moraine 0.3.2

A SlateDB-backed catalog for DuckLake
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251

//! Concurrent-commit behavior through the public API: benign races
//! retry internally; true conflicts surface typed.

use moraine::Error;

use crate::fixtures::{col, datafile, seeded};

#[tokio::test(flavor = "multi_thread")]
async fn disjoint_table_ddl_both_succeed() {
    let (catalog, _s, a, b) = seeded().await;
    let c1 = catalog.clone();
    let c2 = catalog.clone();
    let t1 = tokio::spawn(async move {
        c1.commit(move |tx| tx.add_column(a, &col("a1")).map(|_| ()))
            .await
    });
    let t2 = tokio::spawn(async move {
        c2.commit(move |tx| tx.add_column(b, &col("b1")).map(|_| ()))
            .await
    });
    t1.await.unwrap().unwrap();
    t2.await.unwrap().unwrap();

    let head = catalog.snapshot().await.unwrap();
    assert_eq!(head.columns_of(a).len(), 2);
    assert_eq!(head.columns_of(b).len(), 2);
    catalog.close().await.unwrap();
}

#[tokio::test(flavor = "multi_thread")]
async fn same_table_ddl_races_serialize_or_conflict() {
    let (catalog, _s, a, _b) = seeded().await;
    // Whether a round serializes or conflicts depends on scheduling; the
    // classifier itself is pinned by the `transaction::operations` unit tests. This
    // test only requires that races never corrupt state.
    let mut added = Vec::new();
    for round in 0..20 {
        let c1 = catalog.clone();
        let c2 = catalog.clone();
        let n1 = format!("c1_{round}");
        let n2 = format!("c2_{round}");
        let t1 = tokio::spawn({
            let n1 = n1.clone();
            async move {
                c1.commit(move |tx| tx.add_column(a, &col(&n1)).map(|_| ()))
                    .await
            }
        });
        let t2 = tokio::spawn({
            let n2 = n2.clone();
            async move {
                c2.commit(move |tx| tx.add_column(a, &col(&n2)).map(|_| ()))
                    .await
            }
        });
        let r1 = t1.await.unwrap();
        let r2 = t2.await.unwrap();
        for (r, name) in [(r1, n1), (r2, n2)] {
            match r {
                Ok(_) => added.push(name),
                Err(Error::CommitConflict(_)) => {}
                Err(other) => panic!("unexpected error: {other}"),
            }
        }
    }
    let head = catalog.snapshot().await.unwrap();
    let names: std::collections::HashSet<String> =
        head.columns_of(a).into_iter().map(|c| c.name).collect();
    for name in &added {
        assert!(
            names.contains(name),
            "successfully committed column {name} missing from table a"
        );
    }
    assert_eq!(
        names.len(),
        added.len() + 1,
        "every successful add present exactly once, plus the seeded column"
    );
    catalog.close().await.unwrap();
}

#[tokio::test(flavor = "multi_thread")]
async fn same_name_create_race_yields_one_table() {
    let (catalog, s, _a, _b) = seeded().await;
    let c1 = catalog.clone();
    let c2 = catalog.clone();
    let t1 = tokio::spawn(async move {
        c1.commit(move |tx| tx.create_table(s, "orders", &[col("x")]).map(|_| ()))
            .await
    });
    let t2 = tokio::spawn(async move {
        c2.commit(move |tx| tx.create_table(s, "orders", &[col("x")]).map(|_| ()))
            .await
    });
    let results = [t1.await.unwrap(), t2.await.unwrap()];
    let oks = results.iter().filter(|r| r.is_ok()).count();
    assert_eq!(oks, 1);
    assert!(
        results
            .iter()
            .any(|r| matches!(r, Err(Error::AlreadyExists(_) | Error::CommitConflict(_))))
    );
    // Exactly one live table named "orders" either way.
    let head = catalog.snapshot().await.unwrap();
    let orders: Vec<_> = head
        .tables_in(s)
        .into_iter()
        .filter(|t| t.name == "orders")
        .collect();
    assert_eq!(orders.len(), 1);
    catalog.close().await.unwrap();
}

#[tokio::test(flavor = "multi_thread")]
async fn counters_never_regress_or_collide_under_concurrency() {
    let (catalog, s, _a, _b) = seeded().await;
    let mut handles = Vec::new();
    for i in 0..8 {
        let c = catalog.clone();
        let name = format!("t{i}");
        handles.push(tokio::spawn(async move {
            c.commit(move |tx| tx.create_table(s, &name, &[col("x")]).map(|_| ()))
                .await
        }));
    }
    for h in handles {
        h.await.unwrap().unwrap();
    }

    let head = catalog.snapshot().await.unwrap();
    let mut ids: Vec<u64> = head.tables_in(s).iter().map(|t| t.id.get()).collect();
    ids.sort_unstable();
    let before = ids.len();
    ids.dedup();
    assert_eq!(ids.len(), before, "table ids must be unique");
    assert_eq!(before, 10, "2 seeded + 8 concurrent");
    catalog.close().await.unwrap();
}

