From 80436e731e71400c999ffae1c15c3af527b92c07 Mon Sep 17 00:00:00 2001 From: mergify-ci-bot Date: Mon, 29 Jun 2026 16:25:10 +0000 Subject: [PATCH 1/2] chore: update changelog --- ...te-results-now-works-in-parallel-merge-queue-mode.mdx | 9 --------- 1 file changed, 9 deletions(-) delete mode 100644 src/content/changelog/2026-06-19-skip-intermediate-results-now-works-in-parallel-merge-queue-mode.mdx diff --git a/src/content/changelog/2026-06-19-skip-intermediate-results-now-works-in-parallel-merge-queue-mode.mdx b/src/content/changelog/2026-06-19-skip-intermediate-results-now-works-in-parallel-merge-queue-mode.mdx deleted file mode 100644 index 8db7bdd6eb..0000000000 --- a/src/content/changelog/2026-06-19-skip-intermediate-results-now-works-in-parallel-merge-queue-mode.mdx +++ /dev/null @@ -1,9 +0,0 @@ ---- -title: "skip_intermediate_results now works in parallel merge queue mode" -date: 2026-06-19 -tags: - - Merge Queue -description: "The skip_intermediate_results option, previously serial-only, now also speeds up merges in parallel mode within each scope's dependency chain." ---- - -When `skip_intermediate_results` is enabled, Mergify merges a pull request as soon as a later batch that includes it passes its checks, instead of waiting for the PR's own speculative check. Previously available only in serial mode, it now applies in parallel mode too — within each scope's dependency chain, a passing batch promotes its not-yet-merged ancestors so busy scopes get the same speedup serial queues already enjoy. From 044d6a27a47a9aa13e4f1d1633bc6bae45f4c6e0 Mon Sep 17 00:00:00 2001 From: Mehdi ABAAKOUK Date: Mon, 29 Jun 2026 17:57:05 +0200 Subject: [PATCH 2/2] docs(merge-queue): fix obsolete skip_intermediate_results parallel-mode note The note claimed skip_intermediate_results depends on serial mode's strict cumulative ordering and is not available in parallel mode. That is no longer true: the engine generalizes the mechanism to the parallel scope dependency graph, so a passing batch vouches for the earlier changes it was tested on top of, in serial and parallel modes alike. The real restriction is isolated mode, where batches are fully independent and the config is rejected outright. Update the batches anti-flake note and the queue-modes limitations bullet to scope the restriction to isolated mode, and fix the section intro that still asserted strict single-queue ordering. Change-Id: Ie80c5095b016f23b80087d81653758110d6a56ff --- src/content/docs/merge-queue/batches.mdx | 6 +++--- src/content/docs/merge-queue/queue-modes.mdx | 9 +++++---- 2 files changed, 8 insertions(+), 7 deletions(-) diff --git a/src/content/docs/merge-queue/batches.mdx b/src/content/docs/merge-queue/batches.mdx index 9b4c7b4ac8..b35c9bd940 100644 --- a/src/content/docs/merge-queue/batches.mdx +++ b/src/content/docs/merge-queue/batches.mdx @@ -607,9 +607,9 @@ need every intermediate state validated for compliance, or while debugging CI and want to see all failures. :::note - `skip_intermediate_results` depends on the strict cumulative ordering of serial - mode and is not available in [parallel - mode](/merge-queue/queue-modes#parallel-mode). + `skip_intermediate_results` works in serial and parallel modes. It is not supported in + [isolated mode](/merge-queue/queue-modes#isolated-mode): batches there are fully independent, + so a passing batch can't vouch for any other. ::: ## Important Considerations diff --git a/src/content/docs/merge-queue/queue-modes.mdx b/src/content/docs/merge-queue/queue-modes.mdx index f341f0a1e1..0b2e9a5663 100644 --- a/src/content/docs/merge-queue/queue-modes.mdx +++ b/src/content/docs/merge-queue/queue-modes.mdx @@ -466,8 +466,8 @@ parent batch to wait for. Batch failures are handled the same way as in the othe ## Compatibility and Limitations -Parallel and isolated modes change how the queue operates. Some features that rely on strict -single-queue ordering are not available: +Parallel and isolated modes change how the queue operates, so some features behave differently or +aren't available depending on the mode: - **Scopes are required in parallel mode.** You must configure `scopes.source` (either `files` or `manual`) so Mergify can tell which pull requests are independent. Serial and isolated modes do @@ -476,8 +476,9 @@ single-queue ordering are not available: - **`fast-forward` merge is not supported.** Because batches merge independently, Mergify needs to rebase them. Use `merge` or `rebase` as your `merge_method`. -- **`skip_intermediate_results` is not available.** This feature depends on the strict cumulative - ordering of serial mode. +- **`skip_intermediate_results` is not supported in isolated mode.** Isolated batches are fully + independent, so a passing batch can't vouch for any other. It works in serial and parallel modes, + where a passing batch vouches for the earlier changes it was tested on top of. - **`partition_rules` are not supported.** Partitions rely on serial ordering; use scopes instead.