You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[Due for payment 2026-08-19] Investigate workflow job failing on main: build (New Expensify Dev, DebugDevelopment, false) #98190
🐛 Error Message:
failure: Process completed with exit code 1.
⚠️Action Required⚠️:
🛠️ A recent merge appears to have caused a failure in the job named build (New Expensify Dev, DebugDevelopment, false).
🔍 This issue has been automatically created and labeled with Workflow Failure for investigation.
Immediately after, both the consumer (remote-build-ios.yml) and the publisher (publishReactNativeArtifacts.yml) broke the same way: App's patches/ (and Mobile-Expensify's own patches, used only for the hybrid target) were still named for react-native@0.85.3. patch-package's version-mismatch guard treats that as fatal, so npm ci died before any pod install or artifact publish could even happen — no RN-0.86 artifact existed yet.
Once the patches got renamed to 0.86.0, the publish pipeline ran clean for the hybrid target from 20:29–20:46 that same day and successfully published com.expensify.react-hybrid.react-native-artifacts:0.86.0-1 (I confirmed via the GitHub Packages API that it exists, is visibility: public, and that the publish run completed with conclusion: success end-to-end: build slices → compose XCFramework → publish).
Despite that artifact existing, remote-build-ios.yml's hybrid job now fails differently: CocoaPods' pod install for React-Core-prebuilt issues a bare, unauthenticated GET straight to https://maven.pkg.github.com/Expensify/App/com/expensify/react-hybrid/react-native-artifacts/0.86.0-1/...tar.gz and gets a 401. GitHub's Packages Maven registry requires authentication on every read, even for public packages — so this request was always going to fail, independent of the artifact's actual state. The request signature (--netrc-optional, -A 'CocoaPods/1.16.2 cocoapods-downloader/2.1', no -H Authorization) is unmistakably CocoaPods' own built-in downloader, not the custom Bearer-token curl inside patched_ios_artifacts.rb's download_rncore_tarball.
What's still unexplained (the actual open bug)
The resolver correctly finds the hash-matching 0.86.0-1 version (I can see it decided to use it — the URL, group naming, and version-suffix format all match its design exactly), but somewhere between "resolver says use this artifact" and "download the tarball," the authenticated download path gets bypassed and CocoaPods' plain downloader takes over instead — which can never work against this registry. I couldn't pin the exact mechanism from static analysis (candidates: an exception inside PatchedIOSArtifacts.resolve's bun subprocess swallowed silently, a stale Pods/Local Podspecs/React-Core-prebuilt.podspec.json cache, or a require/timing issue specific to how rock's CocoaPods install sequence loads the podspec).
Recommended next step: this is worth a direct question to mateuuszzzzz — specifically whether download_rncore_tarball's local-download-then-file://-wrap flow has been verified to actually run under rock's pod install (vs. plain bundle exec pod install from scripts/pod-install.sh, which is what's exercised in local dev and may be the only path that's been manually tested).
I’ve prepared two PRs to align our Ruby scripts that apply hot patches to artifact-related code with the changes introduced in RN 0.86. I’ve also introduced a few improvements to make the scripts less error-prone.