#[tokio::test(flavor = "multi_thread")]
async fn same_table_appends_both_land_with_dense_row_ids() {
    let (catalog, _s, a, _b) = seeded().await;
    let c1 = catalog.clone();
    let c2 = catalog.clone();
    let t1 = tokio::spawn(async move {
        c1.commit(move |tx| tx.register_data_file(a, datafile(100), &[]).map(|_| ()))
            .await
    });
    let t2 = tokio::spawn(async move {
        c2.commit(move |tx| tx.register_data_file(a, datafile(50), &[]).map(|_| ()))
            .await
    });
    t1.await.unwrap().unwrap();
    t2.await.unwrap().unwrap();

    let head = catalog.snapshot().await.unwrap();
    let files = head.data_files_of(a);
    assert_eq!(files.len(), 2);
    let mut starts: Vec<(u64, u64)> = files
        .iter()
        .map(|f| {
            (
                f.row_id_start.expect("verb-registered files carry a start"),
                f.record_count,
            )
        })
        .collect();
    starts.sort_unstable();

    // Dense, disjoint ranges regardless of which commit won the race.
    assert_eq!(starts[0].0, 0);
    assert_eq!(starts[1].0, starts[0].1);
    assert_eq!(head.table_stats(a).unwrap().next_row_id, 150);
    catalog.close().await.unwrap();
}

#[tokio::test(flavor = "multi_thread")]
async fn append_vs_drop_is_a_real_error() {
    let (catalog, s, a, _b) = seeded().await;
    let c1 = catalog.clone();
    let c2 = catalog.clone();
    let t1 = tokio::spawn(async move {
        c1.commit(move |tx| tx.register_data_file(a, datafile(10), &[]).map(|_| ()))
            .await
    });
    let t2 = tokio::spawn(async move { c2.commit(move |tx| tx.drop_table(a)).await });
    let r1 = t1.await.unwrap();
    let r2 = t2.await.unwrap();

    // Serialized is fine; a genuine race surfaces CommitConflict or
    // NotFound on the loser.
    for r in [&r1, &r2] {
        match r {
            Ok(_) | Err(Error::CommitConflict(_) | Error::NotFound(_)) => {}
            Err(other) => panic!("unexpected error: {other}"),
        }
    }

    // Whatever the outcome mix, the head must be consistent with it.
    let head = catalog.snapshot().await.unwrap();
    let live = head.tables_in(s).into_iter().any(|t| t.id == a);
    if r2.is_ok() {
        assert!(!live, "drop committed, so the table must be gone");
    } else {
        assert!(live, "drop failed, so the table must survive");
        if r1.is_ok() {
            assert_eq!(head.data_files_of(a).len(), 1);
        }
    }
    catalog.close().await.unwrap();
}

/// An options-only commit re-validates its scope against a racing drop:
/// whichever order the two land in, a dropped table never keeps a live
/// option record (which nothing could ever remove again).
#[tokio::test(flavor = "multi_thread")]
async fn option_set_vs_drop_leaves_no_orphaned_option() {
    use moraine::OptionScope;

    for _ in 0..10 {
        let (catalog, s, a, _b) = seeded().await;
        let c1 = catalog.clone();
        let c2 = catalog.clone();
        let t1 = tokio::spawn(async move {
            c1.commit(move |tx| tx.set_option(OptionScope::Table(a), "k", "v"))
                .await
        });
        let t2 = tokio::spawn(async move { c2.commit(move |tx| tx.drop_table(a)).await });
        let set = t1.await.unwrap();
        let dropped = t2.await.unwrap();

        for r in [&set, &dropped] {
            match r {
                Ok(_) | Err(Error::CommitConflict(_) | Error::NotFound(_)) => {}
                Err(other) => panic!("unexpected error: {other}"),
            }
        }

        let head = catalog.snapshot().await.unwrap();
        let live = head.tables_in(s).into_iter().any(|t| t.id == a);
        if !live {
            assert_eq!(
                head.option(OptionScope::Table(a), "k"),
                None,
                "dropped table kept an orphaned option record"
            );
        }
        catalog.close().await.unwrap();
    }
}