commonware-glue 2026.7.0

Default constructions that span multiple primitives.
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
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
//! Gather a recent finalization to sync from.
//!
//! A node that is starting fresh (or recovering from far behind) needs a recent, trustworthy
//! finalization, the "floor", as the point to begin state sync from. It cannot trust any
//! single peer to name that point, so the [`Probe`] asks many peers and adopts the
//! highest valid finalization from `f + 1` distinct peer replies.
//!
//! # Protocol
//!
//! ## Solicit
//!
//! Once a floor subscriber appears, [`Probe`] broadcasts a `Request` to every
//! connected peer:
//!
//! ```text
//!                    +-- Request --> peer 1
//!                    |
//!   Probe -----------+-- Request --> peer 2
//!                    |
//!                    +-- Request --> peer 3
//!                    |
//!                    +-- Request --> peer 4
//! ```
//!
//! A subscription is the request to discover a floor. If all floor subscribers are dropped before
//! a floor is selected, discovery is cancelled. Attaching marshal after that makes the node a
//! source: it enters service without a cached floor. Callers that need a floor must keep a
//! subscription alive until it resolves and attach marshal only after consuming that floor.
//!
//! ## Collect and select
//!
//! Each peer answers with its own latest finalization (or nothing, if it has none). Every
//! response is verified against the certificate scheme for its epoch. At most one finalization is
//! counted per peer, so no single peer can inflate the sample on its own. Once `f + 1`
//! distinct peers have replied, the highest finalized round becomes the floor:
//!
//! ```text
//!   peer 1 --Response(view 10)-->\                 replies
//!   peer 2 --Response(view 12)--> +-> Probe {10, 12, 13}
//!   peer 3 --Response(view 13)-->/                        |
//!                                                             v
//!                                      sample reached, highest view becomes the floor: 13
//! ```
//!
//! A peer that sends an undecodable or unverifiable first finalization in a request round is
//! blocked. After a peer has already contributed a verified response for that round, later
//! messages from that peer are ignored before validation.
//!
//! ## Retry
//!
//! If too few peers reply, the collected responses are cleared and the request is re-issued after
//! a configurable `retry_timeout`. Retry is not required for safety. It is a liveness mechanism
//! for request rounds that fail to collect enough usable replies because messages were dropped,
//! peers were slow or offline, or a finalization's epoch could not yet be judged.
//!
//! ```text
//!   request --> collect --> sample reached? --yes--> highest floor
//!      ^                         |
//!      |                         no
//!      +----- clear + re-request +
//!            (retry_timeout elapsed)
//! ```
//!
//! # Why the sample is `f + 1`
//!
//! The `f` used here comes from the epoch-scoped committee for each accepted response
//! finalization, not from the initially registered peer set. That lets probe tolerate peer churn
//! after startup without lowering the sample below the epoch's fault bound.
//!
//! Assume at most `f` of the `n` participants in that epoch are faulty. In this protocol, `f` makes
//! no distinction between Byzantine and crashed nodes: a peer that does not answer and a peer that
//! answers adversarially both count against the same fault budget.
//!
//! A finalization is self-certifying: it carries a quorum certificate, so any one that verifies
//! proves the network truly finalized that block. Accepting a single peer's finalization is
//! therefore always *safe* (it names a real block), but it is not necessarily *recent*.
//!
//! That recency gap is the attack. A Byzantine peer can replay an old (but still valid)
//! finalization to drag a joining node's floor far behind the real tip of the chain, forcing it
//! to re-sync a huge range or to settle on a stale view.
//!
//! Under [`simplex`](commonware_consensus::simplex)'s synchrony assumptions, after one honest node
//! advances to a new view, other honest nodes may remain in the previous view until that transition
//! is delivered within the network-delay bound (`delta`). Waiting for `f + 1` replies guarantees at
//! least one honest response in the sample: at most `f` responders can be Byzantine, and crashed
//! nodes do not respond. The selected finalization is therefore at least as recent as that honest
//! response. Byzantine peers can still replay old certificates, but old certificates lose to newer
//! honest replies. If they report something higher, it must still be a valid finalization, so it is
//! a real finalized block rather than a rollback.
//!
//! If fewer than `f + 1` registered peers can answer for that epoch, probe cannot resolve that
//! request round and will retry. This is a liveness tradeoff, not a safety one: using the epoch's
//! `f + 1` threshold lets probe tolerate peer churn while preserving the assumption that every
//! completed sample includes at least one honest response from the relevant historical committee.
//!
//! [`Config::minimum_epoch`] bounds that historical search. A caller that initializes peers from a
//! known epoch can set it to that lower bound; responses from earlier epochs are ignored. Probe
//! still sizes each accepted sample from that finalization's epoch committee.
//!
//! ```text
//!   any f + 1 sample:
//!
//!     [ B   B  ...  B ]  [ H ]       at most f Byzantine
//!      \____ <= f ____/    \_ at least one honest response
//!
//!     choose max(valid finalizations)
//!       => floor is no older than the freshest honest reply in the sample
//! ```
//!
//! # Resource Bounds
//!
//! The actor retains at most one finalization candidate per peer per request round. Additional
//! messages from the same peer are ignored before validation, preserving correctness when network
//! delivery lags across request rounds without wasting certificate work or allowing one peer to
//! inflate a sample. The p2p channel supplies rate limiting and maximum encoded message size
//! enforcement.

mod actor;
pub use actor::{Config, Probe};

mod mailbox;
pub use mailbox::Mailbox;

mod wire;

