Artifact [4d2353d61e]

Artifact 4d2353d61eff5f70cb1438dff9996e05f6ab7c95bc0255b09f7b375d172b14ef:

• File repo_timeline_test.go — part of check-in [6273a0750d] at 2026-07-22 23:03:24 on branch trunk — fix: correct go.mod module path to go-libfossil (#76) (#117)

  * fix: correct go.mod module path to go-libfossil (#76)

  The module directive still declared github.com/danmestas/libfossil after the repository was renamed to go-libfossil. pkg.go.dev/godoc scraping matches the declared module path against its hosting path, so the stale path broke SDK doc generation.

  Rename the module path across all four in-repo modules (root, the two db/driver submodules, observer/otel, and docs/site) and every internal import, plus the release workflow, the doc-generation script, and the docs.

  No tag has been published under the go-libfossil path yet, so the cross-module requires cannot resolve from the proxy. Add replace directives pointing the in-repo requires at their local directories so GOWORK=off builds and the CI matrix resolve without a published version. These replaces are ignored by downstream consumers and can be dropped once new-path tags are cut.

  * Guard releases against replace directives; regenerate SDK docs

  The go-libfossil rename bridges the in-repo modules with `replace` directives. Go refuses a main module that carries them, so a tag cut while they are present breaks `go install .../cmd/libfossil@<tag>` and every `@version` fetch of the driver submodules. Add a Release workflow step that fails the job before anything is published if any tracked go.mod still contains a replace directive, so the bridge is a CI-enforced TODO rather than a silent landmine. The step comment records what unblocks a release: re-tag the submodules with the corrected module paths, bump the root requires, then drop the replace blocks.

  Regenerate the SDK reference docs. The previous pass only substituted URLs over the pre-existing generated files, leaving stale signatures (Repo.Timeline was still documented as taking LogOpts rather than TimelineOpts). Note that gomarkdoc must be run outside a git worktree — inside one it silently strips every source link.

  Also reword the README rename note, which read as though the import path had not changed, and re-sort the ncruces require block. (user: [email protected], size: 9243) [annotate] [blame] [check-ins using]

package libfossil

import (
	"fmt"
	"path/filepath"
	"testing"
	"time"

	_ "github.com/danmestas/go-libfossil/internal/testdriver"
)

// TestTimelineEnumeratesEveryEvent is the acceptance criterion for the bug
// this issue fixes: Timeline with no type filter must return every row of
// event, including check-ins that are not first-parent ancestors of any
// single start rid — the thing the old Timeline (now Ancestry) could
// never see.
func TestTimelineEnumeratesEveryEvent(t *testing.T) {
	dir := t.TempDir()
	r, err := Create(filepath.Join(dir, "test.fossil"), CreateOpts{User: "test"})
	if err != nil {
		t.Fatal(err)
	}
	defer r.Close()

	rid1, _, err := r.Commit(CommitOpts{
		Files:   []FileToCommit{{Name: "a.txt", Content: []byte("v1")}},
		Comment: "trunk root", User: "test",
	})
	if err != nil {
		t.Fatal(err)
	}
	if _, _, err := r.Commit(CommitOpts{
		ParentID: rid1,
		Files:    []FileToCommit{{Name: "a.txt", Content: []byte("v2")}},
		Comment:  "trunk tip", User: "test",
	}); err != nil {
		t.Fatal(err)
	}
	branchRID, branchUUID, err := r.Commit(CommitOpts{
		ParentID: rid1,
		Files:    []FileToCommit{{Name: "a.txt", Content: []byte("v3")}},
		Comment:  "sibling branch head", User: "test",
		Tags: []TagSpec{{Name: "branch", Value: "feature-x"}},
	})
	if err != nil {
		t.Fatal(err)
	}
	if branchRID <= 0 {
		t.Fatal("branch commit returned rid <= 0")
	}

	var wantCount int
	if err := r.DB().QueryRow("SELECT count(*) FROM event").Scan(&wantCount); err != nil {
		t.Fatal(err)
	}

	entries, err := r.Timeline(TimelineOpts{})
	if err != nil {
		t.Fatal(err)
	}
	if len(entries) != wantCount {
		t.Fatalf("Timeline returned %d entries, want %d (SELECT count(*) FROM event)", len(entries), wantCount)
	}

	found := false
	for _, e := range entries {
		if e.UUID == branchUUID {
			found = true
		}
	}
	if !found {
		t.Fatalf("sibling branch head %s missing from Timeline output", branchUUID)
	}
}

