from | to | | ---------- | ------- | ------ | ------ | | docker | golang | 1.26.4 | 1.26.5 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
* Fix dispose and getContext data race. Closes #5328 * Expand contextMu coverage to all b.context accesses * Cover public context getter in regression test and add result assertions * update stale comment
* Fix implicit area role test fixture * Make implicit area fixture accessible * Include hidden image map areas in role test
…rror (#6408) When group() is called with an async callback and the async-metric-context experimental feature is not enabled, the returned error now tells the user how to enable it (--features async-metric-context / K6_FEATURES) instead of implying async callbacks are unsupported outright. Addresses review feedback on #6341. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
…266) | datasource | package | from | to | | ---------- | ----------------------------- | ------- | ------- | | go | github.com/klauspost/compress | v1.19.1 | v1.20.0 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
* fix(webcrypto): Validate getRandomValues input * test(webcrypto): cover invalid getRandomValues inputs * fix(webcrypto): throw ECMAScript TypeErrors for invalid input
Hidden and detached waits should succeed when no element matches. Return null to JavaScript in that case and preserve element handles when a hidden element still exists.
* Bound screenshot capture by configured timeouts Unanswered layout and capture requests used the page lifetime, leaving scripts unable to handle screenshot failures before teardown. Apply the screenshot deadline to shared capture work and restore the viewport and background with a bounded cleanup context. Keep the page usable after the operation times out and preserve cancellation. * Simplify screenshot regression coverage Exercise deadline and cancellation behavior through page screenshots in one table. Check the actual background reset and keep failed viewport restoration from reporting a size the browser did not accept. Drop the redundant viewport return value now that restoration preserves the original emulation settings.
Remove weekly k6packager schedule
| datasource | package | from | to | | ---------- | ---------------------- | ------- | ------- | | go | google.golang.org/grpc | v1.83.2 | v1.84.0 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
| datasource | package | from | to | | ---------- | ------- | ------ | ------ | | docker | golang | 1.27.0 | 1.27.1 | Signed-off-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com> Co-authored-by: renovate-sh-app[bot] <219655108+renovate-sh-app[bot]@users.noreply.github.com>
…6345) Browser callbacks and CDP network events cross execution boundaries that Sobek's Promise tracking cannot observe. Requests and Web Vitals can be reported after the JavaScript call or group that initiated them has returned. Reading the VU's live tags at that point makes attribution depend on response timing and unrelated subsequent operations. Capture the complete metric context when browser callbacks are registered and restore it while invoking page events, waitForEvent predicates, routes, and browser-context callbacks. Track the context of browser operations that can initiate network traffic. Keep launched operations active until their returned Promise settles and retain the most recently finished operation for delayed CDP events. Discard calls that fail before launching an operation. Cover pages, frames, locators, element and JavaScript handles, keyboard, mouse, and touchscreen APIs. Explicitly classify every mapped callable as network or passive. Passive observers such as waitForResponse must remain unwrapped; otherwise, they can compete with the operation that initiated a request and make attribution timing-dependent. Requiring classification also makes newly added APIs fail closed instead of silently receiving the wrong behavior. CDP does not expose a general causal link between an API call and all traffic it triggers, so network attribution remains best-effort. Prefer an applicable frame-loader binding, then the latest active operation, the most recently finished operation, the page fallback, and finally the live VU context. Snapshot the selected context when NetworkManager creates a request so later operations cannot retag its response metrics. Preserve that snapshot across redirects, bind document navigations to their loaders, propagate opener context to popup pages, and use the navigation context for asynchronously reported Web Vitals. When async-metric-context is disabled, skip callback snapshots, operation wrappers, loader bindings, and request-owned snapshots, preserving the existing emit-time behavior. Add unit and integration coverage for callbacks, passive and network mapping classification, delayed requests, redirects, input actions, popups, standalone JavaScript handles, tags, metadata, Web Vitals, and teardown-safe metric emission. Add the metric-context example to the browser E2E workflow.
* Add v2.3.0 release notes Starts the v2.3.0 cycle. The release notes file has to exist on master before the tag is pushed, because build.yml builds the GitHub release body from 'release notes/v2.3.0.md'. Seeded from the template; entries land as PRs merge. * Update v2.3.0 release notes * Show how once transforms a duration scenario * Include the latest merged release fixes * Explain scenario selection before single-run mode Show how to run an existing workload without script edits, then combine selection with once for focused smoke tests. Related: #6360 * Preserve the existing once release documentation Keep the combination example under scenario selection and restore the existing once section unchanged. * Add JavaScript features and proxy timing fix * Document async groups and timer fixes * Include usage-report help fix in release notes * Add browser context race fix to release notes * Pin release notes docs link to v2.3 * Note browser role-selector fixture update * Use straight quotes in release notes * Link to docs for scenario selection with once * Update async guidance and dependency notes * Add browser and random-value fixes to notes * Document REST execution results in release notes * Credit REST status external contributor * Update release dependency versions * Remove blank lines between release note items
Preserve lean exclusions, browser timing metrics and error payloads. Carry async metric context into fork-specific samples and add regression coverage. Exclude installation-ID/build-origin reporting and Grafana Renovate approval workflows. Keep browser admission control out of this merge.
Consolidate the custom fork on master with k6 v2.3.0
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What?
Integrate jgaalen/k6 master with its full history and use Breaking-IT/k6 for future development. Replace the upstream workflows with the four-target lean release build, race-enabled fork tests and focused Chromium regressions on standard Ubuntu runners.
Why?
Build and test the binaries used by BreakTest under the organization that distributes them. Preserve the historical breakingit branch and existing release tags.
Validation
The Lean CI jobs exercise the release script and the affected runtime, CLI, output, HTTP, metrics and browser paths.