#[cfg(test)]
mod test {
    use super::{wire, Config, Mailbox, Probe};
    use bytes::{Buf, BufMut};
    use commonware_actor::Feedback;
    use commonware_codec::{Encode, EncodeSize, Error as CodecError, Read, ReadExt as _, Write};
    use commonware_consensus::{
        marshal::{
            self,
            core::{Actor as MarshalActor, Mailbox as MarshalMailbox},
            resolver::p2p as marshal_resolver,
            standard::Standard,
            Start, Update,
        },
        simplex::{
            mocks::scheme::{self as scheme_mocks, Scheme as MockScheme},
            types::{Activity, Context as SimplexContext, Finalization, Finalize, Proposal},
        },
        types::{Epoch, FixedEpocher, Height, Round, View, ViewDelta},
        Block as ConsensusBlock, CertifiableBlock, Heightable, Reporter,
    };
    use commonware_cryptography::{
        certificate::{
            self, mocks::Fixture, Attestation, ConstantProvider, Provider, Scoped, Verification,
            Verifier,
        },
        ed25519,
        sha256::Digest as Sha256Digest,
        Digest as _, Digestible, Signer as _,
    };
    use commonware_macros::test_collect_traces;
    use commonware_p2p::{
        simulated::{Config as SimConfig, Link, Network, Oracle, Sender},
        Recipients, Sender as _,
    };
    use commonware_parallel::{Sequential, Strategy as ParallelStrategy};
    use commonware_runtime::{
        buffer::paged::CacheRef, deterministic, telemetry::traces::collector::TraceStorage, Clock,
        Handle, Metrics, Quota, Runner as _, Supervisor,
    };
    use commonware_storage::archive::immutable;
    use commonware_utils::{
        channel::oneshot, sync::Mutex, test_rng, Acknowledgement, NZDuration, NZUsize,
        NonZeroDuration, NZU16, NZU64,
    };
    use std::{
        collections::BTreeMap,
        num::{NonZeroU32, NonZeroU64},
        sync::Arc,
        time::Duration,
    };

    const NAMESPACE: &[u8] = b"_COMMONWARE_GLUE_PROBE_TEST";
    const EPOCH_LENGTH: NonZeroU64 = NZU64!(u64::MAX);
    const TEST_QUOTA: Quota = Quota::per_second(NonZeroU32::MAX);
    const LINK: Link = Link {
        latency: Duration::from_millis(10),
        jitter: Duration::from_millis(1),
        success_rate: 1.0,
    };
    const PROBE_CHANNEL: u64 = 0;
    const BACKFILL_CHANNEL: u64 = 1;

    type Scheme = MockScheme<ed25519::PublicKey>;
    type Variant = Standard<Block>;

    #[derive(Clone, Debug, PartialEq, Eq)]
    struct Block {
        context: SimplexContext<Sha256Digest, ed25519::PublicKey>,
        height: Height,
        digest: Sha256Digest,
    }

    impl Block {
        fn new(height: u64, digest_byte: u8) -> Self {
            Self {
                context: SimplexContext {
                    round: Round::new(Epoch::zero(), View::new(height)),
                    leader: ed25519::PrivateKey::from_seed(0).public_key(),
                    parent: (View::zero(), Sha256Digest::EMPTY),
                },
                height: Height::new(height),
                digest: Sha256Digest::from([digest_byte; 32]),
            }
        }
    }

    impl Write for Block {
        fn write(&self, buf: &mut impl BufMut) {
            self.context.write(buf);
            buf.put_u64(self.height.get());
            buf.put_slice(self.digest.as_ref());
        }
    }

    impl EncodeSize for Block {
        fn encode_size(&self) -> usize {
            self.context.encode_size() + 8 + 32
        }
    }

    impl Read for Block {
        type Cfg = ();

        fn read_cfg(buf: &mut impl Buf, _: &Self::Cfg) -> Result<Self, CodecError> {
            let context = SimplexContext::read(buf)?;
            let height = Height::new(buf.get_u64());
            let mut digest = [0u8; 32];
            buf.copy_to_slice(&mut digest);
            Ok(Self {
                context,
                height,
                digest: Sha256Digest::from(digest),
            })
        }
    }

    impl Digestible for Block {
        type Digest = Sha256Digest;

        fn digest(&self) -> Self::Digest {
            self.digest
        }
    }

    impl Heightable for Block {
        fn height(&self) -> Height {
            self.height
        }
    }

    impl ConsensusBlock for Block {
        fn parent(&self) -> Self::Digest {
            Sha256Digest::EMPTY
        }
    }

    impl CertifiableBlock for Block {
        type Context = SimplexContext<Sha256Digest, ed25519::PublicKey>;

        fn context(&self) -> Self::Context {
            self.context.clone()
        }
    }

    #[derive(Clone)]
    struct NoopReporter;

    impl Reporter for NoopReporter {
        type Activity = Update<Block>;

        fn report(&mut self, activity: Self::Activity) -> Feedback {
            if let Update::Block(_, ack) = activity {
                ack.acknowledge();
            }
            Feedback::Ok
        }
    }

