Thread (5 messages) 5 messages, 3 authors, 2025-10-29

Re: [PATCH v2] commit-graph: add new config for changed-paths & recommend it in scalar

From: SZEDER Gábor <hidden>
Date: 2025-10-29 21:04:29

On Fri, Oct 17, 2025 at 08:58:59PM +0000, Emily Yang via GitGitGadget wrote:
From: Emily Yang <redacted>

The changed-path Bloom filters feature has proven stable and reliable
over several years of use, delivering significant performance
improvement for file history computation in large monorepos. Currently
a user can opt-in to writing the changed-path Bloom filters using the
"--changed-paths" option to "git commit-graph write". The filters will
be persisted until the user drops the filters using the
"--no-changed-paths" option. For this functionality, refer to 0087a87ba8
(commit-graph: persist existence of changed-paths, 2020-07-01).

Large monorepos using Git's background maintenance to build and update
commit-graph files could use an easy switch to enable this feature
without a foreground computation. In this commit, we're proposing a new
config option "commitGraph.changedPaths":

* If "true", "git commit-graph write" will write Bloom filters,
  equivalent to passing "--changed-paths";
* If "false" or "unset", Bloom filters will be written during "git
  commit-graph write" only if the filters already exist in the current
  commit-graph file. This matches the default behaviour of "git
  commit-graph write" without any "--[no-]changed-paths" option. Note
  "false" can disable a previous "true" config value but doesn't imply
  "--no-changed-paths".
So if the commit-graph contains changed path Bloom filters, and the
user takes the effort, and explicitly sets this config variable to
false, then Git will just ignore that, and will continue to waste
resources to compute the Bloom filters?!  This doesn't seems like a
sensible behavior to me.

quoted hunk ↗ jump to hunk
diff --git a/t/t5318-commit-graph.sh b/t/t5318-commit-graph.sh
index 0b3404f58f..98c6910963 100755
--- a/t/t5318-commit-graph.sh
+++ b/t/t5318-commit-graph.sh
@@ -946,4 +946,48 @@ test_expect_success 'stale commit cannot be parsed when traversing graph' '
 	)
 '
 
+test_expect_success 'config commitGraph.changedPaths acts like --changed-paths' '
+	git init config-changed-paths &&
+	(
+		cd config-changed-paths &&
+
+		# commitGraph.changedPaths is not set and it should not write Bloom filters
+		test_commit first &&
+		GIT_PROGRESS_DELAY=0 git commit-graph write --reachable --progress 2>error &&
+		test_grep ! "Bloom filters" error &&
+
+		# Set commitGraph.changedPaths to true and it should write Bloom filters
+		test_commit second &&
+		git config commitGraph.changedPaths true &&
+		GIT_PROGRESS_DELAY=0 git commit-graph write --reachable --progress 2>error &&
+		test_grep "Bloom filters" error &&
+
+		# Add one more config commitGraph.changedPaths as false to disable the previous true config value
+		# It should still write Bloom filters due to existing filters
+		test_commit third &&
+		git config --add commitGraph.changedPaths false &&
+		GIT_PROGRESS_DELAY=0 git commit-graph write --reachable --progress 2>error &&
+		test_grep "Bloom filters" error &&
+
+		# commitGraph.changedPaths is still false and command line options should take precedence
+		test_commit fourth &&
+		GIT_PROGRESS_DELAY=0 git commit-graph write --no-changed-paths --reachable --progress 2>error &&
+		test_grep ! "Bloom filters" error &&
+		GIT_PROGRESS_DELAY=0 git commit-graph write --reachable --progress 2>error &&
+		test_grep ! "Bloom filters" error &&
+
+		# commitGraph.changedPaths is all cleared and then set to false again, command line options should take precedence
+		test_commit fifth &&
+		git config --unset-all commitGraph.changedPaths &&
+		git config commitGraph.changedPaths false &&
+		GIT_PROGRESS_DELAY=0 git commit-graph write --changed-paths --reachable --progress 2>error &&
+		test_grep "Bloom filters" error &&
+
+		# commitGraph.changedPaths is still false and it should write Bloom filters due to existing filters
+		test_commit sixth &&
+		GIT_PROGRESS_DELAY=0 git commit-graph write --reachable --progress 2>error &&
+		test_grep "Bloom filters" error
+	)
+'
The interaction of split commit-graphs and changed path Bloom filters
used to be buggy, even after attempts to fix it.  Therefore, I think
this config variable should be tested with split commit-graphs as
well.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help