// TestTimelineTypeFilter covers Type=EventKindCheckin returning exactly
// the check-in rows.
func TestTimelineTypeFilter(t *testing.T) {
	dir := t.TempDir()
	r, err := Create(filepath.Join(dir, "test.fossil"), CreateOpts{User: "test"})
	if err != nil {
		t.Fatal(err)
	}
	defer r.Close()

	if _, _, err := r.Commit(CommitOpts{
		Files:   []FileToCommit{{Name: "a.txt", Content: []byte("v1")}},
		Comment: "checkin", User: "test",
	}); err != nil {
		t.Fatal(err)
	}

	var wantCi int
	if err := r.DB().QueryRow("SELECT count(*) FROM event WHERE type='ci'").Scan(&wantCi); err != nil {
		t.Fatal(err)
	}

	entries, err := r.Timeline(TimelineOpts{Type: EventKindCheckin})
	if err != nil {
		t.Fatal(err)
	}
	if len(entries) != wantCi {
		t.Fatalf("Timeline(Type=ci) returned %d entries, want %d", len(entries), wantCi)
	}
	for _, e := range entries {
		if e.Kind != EventKindCheckin {
			t.Fatalf("entry kind = %q, want %q", e.Kind, EventKindCheckin)
		}
	}
}

// TestTimelineOrderingAndLimit covers (mtime DESC, rid DESC) ordering and
// that Limit both bounds the result and is reachable — the defining
// symptom of the original bug was a limit that never bound.
func TestTimelineOrderingAndLimit(t *testing.T) {
	dir := t.TempDir()
	r, err := Create(filepath.Join(dir, "test.fossil"), CreateOpts{User: "test"})
	if err != nil {
		t.Fatal(err)
	}
	defer r.Close()

	var parent int64
	var rids []int64
	for i := 0; i < 5; i++ {
		rid, _, err := r.Commit(CommitOpts{
			ParentID: parent,
			Files:    []FileToCommit{{Name: "a.txt", Content: []byte(fmt.Sprintf("v%d", i))}},
			Comment:  fmt.Sprintf("commit %d", i), User: "test",
			Time: time.Date(2024, 1, 15, 10, i, 0, 0, time.UTC),
		})
		if err != nil {
			t.Fatalf("commit %d: %v", i, err)
		}
		parent = rid
		rids = append(rids, rid)
	}

	entries, err := r.Timeline(TimelineOpts{Limit: 2})
	if err != nil {
		t.Fatal(err)
	}
	if len(entries) != 2 {
		t.Fatalf("len = %d, want 2", len(entries))
	}
	// Newest first: the last two commits, newest rid first.
	if entries[0].RID != rids[4] || entries[1].RID != rids[3] {
		t.Fatalf("order = [%d, %d], want [%d, %d]", entries[0].RID, entries[1].RID, rids[4], rids[3])
	}
}

// TestTimelinePaginationExactlyOnce covers the acceptance criterion that
// paginating the full set yields each event exactly once, including when
// several events share an identical mtime.
func TestTimelinePaginationExactlyOnce(t *testing.T) {
	dir := t.TempDir()
	r, err := Create(filepath.Join(dir, "test.fossil"), CreateOpts{User: "test"})
	if err != nil {
		t.Fatal(err)
	}
	defer r.Close()

	sameTime := time.Date(2024, 1, 15, 10, 0, 0, 0, time.UTC)
	var parent int64
	var allUUIDs []string
	for i := 0; i < 6; i++ {
		rid, uuid, err := r.Commit(CommitOpts{
			ParentID: parent,
			Files:    []FileToCommit{{Name: "a.txt", Content: []byte(fmt.Sprintf("v%d", i))}},
			Comment:  fmt.Sprintf("commit %d", i), User: "test",
			Time: sameTime, // every commit shares the exact same mtime
		})
		if err != nil {
			t.Fatalf("commit %d: %v", i, err)
		}
		parent = rid
		allUUIDs = append(allUUIDs, uuid)
	}