    /// A certificate-scheme provider keyed by epoch, with no all-epoch verifier, so every lookup
    /// resolves to the epoch's signing scheme.
    #[derive(Clone, Default)]
    struct EpochProvider(Arc<Mutex<BTreeMap<Epoch, Arc<Scheme>>>>);

    impl EpochProvider {
        fn insert(&self, epoch: Epoch, scheme: Scheme) {
            self.0.lock().insert(epoch, Arc::new(scheme));
        }

        fn forget(&self, epoch: Epoch) {
            self.0.lock().remove(&epoch);
        }
    }

    impl Provider for EpochProvider {
        type Scope = Epoch;
        type Scheme = Scheme;

        fn scoped(&self, scope: Epoch) -> Option<Scoped<Scheme>> {
            self.0.lock().get(&scope).cloned().map(Scoped::scheme)
        }
    }

    /// Wraps the mock scheme so it can verify certificates without exposing participants.
    #[derive(Clone, Debug)]
    struct MaybeEnumerableScheme {
        inner: Scheme,
        enumerable: bool,
    }

    impl MaybeEnumerableScheme {
        const fn new(inner: Scheme, enumerable: bool) -> Self {
            Self { inner, enumerable }
        }

        fn wrap_attestation(attestation: Attestation<Scheme>) -> Attestation<Self> {
            Attestation {
                signer: attestation.signer,
                signature: attestation.signature,
            }
        }

        fn unwrap_attestation(attestation: Attestation<Self>) -> Attestation<Scheme> {
            Attestation {
                signer: attestation.signer,
                signature: attestation.signature,
            }
        }
    }

    impl Verifier for MaybeEnumerableScheme {
        type Subject<'a, D: commonware_cryptography::Digest> =
            commonware_consensus::simplex::types::Subject<'a, D>;
        type PublicKey = ed25519::PublicKey;
        type Certificate = <Scheme as Verifier>::Certificate;

        fn verify_certificate<R, D, M>(
            &self,
            rng: &mut R,
            subject: Self::Subject<'_, D>,
            certificate: &Self::Certificate,
            strategy: &impl ParallelStrategy,
        ) -> bool
        where
            R: rand_core::CryptoRng,
            D: commonware_cryptography::Digest,
            M: commonware_utils::Faults,
        {
            self.inner
                .verify_certificate::<_, D, M>(rng, subject, certificate, strategy)
        }

        fn verify_certificates<'a, R, D, I, M>(
            &self,
            rng: &mut R,
            certificates: I,
            strategy: &impl ParallelStrategy,
        ) -> bool
        where
            R: rand_core::CryptoRng,
            D: commonware_cryptography::Digest,
            I: Iterator<Item = (Self::Subject<'a, D>, &'a Self::Certificate)>,
            M: commonware_utils::Faults,
        {
            self.inner
                .verify_certificates::<_, D, _, M>(rng, certificates, strategy)
        }

        fn is_batchable() -> bool {
            Scheme::is_batchable()
        }

        fn certificate_codec_config(&self) -> <Self::Certificate as commonware_codec::Read>::Cfg {
            self.inner.certificate_codec_config()
        }

        fn certificate_codec_config_unbounded() -> <Self::Certificate as commonware_codec::Read>::Cfg
        {
            Scheme::certificate_codec_config_unbounded()
        }
    }

    impl certificate::Scheme for MaybeEnumerableScheme {
        type Signature = <Scheme as certificate::Scheme>::Signature;

        fn me(&self) -> Option<commonware_utils::Participant> {
            self.inner.me()
        }

        fn participants(&self) -> &commonware_utils::ordered::Set<Self::PublicKey> {
            assert!(
                self.enumerable,
                "verify-only scheme has no participant metadata"
            );
            self.inner.participants()
        }

        fn sign<D: commonware_cryptography::Digest>(
            &self,
            subject: Self::Subject<'_, D>,
        ) -> Option<Attestation<Self>> {
            self.inner.sign(subject).map(Self::wrap_attestation)
        }

        fn verify_attestation<R, D>(
            &self,
            rng: &mut R,
            subject: Self::Subject<'_, D>,
            attestation: &Attestation<Self>,
            strategy: &impl ParallelStrategy,
        ) -> bool
        where
            R: rand_core::CryptoRng,
            D: commonware_cryptography::Digest,
        {
            let attestation = Attestation::<Scheme> {
                signer: attestation.signer,
                signature: attestation.signature.clone(),
            };
            self.inner
                .verify_attestation(rng, subject, &attestation, strategy)
        }

        fn verify_attestations<R, D, I>(
            &self,
            rng: &mut R,
            subject: Self::Subject<'_, D>,
            attestations: I,
            strategy: &impl ParallelStrategy,
        ) -> Verification<Self>
        where
            R: rand_core::CryptoRng,
            D: commonware_cryptography::Digest,
            I: IntoIterator<Item = Attestation<Self>>,
            I::IntoIter: Send,
        {
            let verification = self.inner.verify_attestations(
                rng,
                subject,
                attestations.into_iter().map(Self::unwrap_attestation),
                strategy,
            );
            Verification::new(
                verification
                    .verified
                    .into_iter()
                    .map(Self::wrap_attestation)
                    .collect(),
                verification.invalid,
            )
        }

        fn assemble<I, M>(
            &self,
            attestations: I,
            strategy: &impl ParallelStrategy,
        ) -> Option<Self::Certificate>
        where
            I: IntoIterator<Item = Attestation<Self>>,
            I::IntoIter: Send,
            M: commonware_utils::Faults,
        {
            self.inner.assemble::<_, M>(
                attestations.into_iter().map(Self::unwrap_attestation),
                strategy,
            )
        }

        fn is_attributable() -> bool {
            Scheme::is_attributable()
        }
    }

