backfill: default to grabbing edge blobs too

Commit 302aff09223f (backfill: accept revision arguments, 2026-03-26) added
support for accepting revision arguments to backfill.  This allows users
to do things like

   git backfill --remotes ^v2.3.0

and then run many commands without triggering on-demand downloads of
blobs.  However, if they have topics based on v2.3.0, they will likely
still trigger on-demand downloads.  Consider, for example, the command

   git log -p v2.3.0..topic

This would still trigger on-demand blob loadings after the backfill
command above, because the commit(s) with A as a parent will need to
diff against the blobs in A.  In fact, multiple commands need blobs from
the lower boundary of the revision range:

   * git log -p A..B                # After backfill A..B
   * git replay --onto TARGET A..B  # After backfill TARGET^! A..B
   * git checkout A && git merge B  # After backfill A...B

Add an extra --[no-]include-edges flag to allow grabbing blobs from
edge commits.  Since the point of backfill is to prevent on-demand blob
loading and these are common commands, default to --include-edges.

Signed-off-by: Elijah Newren <newren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

diff --git a/Documentation/git-backfill.adoc b/Documentation/git-backfill.adoc
index bf26d7694f402d27b78ba664ca67c9155a392f50..c0a3b80615e0342a154877551dc601431de5b7c5 100644 (file)
--- a/Documentation/git-backfill.adoc
+++ b/Documentation/git-backfill.adoc
@@ -9,7 +9,7 @@ git-backfill - Download missing objects in a partial clone
 SYNOPSIS
 --------
 [synopsis]
-git backfill [--min-batch-size=<n>] [--[no-]sparse] [<revision-range>]
+git backfill [--min-batch-size=<n>] [--[no-]sparse] [--[no-]include-edges] [<revision-range>]
 
 DESCRIPTION
 -----------
@@ -63,6 +63,13 @@ OPTIONS
        current sparse-checkout. If the sparse-checkout feature is enabled,
        then `--sparse` is assumed and can be disabled with `--no-sparse`.
 
+`--include-edges`::
+`--no-include-edges`::
+       Include blobs from boundary commits in the backfill.  Useful in
+       preparation for commands like `git log -p A..B` or `git replay
+       --onto TARGET A..B`, where A..B normally excludes A but you need
+       the blobs from A as well.  `--include-edges` is the default.
+
 `<revision-range>`::
        Backfill only blobs reachable from commits in the specified
        revision range.  When no _<revision-range>_ is specified, it

diff --git a/builtin/backfill.c b/builtin/backfill.c
index e934d360fdd9d0b2aaf0aa4473df2bee47a5e002..7ffab2ea74f5cc7ed98587fa8e26f3746d18a163 100644 (file)
--- a/builtin/backfill.c
+++ b/builtin/backfill.c
@@ -26,7 +26,7 @@
 #include "path-walk.h"
 
 static const char * const builtin_backfill_usage[] = {
-       N_("git backfill [--min-batch-size=<n>] [--[no-]sparse] [<revision-range>]"),
+       N_("git backfill [--min-batch-size=<n>] [--[no-]sparse] [--[no-]include-edges] [<revision-range>]"),
        NULL
 };
 
@@ -35,6 +35,7 @@ struct backfill_context {
        struct oid_array current_batch;
        size_t min_batch_size;
        int sparse;
+       int include_edges;
        struct rev_info revs;
 };
 
@@ -116,6 +117,8 @@ static int do_backfill(struct backfill_context *ctx)
        /* Walk from HEAD if otherwise unspecified. */
        if (!ctx->revs.pending.nr)
                add_head_to_pending(&ctx->revs);
+       if (ctx->include_edges)
+               ctx->revs.edge_hint = 1;
 
        info.blobs = 1;
        info.tags = info.commits = info.trees = 0;
@@ -143,12 +146,15 @@ int cmd_backfill(int argc, const char **argv, const char *prefix, struct reposit
                .min_batch_size = 50000,
                .sparse = -1,
                .revs = REV_INFO_INIT,
+               .include_edges = 1,
        };
        struct option options[] = {
                OPT_UNSIGNED(0, "min-batch-size", &ctx.min_batch_size,
                             N_("Minimum number of objects to request at a time")),
                OPT_BOOL(0, "sparse", &ctx.sparse,
                         N_("Restrict the missing objects to the current sparse-checkout")),
+               OPT_BOOL(0, "include-edges", &ctx.include_edges,
+                        N_("Include blobs from boundary commits in the backfill")),
                OPT_END(),
        };
        struct repo_config_values *cfg = repo_config_values(the_repository);

diff --git a/t/t5620-backfill.sh b/t/t5620-backfill.sh
index f3b5e39493677b0daa917d1041b11d07c9cb0035..94f35ce19016716b2ca8c22faffaa8a05fb3e7a6 100755 (executable)
--- a/t/t5620-backfill.sh
+++ b/t/t5620-backfill.sh
@@ -257,11 +257,12 @@ test_expect_success 'backfill with revision range' '
        git -C backfill-revs rev-list --quiet --objects --missing=print HEAD >missing &&
        test_line_count = 48 missing &&
 
-       git -C backfill-revs backfill HEAD~2..HEAD &&
+       GIT_TRACE2_EVENT="$(pwd)/backfill-trace" git -C backfill-revs backfill HEAD~2..HEAD &&
 