	const pageSize = 2
	// Exactly ceil(n/pageSize) pages: a fixed bound, not a "while non-empty"
	// loop, so a cursor that fails to advance fails this test immediately
	// instead of hanging it.
	maxPages := (len(allUUIDs) + pageSize - 1) / pageSize
	seen := make(map[string]int)
	var total int
	var after Cursor
	for page := 0; page < maxPages; page++ {
		entries, err := r.Timeline(TimelineOpts{Limit: pageSize, After: after})
		if err != nil {
			t.Fatalf("page %d: %v", page, err)
		}
		if len(entries) == 0 {
			t.Fatalf("page %d returned 0 entries after only %d of %d events seen — cursor did not advance", page, total, len(allUUIDs))
		}
		for _, e := range entries {
			seen[e.UUID]++
			total++
		}
		after = entries[len(entries)-1].Cursor
	}

	if total != len(allUUIDs) {
		t.Fatalf("paginated total = %d across exactly %d pages, want %d", total, maxPages, len(allUUIDs))
	}
	for _, uuid := range allUUIDs {
		if seen[uuid] != 1 {
			t.Fatalf("event %s seen %d times across %d pages, want exactly 1", uuid, seen[uuid], maxPages)
		}
	}

	// The defining symptom of a broken cursor is non-termination: every
	// row enumerated correctly, but the cursor never signals exhaustion.
	// One more call past the last real page must come back empty.
	final, err := r.Timeline(TimelineOpts{Limit: pageSize, After: after})
	if err != nil {
		t.Fatalf("final page: %v", err)
	}
	if len(final) != 0 {
		t.Fatalf("pagination did not terminate: page after the last expected page returned %d entries, want 0: %v", len(final), final)
	}
}

// TestTimelinePaginationSubMillisecondBoundary is the public-API-level
// regression test for a cursor predicate that only distinguishes rows via
// a tolerance window instead of treating rid as a true tie-break at exact
// mtime equality: such a predicate can skip a row, duplicate another, or
// fail to advance the cursor entirely when two rows are close together
// but not equal. rid (checkin/insertion order) is deliberately decoupled
// from mtime order here, since a predicate bug that only shows up when
// they disagree would otherwise pass unnoticed.
func TestTimelinePaginationSubMillisecondBoundary(t *testing.T) {
	dir := t.TempDir()
	r, err := Create(filepath.Join(dir, "test.fossil"), CreateOpts{User: "test"})
	if err != nil {
		t.Fatal(err)
	}
	defer r.Close()

	const n = 6
	const pageSize = 2
	base := time.Date(2024, 1, 15, 10, 0, 0, 0, time.UTC)
	var parent int64
	var allUUIDs []string
	for i := 0; i < n; i++ {
		ts := base.Add(time.Duration(n-1-i) * 3 * time.Millisecond) // 3ms apart, descending as rid ascends
		rid, uuid, err := r.Commit(CommitOpts{
			ParentID: parent,
			Files:    []FileToCommit{{Name: "a.txt", Content: []byte(fmt.Sprintf("v%d", i))}},
			Comment:  fmt.Sprintf("commit %d", i), User: "test",
			Time: ts,
		})
		if err != nil {
			t.Fatalf("commit %d: %v", i, err)
		}
		parent = rid
		allUUIDs = append(allUUIDs, uuid)
	}

	maxPages := (n + pageSize - 1) / pageSize // exactly ceil(n/pageSize) = 3
	seen := make(map[string]int)
	var total int
	var after Cursor
	for page := 0; page < maxPages; page++ {
		entries, err := r.Timeline(TimelineOpts{Limit: pageSize, After: after})
		if err != nil {
			t.Fatalf("page %d: %v", page, err)
		}
		if len(entries) == 0 {
			t.Fatalf("page %d returned 0 entries after only %d of %d events seen — cursor did not advance", page, total, n)
		}
		for _, e := range entries {
			seen[e.UUID]++
			total++
		}
		after = entries[len(entries)-1].Cursor
	}

	if total != n {
		t.Fatalf("paginated total = %d across exactly %d pages, want %d", total, maxPages, n)
	}
	for _, uuid := range allUUIDs {
		if seen[uuid] != 1 {
			t.Fatalf("event %s seen %d times across %d pages, want exactly 1", uuid, seen[uuid], maxPages)
		}
	}

	// Non-termination is the actual user-facing failure mode: every row
	// enumerated correctly but the cursor never signals exhaustion. Assert
	// the page past the last expected one comes back empty.
	final, err := r.Timeline(TimelineOpts{Limit: pageSize, After: after})
	if err != nil {
		t.Fatalf("final page: %v", err)
	}
	if len(final) != 0 {
		t.Fatalf("pagination did not terminate: page after the last expected page returned %d entries, want 0: %v", len(final), final)
	}
}