    /// Serves a participant-less verifier from `scoped` and the full signing scheme from `scheme`.
    #[derive(Clone)]
    struct ParticipantlessAllProvider {
        verifier: Arc<MaybeEnumerableScheme>,
        scheme: Arc<MaybeEnumerableScheme>,
    }

    impl Provider for ParticipantlessAllProvider {
        type Scope = Epoch;
        type Scheme = MaybeEnumerableScheme;

        fn scoped(&self, _: Epoch) -> Option<Scoped<MaybeEnumerableScheme>> {
            Some(Scoped::verifier(self.verifier.clone()))
        }

        fn scheme(&self, _: Epoch) -> Option<Arc<MaybeEnumerableScheme>> {
            Some(self.scheme.clone())
        }
    }

    /// A single node in the harness: its probe ingress and its marshal mailbox.
    struct Node {
        probe: Mailbox<Scheme, Variant>,
        marshal: MarshalMailbox<Scheme, Variant>,
        // A clone of the node's probe channel sender, used by tests to inject raw
        // bytes that appear to originate from this node.
        probe_sender: Sender<ed25519::PublicKey, deterministic::Context>,
        // The probe actor, started on demand via `start_probes` once peers have
        // been seeded with finalizations. `None` once started.
        start: Option<Box<dyn FnOnce() -> Handle<()>>>,
        // Held to keep the spawned actors alive for the duration of the test.
        _handles: Vec<Handle<()>>,
    }

    /// A reusable harness of several real (unbuffered) marshal actors, each paired with a
    /// [`Probe`], wired over an all-to-all simulated p2p network.
    struct Harness {
        participants: Vec<ed25519::PublicKey>,
        schemes: Vec<Scheme>,
        nodes: Vec<Node>,
        oracle: Oracle<ed25519::PublicKey, deterministic::Context>,
        // Held to keep the network alive.
        _network: Handle<()>,
    }

    impl Harness {
        /// Spin up `n` nodes over a simulated network. Each node runs a real marshal actor
        /// (seeded with only a genesis block) and a [`Probe`] using `retry_timeout`,
        /// configured with a [`ConstantProvider`] (single scheme, all epochs).
        async fn setup(
            context: &deterministic::Context,
            n: u32,
            retry_timeout: NonZeroDuration,
        ) -> Self {
            Self::setup_with(context, n, retry_timeout, Epoch::zero(), |scheme| {
                ConstantProvider::new(scheme.clone())
            })
            .await
        }

        /// Like [`Harness::setup`], but `make_provider` builds each node's [`Probe`]
        /// certificate provider from that node's scheme. Lets tests supply an epoch-keyed
        /// provider to exercise multi-epoch verification.
        async fn setup_with<D, F>(
            context: &deterministic::Context,
            n: u32,
            retry_timeout: NonZeroDuration,
            minimum_epoch: Epoch,
            make_provider: F,
        ) -> Self
        where
            D: Provider<Scope = Epoch, Scheme = Scheme>,
            F: Fn(&Scheme) -> D,
        {
            let mut rng = test_rng();
            let Fixture {
                participants,
                schemes,
                ..
            } = scheme_mocks::fixture(&mut rng, NAMESPACE, n);

            // Simulated network with all participants tracked in a single peer set.
            let (network, oracle) = Network::new_with_peers(
                context.child("network"),
                SimConfig {
                    max_size: 1024 * 1024,
                    disconnect_on_block: true,
                    tracked_peer_sets: NZUsize!(1),
                },
                participants.clone(),
            )
            .await;
            let network = network.start();

            // All-to-all links so every node can reach every other node.
            for a in &participants {
                for b in &participants {
                    if a != b {
                        oracle
                            .add_link(a.clone(), b.clone(), LINK)
                            .await
                            .expect("failed to add link");
                    }
                }
            }

            let genesis = Block::new(0, 0);
            let mut nodes = Vec::with_capacity(n as usize);
            for (index, public_key) in participants.iter().enumerate() {
                let scheme = schemes[index].clone();
                let node_ctx = context.child("node").with_attribute("index", index);
                let partition_prefix = format!("node-{index}");
                let page_cache = CacheRef::from_pooler(&node_ctx, NZU16!(1024), NZUsize!(10));
                let control = oracle.control(public_key.clone());

                // Marshal backfill resolver.
                let backfill = control
                    .register(BACKFILL_CHANNEL, TEST_QUOTA)
                    .await
                    .expect("failed to register backfill channel");
                let resolver = marshal_resolver::init(
                    node_ctx.child("marshal_resolver"),
                    marshal_resolver::Config {
                        public_key: public_key.clone(),
                        peer_provider: oracle.manager(),
                        blocker: oracle.control(public_key.clone()),
                        mailbox_size: NZUsize!(100),
                        initial: Duration::from_secs(1),
                        timeout: Duration::from_secs(2),
                        fetch_retry_timeout: Duration::from_millis(100),
                        priority_requests: false,
                        priority_responses: false,
                    },
                    backfill,
                );

                // Marshal storage archives.
                let finalizations_by_height = immutable::Archive::init(
                    node_ctx.child("finalizations_by_height"),
                    archive_config(&partition_prefix, "finalizations", page_cache.clone()),
                )
                .await
                .expect("failed to init finalizations archive");
                let finalized_blocks = immutable::Archive::init(
                    node_ctx.child("finalized_blocks"),
                    archive_config(&partition_prefix, "blocks", page_cache.clone()),
                )
                .await
                .expect("failed to init blocks archive");