changed the title [-]Investigate workflow job failing on main: build (New Expensify Dev, DebugDevelopment, false)[/-][+][Due for payment 2026-08-19] Investigate workflow job failing on main: build (New Expensify Dev, DebugDevelopment, false)[/+]on Aug 12, 2026
The solution for this issue has been 🚀 deployed to production 🚀 in version 9.4.52-11 and is now subject to a 7-day regression period 📆. Here is the list of pull requests that resolve this issue:
If no regressions arise, payment will be issued on 2026-08-19. 🎊
The following checklist (instructions) will need to be completed before the issue can be closed. Please copy/paste the BugZero Checklist from here into a new comment on this GH and complete it. If you have the K2 extension, you can simply click: [this button]. If no checklist is needed for this issue, you can click: [no checklist button]
Activity
roryabraham commentedon Aug 11, 2026
ok, I do see iOS builds failing on main: https://github.com/Expensify/App/actions/workflows/remote-build-ios.yml since the RN 0.86 upgrade, even after the Mobile-Expensify piece was merged too
roryabraham commentedon Aug 11, 2026
What actually happened, in order
jakubstec/upgrade-to-rn-0.86PR (Update version to 1.2.49-0 on main #14021).remote-build-ios.yml) and the publisher (publishReactNativeArtifacts.yml) broke the same way: App'spatches/(and Mobile-Expensify's own patches, used only for the hybrid target) were still named forreact-native@0.85.3.patch-package's version-mismatch guard treats that as fatal, sonpm cidied before any pod install or artifact publish could even happen — no RN-0.86 artifact existed yet.0.86.0, the publish pipeline ran clean for the hybrid target from 20:29–20:46 that same day and successfully publishedcom.expensify.react-hybrid.react-native-artifacts:0.86.0-1(I confirmed via the GitHub Packages API that it exists, isvisibility: public, and that the publish run completed withconclusion: successend-to-end: build slices → compose XCFramework → publish).remote-build-ios.yml's hybrid job now fails differently: CocoaPods'pod installforReact-Core-prebuiltissues a bare, unauthenticated GET straight tohttps://maven.pkg.github.com/Expensify/App/com/expensify/react-hybrid/react-native-artifacts/0.86.0-1/...tar.gzand gets a 401. GitHub's Packages Maven registry requires authentication on every read, even for public packages — so this request was always going to fail, independent of the artifact's actual state. The request signature (--netrc-optional,-A 'CocoaPods/1.16.2 cocoapods-downloader/2.1', no-H Authorization) is unmistakably CocoaPods' own built-in downloader, not the custom Bearer-token curl insidepatched_ios_artifacts.rb'sdownload_rncore_tarball.What's still unexplained (the actual open bug)
The resolver correctly finds the hash-matching
0.86.0-1version (I can see it decided to use it — the URL, group naming, and version-suffix format all match its design exactly), but somewhere between "resolver says use this artifact" and "download the tarball," the authenticated download path gets bypassed and CocoaPods' plain downloader takes over instead — which can never work against this registry. I couldn't pin the exact mechanism from static analysis (candidates: an exception insidePatchedIOSArtifacts.resolve'sbunsubprocess swallowed silently, a stalePods/Local Podspecs/React-Core-prebuilt.podspec.jsoncache, or a require/timing issue specific to how rock's CocoaPods install sequence loads the podspec).Recommended next step: this is worth a direct question to mateuuszzzzz — specifically whether
download_rncore_tarball's local-download-then-file://-wrap flow has been verified to actually run underrock's pod install (vs. plainbundle exec pod installfromscripts/pod-install.sh, which is what's exercised in local dev and may be the only path that's been manually tested).jakubstec commentedon Aug 11, 2026
looking into it right now!
mateuuszzzzz commentedon Aug 11, 2026
I’ve prepared two PRs to align our Ruby scripts that apply hot patches to artifact-related code with the changes introduced in RN 0.86. I’ve also introduced a few improvements to make the scripts less error-prone.
App: #98293
Mobile-Expensify: https://github.com/Expensify/Mobile-Expensify/pull/14052
roryabraham commentedon Aug 11, 2026
This is fixed
[-]Investigate workflow job failing on main: build (New Expensify Dev, DebugDevelopment, false)[/-][+][Due for payment 2026-08-19] Investigate workflow job failing on main: build (New Expensify Dev, DebugDevelopment, false)[/+]melvin-bot commentedon Aug 12, 2026
@roryabraham @mateuuszzzzz
The solution for this issue has been 🚀 deployed to production 🚀 in version 9.4.52-11 and is now subject to a 7-day regression period 📆. Here is the list of pull requests that resolve this issue:
If no regressions arise, payment will be issued on 2026-08-19. 🎊
The following checklist (instructions) will need to be completed before the issue can be closed. Please copy/paste the BugZero Checklist from here into a new comment on this GH and complete it. If you have the K2 extension, you can simply click: [this button]. If no checklist is needed for this issue, you can click: [no checklist button]
melvin-bot commentedon Aug 13, 2026
If you are the assigned CME please investigate whether the linked PR caused a regression and leave a comment with the results.
If a regression has occurred and you are the assigned CM follow the instructions here.
If this regression could have been avoided please consider also proposing a recommendation to the PR checklist so that we can avoid it in the future.