-       # 30 objects downloaded.
+       # 36 objects downloaded, 12 still missing
+       test_trace2_data promisor fetch_count 36 <backfill-trace &&
        git -C backfill-revs rev-list --quiet --objects --missing=print HEAD >missing &&
-       test_line_count = 18 missing
+       test_line_count = 12 missing
 '
 
 test_expect_success 'backfill with revisions over stdin' '
@@ -279,11 +280,12 @@ test_expect_success 'backfill with revisions over stdin' '
        ^HEAD~2
        EOF
 
-       git -C backfill-revs backfill --stdin <in &&
+       GIT_TRACE2_EVENT="$(pwd)/backfill-trace" git -C backfill-revs backfill --stdin <in &&
 
-       # 30 objects downloaded.
+       # 36 objects downloaded, 12 still missing
+       test_trace2_data promisor fetch_count 36 <backfill-trace &&
        git -C backfill-revs rev-list --quiet --objects --missing=print HEAD >missing &&
-       test_line_count = 18 missing
+       test_line_count = 12 missing
 '
 
 test_expect_success 'backfill with prefix pathspec' '
@@ -398,6 +400,102 @@ test_expect_success 'backfill with --since' '
        test_line_count = 6 missing
 '
 
+test_expect_success 'backfill range with include-edges enables fetch-free git-log' '
+       git clone --no-checkout --filter=blob:none      \
+               --single-branch --branch=main           \
+               "file://$(pwd)/srv.bare" backfill-log &&
+
+       # Backfill the range with default include edges.
+       git -C backfill-log backfill HEAD~2..HEAD &&
+
+       # git log -p needs edge blobs for the "before" side of
+       # diffs.  With edge inclusion, all needed blobs are local.
+       GIT_TRACE2_EVENT="$(pwd)/log-trace" git \
+               -C backfill-log log -p HEAD~2..HEAD >log-output &&
+
+       # No promisor fetches should have been needed.
+       ! grep "fetch_count" log-trace
+'
+
+test_expect_success 'backfill range without include edges causes on-demand fetches in git-log' '
+       git clone --no-checkout --filter=blob:none      \
+               --single-branch --branch=main           \
+               "file://$(pwd)/srv.bare" backfill-log-no-bdy &&
+
+       # Backfill WITHOUT include edges -- file.3 v1 blobs are missing.
+       git -C backfill-log-no-bdy backfill --no-include-edges HEAD~2..HEAD &&
+
+       # git log -p HEAD~2..HEAD computes diff of commit 7 against
+       # commit 6.  It needs file.3 v1 (the "before" side), which was
+       # not backfilled.  This triggers on-demand promisor fetches.
+       GIT_TRACE2_EVENT="$(pwd)/log-no-bdy-trace" git \
+               -C backfill-log-no-bdy log -p HEAD~2..HEAD >log-output &&
+
+       grep "fetch_count" log-no-bdy-trace
+'
+
+test_expect_success 'backfill range enables fetch-free replay' '
+       # Create a repo with a branch to replay.
+       git init replay-src &&
+       (
+               cd replay-src &&
+               git config uploadpack.allowfilter 1 &&
+               git config uploadpack.allowanysha1inwant 1 &&
+               test_commit base &&
+               git checkout -b topic &&
+               test_commit topic-change &&
+               git checkout main &&
+               test_commit main-change
+       ) &&
+       git clone --bare --filter=blob:none \
+               "file://$(pwd)/replay-src" replay-dest.git &&
+
+       # Backfill the replay range: --onto main, replaying topic~1..topic.
+       # For replay, we need TARGET^! plus the range.
+       main_oid=$(git -C replay-dest.git rev-parse main) &&
+       topic_oid=$(git -C replay-dest.git rev-parse topic) &&
+       base_oid=$(git -C replay-dest.git rev-parse topic~1) &&
+       git -C replay-dest.git backfill \
+               "$main_oid^!" "$base_oid..$topic_oid" &&
+
+       # Now replay should complete without any promisor fetches.
+       GIT_TRACE2_EVENT="$(pwd)/replay-trace" git -C replay-dest.git \
+               replay --onto main topic~1..topic >replay-out &&
+
+       ! grep "fetch_count" replay-trace
+'
+
+test_expect_success 'backfill enables fetch-free merge' '
+       # Create a repo with two branches to merge.
+       git init merge-src &&
+       (
+               cd merge-src &&
+               git config uploadpack.allowfilter 1 &&
+               git config uploadpack.allowanysha1inwant 1 &&
+               test_commit merge-base &&
+               git checkout -b side &&
+               test_commit side-change &&
+               git checkout main &&
+               test_commit main-side-change
+       ) &&
+       git clone --filter=blob:none \
+               "file://$(pwd)/merge-src" merge-dest &&
+
+       # The clone checked out main, fetching its blobs.
+       # Backfill the three endpoint commits needed for merge.
+       main_oid=$(git -C merge-dest rev-parse origin/main) &&
+       side_oid=$(git -C merge-dest rev-parse origin/side) &&
+       mbase=$(git -C merge-dest merge-base origin/main origin/side) &&
+       git -C merge-dest backfill --no-include-edges \
+               "$main_oid^!" "$side_oid^!" "$mbase^!" &&
+
+       # Merge should complete without promisor fetches.
+       GIT_TRACE2_EVENT="$(pwd)/merge-trace" git -C merge-dest \
+               merge origin/side -m "test merge" &&
+
+       ! grep "fetch_count" merge-trace
+'
+
 . "$TEST_DIRECTORY"/lib-httpd.sh
 start_httpd