                // Marshal actor (unbuffered: blocks arrive via `proposed`/resolver, not a buffer).
                let marshal_config = marshal::Config {
                    provider: ConstantProvider::new(scheme.clone()),
                    epocher: FixedEpocher::new(EPOCH_LENGTH),
                    start: Start::Genesis(genesis.clone()),
                    partition_prefix: partition_prefix.clone(),
                    mailbox_size: NZUsize!(100),
                    view_retention_timeout: ViewDelta::new(10),
                    prunable_items_per_section: NZU64!(10),
                    page_cache,
                    replay_buffer: NZUsize!(2048),
                    key_write_buffer: NZUsize!(2048),
                    value_write_buffer: NZUsize!(2048),
                    block_codec_config: (),
                    max_repair: NZUsize!(10),
                    max_pending_acks: NZUsize!(1),
                    strategy: Sequential,
                };
                let (marshal_actor, marshal_mailbox, _) =
                    MarshalActor::<_, Variant, _, _, _, _, _>::init(
                        node_ctx.child("marshal"),
                        finalizations_by_height,
                        finalized_blocks,
                        marshal_config,
                    )
                    .await;
                let marshal_handle = marshal_actor.start_unbuffered(NoopReporter, resolver);

                // Probe.
                let probe_network = control
                    .register(PROBE_CHANNEL, TEST_QUOTA)
                    .await
                    .expect("failed to register probe channel");
                let probe_sender = probe_network.0.clone();
                let (probe, probe_mailbox) = Probe::new(Config {
                    context: node_ctx.child("probe"),
                    provider: make_provider(&scheme),
                    strategy: Sequential,
                    capacity: NZUsize!(100),
                    blocker: oracle.control(public_key.clone()),
                    minimum_epoch,
                    retry_timeout,
                });
                // Node 0 is the discoverer and is left in discovery for the test to drive. Every
                // other node is a source: attach its marshal so it enters service without
                // soliciting peers.
                if index != 0 {
                    probe_mailbox.attach(marshal_mailbox.clone());
                }
                // Defer the actor's start so tests can seed peer marshals before node 0 issues
                // its first request.
                let start: Box<dyn FnOnce() -> Handle<()>> =
                    Box::new(move || probe.start(probe_network));

                nodes.push(Node {
                    probe: probe_mailbox,
                    marshal: marshal_mailbox,
                    probe_sender,
                    start: Some(start),
                    _handles: vec![marshal_handle],
                });
            }

            Self {
                participants,
                schemes,
                nodes,
                oracle,
                _network: network,
            }
        }

        /// Starts every node's probe actor. Call after seeding peer marshals so each
        /// actor's first request observes the intended finalizations.
        fn start_probes(&mut self) {
            for node in &mut self.nodes {
                if let Some(start) = node.start.take() {
                    node._handles.push(start());
                }
            }
        }

        /// Builds a verifiable finalization at `height` committing to a block whose digest is
        /// `[digest_byte; 32]`, signed by this harness's scheme set. The returned block must be
        /// injected alongside it so the marshal can serve the finalization.
        fn finalization(
            &self,
            height: u64,
            digest_byte: u8,
        ) -> (Block, Finalization<Scheme, Sha256Digest>) {
            build_finalization(&self.schemes, height, digest_byte)
        }

        /// Injects a finalized block into node `index`'s marshal: caches the block, then reports
        /// the finalization so the marshal stores and serves it via `get_info`/`get_finalization`.
        async fn inject(
            &self,
            index: usize,
            block: Block,
            finalization: Finalization<Scheme, Sha256Digest>,
        ) {
            let mut marshal = self.nodes[index].marshal.clone();
            let round = finalization.proposal.round;
            assert!(marshal.verified(round, block).await);
            let _ = marshal.report(Activity::Finalization(finalization));
        }

        /// Sends raw `bytes` on the probe channel from node `from` to node `to`,
        /// bypassing the wire encoding. Used to deliver malformed messages to a [`Probe`].
        fn send_raw(&self, from: usize, to: usize, bytes: Vec<u8>) {
            let mut sender = self.nodes[from].probe_sender.clone();
            sender.send(Recipients::One(self.participants[to].clone()), bytes, false);
        }
    }

    /// Builds an epoch-0 finalization. See [`build_finalization_at`].
    fn build_finalization(
        schemes: &[Scheme],
        height: u64,
        digest_byte: u8,
    ) -> (Block, Finalization<Scheme, Sha256Digest>) {
        build_finalization_at(schemes, Epoch::zero(), height, digest_byte)
    }

    /// Builds a finalization in `epoch` at `height` over a block whose digest is
    /// `[digest_byte; 32]`, signed by `schemes`. Verifiable only against the matching verifier;
    /// signing with a foreign scheme set yields a structurally valid but unverifiable
    /// finalization.
    fn build_finalization_at<S>(
        schemes: &[S],
        epoch: Epoch,
        height: u64,
        digest_byte: u8,
    ) -> (Block, Finalization<S, Sha256Digest>)
    where
        S: commonware_consensus::simplex::scheme::Scheme<Sha256Digest>,
    {
        let block = Block::new(height, digest_byte);
        let round = Round::new(epoch, View::new(height));
        let proposal = Proposal {
            round,
            parent: View::new(height.saturating_sub(1)),
            payload: block.digest(),
        };
        let finalizes: Vec<_> = schemes
            .iter()
            .map(|scheme| Finalize::sign(scheme, proposal.clone()).expect("sign finalize"))
            .collect();
        let finalization =
            Finalization::from_finalizes(&schemes[0], &finalizes, &Sequential).expect("recover");
        (block, finalization)
    }

    /// Builds an [`EpochProvider`] seeded from `entries`, where each entry's scheme set comes
    /// from its own fixture so the scopes verify independently.
    fn epoch_provider(entries: impl IntoIterator<Item = (Epoch, Scheme)>) -> EpochProvider {
        let provider = EpochProvider::default();
        for (epoch, scheme) in entries {
            provider.insert(epoch, scheme);
        }
        provider
    }

    /// Encodes a finalization as the bytes of a [`wire::Message::Response`].
    fn finalization_bytes<S>(finalization: Finalization<S, Sha256Digest>) -> Vec<u8>
    where
        S: commonware_consensus::simplex::scheme::Scheme<Sha256Digest>,
    {
        wire::Message::<S, Variant>::Response(finalization)
            .encode()
            .to_vec()
    }

    #[cfg(feature = "arbitrary")]
    mod conformance {
        use super::{wire, Scheme, Variant};
        use commonware_codec::conformance::CodecConformance;

        commonware_conformance::conformance_tests! {
            CodecConformance<wire::Tag>,
            CodecConformance<wire::Message<Scheme, Variant>>,
        }
    }

    /// Storage configuration for one of marshal's immutable archives.
    fn archive_config(prefix: &str, name: &str, page_cache: CacheRef) -> immutable::Config<()> {
        immutable::Config {
            metadata_partition: format!("{prefix}-{name}-metadata"),
            freezer_table_partition: format!("{prefix}-{name}-freezer-table"),
            freezer_table_initial_size: 64,
            freezer_table_resize_frequency: 10,
            freezer_table_resize_chunk_size: 10,
            freezer_key_partition: format!("{prefix}-{name}-freezer-key"),
            freezer_key_page_cache: page_cache,
            freezer_value_partition: format!("{prefix}-{name}-freezer-value"),
            freezer_value_target_size: 1024,
            freezer_value_compression: None,
            ordinal_partition: format!("{prefix}-{name}-ordinal"),
            items_per_section: NZU64!(10),
            codec_config: (),
            replay_buffer: NZUsize!(2048),
            freezer_key_write_buffer: NZUsize!(2048),
            freezer_value_write_buffer: NZUsize!(2048),
            ordinal_write_buffer: NZUsize!(2048),
        }
    }

    /// A sample of peers with the same finalization lets a subscribing node resolve it as
    /// the floor.
    #[test]
    fn test_resolves_floor_from_sample_peers() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut harness =
                Harness::setup(&context, 4, NZDuration!(Duration::from_millis(500))).await;
            assert_eq!(harness.participants.len(), 4);

            // Seed a sample (f + 1 = 2) of peers with the same finalization, then start the
            // probes so node 0's first request already observes agreement.
            let (block, finalization) = harness.finalization(1, 1);
            for index in [1, 2] {
                harness
                    .inject(index, block.clone(), finalization.clone())
                    .await;
            }
            harness.start_probes();

            // The remaining node should resolve the agreed finalization as its floor.
            let floor = harness.nodes[0]
                .probe
                .subscribe()
                .await
                .expect("floor resolved");
            assert_eq!(floor, finalization);
        });
    }

    /// A sample of distinct valid replies resolves to the highest finalization, even without
    /// exact agreement.
    #[test]
    fn test_resolves_highest_floor_from_sample_replies() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut harness =
                Harness::setup(&context, 7, NZDuration!(Duration::from_secs(3600))).await;
            harness.start_probes();
            let mut subscription = harness.nodes[0].probe.subscribe();

            let mut expected = None;
            for index in 1..=3u8 {
                let (_, finalization) = harness.finalization(index.into(), index);
                if index == 3 {
                    expected = Some(finalization.clone());
                }
                harness.send_raw(index as usize, 0, finalization_bytes(finalization));
            }
            let expected = expected.expect("highest finalization present");

            context.sleep(Duration::from_millis(100)).await;
            let floor = subscription
                .try_recv()
                .expect("floor should resolve once enough replies arrive");
            assert_eq!(floor, expected);
        });
    }

    /// When too few peers reply, the node waits out its retry deadline before re-requesting.
    #[test_collect_traces]
    fn test_retries_until_enough_replies(traces: TraceStorage) {
        let retry_timeout = NZDuration!(Duration::from_millis(500));
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(move |context| async move {
            let mut harness = Harness::setup(&context, 4, retry_timeout).await;
            harness.start_probes();
            let start = context.current();
            let mut subscription = harness.nodes[0].probe.subscribe();

            // Node 0's first request gathers one reply, which is below the sample size (f + 1 = 2).
            let (_, finalization_f) = harness.finalization(1, 0xF);
            let (_, finalization_g) = harness.finalization(2, 0x6);
            harness.send_raw(1, 0, finalization_bytes(finalization_f.clone()));

            // Let node 0 finish its first request round. With too few finalizations available,
            // the subscription must remain pending.
            context.sleep(Duration::from_millis(100)).await;
            assert!(
                matches!(
                    subscription.try_recv(),
                    Err(oneshot::error::TryRecvError::Empty)
                ),
                "floor resolved before enough replies arrived"
            );

            // After the retry deadline, send a full sample. The actor should select the
            // highest valid reply.
            context.sleep(retry_timeout.get()).await;
            harness.send_raw(1, 0, finalization_bytes(finalization_f));
            harness.send_raw(2, 0, finalization_bytes(finalization_g.clone()));

            // The floor resolves to G, and only after the retry deadline has elapsed.
            context.sleep(Duration::from_millis(100)).await;
            let floor = subscription.try_recv().expect("floor resolved");
            assert_eq!(floor, finalization_g);
            let elapsed = context.current().duration_since(start).unwrap();
            assert!(
                elapsed >= retry_timeout.get(),
                "floor resolved before a retry could occur ({elapsed:?})"
            );
        });

        // The retry was driven by the deadline.
        let events = traces.get_all();
        events
            .expect_event(|event| {
                event.metadata.content == "re-requesting finalizations"
                    && event
                        .metadata
                        .expect_field_exact("reason", "deadline elapsed")
                        .is_ok()
            })
            .expect("a deadline-driven retry should have occurred");
    }

    /// Agreement below the sample size is not enough to resolve a floor.
    #[test]
    fn test_waits_for_sample_size_even_with_matching_replies() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut harness =
                Harness::setup(&context, 7, NZDuration!(Duration::from_secs(3600))).await;
            harness.start_probes();
            let mut subscription = harness.nodes[0].probe.subscribe();

            // Seven participants => f + 1 = 3. Two matching replies are still not enough.
            let (_, finalization) = harness.finalization(1, 1);
            for index in 1..=2 {
                harness.send_raw(index, 0, finalization_bytes(finalization.clone()));
            }

            context.sleep(Duration::from_millis(100)).await;
            assert!(
                matches!(
                    subscription.try_recv(),
                    Err(oneshot::error::TryRecvError::Empty)
                ),
                "matching replies below the sample size must not resolve the floor"
            );
        });
    }

    /// Starting without a floor subscriber does not solicit peers. This lets a source node start
    /// the actor before attaching marshal without sending a useless initial request.
    #[test]
    fn test_does_not_request_without_subscriber() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut harness =
                Harness::setup(&context, 4, NZDuration!(Duration::from_millis(500))).await;
            harness.start_probes();

            context.sleep(Duration::from_millis(100)).await;

            let metrics = context.encode();
            assert!(
                !metrics.contains("network_messages_sent_total"),
                "unexpected network messages before subscription: {metrics}"
            );
        });
    }

    /// A peer that sends a finalization-tagged payload that cannot be decoded is blocked.
    #[test]
    fn test_blocks_peer_sending_malformed_finalization() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut harness =
                Harness::setup(&context, 4, NZDuration!(Duration::from_millis(500))).await;
            harness.start_probes();

            let mut junk = vec![1u8];
            junk.extend_from_slice(&[0xAB; 32]);
            harness.send_raw(1, 0, junk);

            context.sleep(Duration::from_millis(100)).await;

            let blocked = harness.oracle.blocked().await.unwrap();
            assert!(
                blocked.contains(&(
                    harness.participants[0].clone(),
                    harness.participants[1].clone(),
                )),
                "node 0 should have blocked node 1"
            );
        });
    }

    /// A peer that sends a correctly encoded but unverifiable finalization (signed by a
    /// foreign key set) is blocked.
    #[test]
    fn test_blocks_peer_sending_invalid_finalization() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut harness =
                Harness::setup(&context, 4, NZDuration!(Duration::from_millis(500))).await;
            harness.start_probes();

            // A finalization signed by a foreign key set: it decodes cleanly under node 0's
            // codec config but fails verification against node 0's scheme.
            let mut rng = test_rng();
            let Fixture {
                schemes: foreign, ..
            } = scheme_mocks::fixture(&mut rng, b"_COMMONWARE_GLUE_PROBE_FOREIGN", 4);
            let (_, finalization) = build_finalization(&foreign, 1, 1);
            harness.send_raw(1, 0, finalization_bytes(finalization));

            context.sleep(Duration::from_millis(100)).await;

            let blocked = harness.oracle.blocked().await.unwrap();
            assert!(
                blocked.contains(&(
                    harness.participants[0].clone(),
                    harness.participants[1].clone(),
                )),
                "node 0 should have blocked node 1"
            );
        });
    }

    #[test]
    fn test_blocks_peer_sending_invalid_message() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut harness =
                Harness::setup(&context, 4, NZDuration!(Duration::from_millis(500))).await;
            harness.start_probes();

            // An unrecognized wire tag is rejected by the decoder.
            harness.send_raw(1, 0, vec![0xFF]);

            context.sleep(Duration::from_millis(100)).await;

            let blocked = harness.oracle.blocked().await.unwrap();
            assert!(
                blocked.contains(&(
                    harness.participants[0].clone(),
                    harness.participants[1].clone(),
                )),
                "node 0 should have blocked node 1"
            );
        });
    }

    /// A subscriber that arrives after the floor is already resolved receives it immediately.
    #[test]
    fn test_late_subscriber_receives_resolved_floor() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut harness =
                Harness::setup(&context, 4, NZDuration!(Duration::from_millis(500))).await;
            let (block, finalization) = harness.finalization(1, 1);
            for index in [1, 2, 3] {
                harness
                    .inject(index, block.clone(), finalization.clone())
                    .await;
            }
            harness.start_probes();

            // Resolve the floor via a first subscriber.
            let floor = harness.nodes[0]
                .probe
                .subscribe()
                .await
                .expect("floor resolved");
            assert_eq!(floor, finalization);

            // A subscriber arriving afterwards is served the cached floor immediately.
            let late = harness.nodes[0]
                .probe
                .subscribe()
                .await
                .expect("late subscriber served");
            assert_eq!(late, finalization);
        });
    }

    /// Once a floor is set, further finalizations are ignored without verification: a peer
    /// that sends an otherwise-blockable (invalid) finalization is not blocked.
    #[test]
    fn test_ignores_finalizations_after_floor_set() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut harness =
                Harness::setup(&context, 4, NZDuration!(Duration::from_millis(500))).await;
            let (block, finalization) = harness.finalization(1, 1);
            for index in [1, 2, 3] {
                harness
                    .inject(index, block.clone(), finalization.clone())
                    .await;
            }
            harness.start_probes();

            let floor = harness.nodes[0]
                .probe
                .subscribe()
                .await
                .expect("floor resolved");
            assert_eq!(floor, finalization);

            // Node 3 sends an invalid (foreign) finalization. Were the floor unset, this would
            // fail verification and block node 3; with the floor set it is ignored entirely.
            let mut rng = test_rng();
            let Fixture {
                schemes: foreign, ..
            } = scheme_mocks::fixture(&mut rng, b"_COMMONWARE_GLUE_PROBE_FOREIGN", 4);
            let (_, invalid) = build_finalization(&foreign, 2, 9);
            harness.send_raw(3, 0, finalization_bytes(invalid));

            context.sleep(Duration::from_millis(100)).await;

            let blocked = harness.oracle.blocked().await.unwrap();
            assert!(
                !blocked.contains(&(
                    harness.participants[0].clone(),
                    harness.participants[3].clone(),
                )),
                "finalizations after the floor is set must be ignored, not verified"
            );
        });
    }

    /// A second finalization from a peer already counted this round is ignored and does not let a
    /// single peer inflate the sample.
    #[test]
    fn test_duplicate_finalization_from_peer_is_ignored() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut harness =
                Harness::setup(&context, 7, NZDuration!(Duration::from_secs(3600))).await;
            harness.start_probes();
            let mut subscription = harness.nodes[0].probe.subscribe();

            // Deliver two different valid finalizations from node 1, then the second from node 2.
            // If node 1's second answer were counted, the second finalization would incorrectly
            // reach the sample size.
            let (_, first) = harness.finalization(1, 1);
            let (_, second) = harness.finalization(2, 2);
            harness.send_raw(1, 0, finalization_bytes(first));
            harness.send_raw(1, 0, finalization_bytes(second.clone()));
            harness.send_raw(2, 0, finalization_bytes(second));

            context.sleep(Duration::from_millis(100)).await;

            // The duplicate is ignored, and one peer cannot inflate the sample.
            let blocked = harness.oracle.blocked().await.unwrap();
            assert!(
                !blocked.contains(&(
                    harness.participants[0].clone(),
                    harness.participants[1].clone(),
                )),
                "a duplicate finalization must be ignored, not treated as a fault"
            );
            assert!(
                matches!(
                    subscription.try_recv(),
                    Err(oneshot::error::TryRecvError::Empty)
                ),
                "a duplicate finalization must not satisfy the sample size"
            );
        });
    }

    /// Duplicates are skipped before validation: a peer that follows a valid finalization with an
    /// invalid one is not blocked because it cannot affect the sample.
    #[test]
    fn test_invalid_duplicate_finalization_is_ignored() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut harness =
                Harness::setup(&context, 4, NZDuration!(Duration::from_secs(3600))).await;
            harness.start_probes();
            let _subscription = harness.nodes[0].probe.subscribe();

            // Node 1 first sends a valid finalization, then an invalid (foreign) one.
            let (_, valid) = harness.finalization(1, 1);
            harness.send_raw(1, 0, finalization_bytes(valid));

            let mut rng = test_rng();
            let Fixture {
                schemes: foreign, ..
            } = scheme_mocks::fixture(&mut rng, b"_COMMONWARE_GLUE_PROBE_FOREIGN", 4);
            let (_, invalid) = build_finalization(&foreign, 1, 2);
            harness.send_raw(1, 0, finalization_bytes(invalid));

            context.sleep(Duration::from_millis(100)).await;

            let blocked = harness.oracle.blocked().await.unwrap();
            assert!(
                !blocked.contains(&(
                    harness.participants[0].clone(),
                    harness.participants[1].clone(),
                )),
                "an invalid duplicate finalization must be ignored, not blocked"
            );
        });
    }

    /// A peer sample resolves to its highest finalization, not to the most common stale one.
    #[test]
    fn test_sample_selects_highest_over_stale_agreement() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            // Seven participants => f + 1 = 3.
            let mut harness =
                Harness::setup(&context, 7, NZDuration!(Duration::from_secs(3600))).await;
            harness.start_probes();
            let mut subscription = harness.nodes[0].probe.subscribe();

            // Two peers agree on a stale finalization, but the sample also includes a newer
            // valid finalization. The highest reply should win.
            let (_, stale) = harness.finalization(1, 0x0F);
            let (_, newest) = harness.finalization(2, 0xA2);
            harness.send_raw(1, 0, finalization_bytes(stale.clone()));
            harness.send_raw(2, 0, finalization_bytes(stale.clone()));
            harness.send_raw(3, 0, finalization_bytes(newest.clone()));

            context.sleep(Duration::from_millis(100)).await;
            let floor = subscription.try_recv().expect("floor resolved");
            assert_eq!(floor, newest);
        });
    }

    /// Finalizations are verified against the scheme for their own epoch: a non-zero-epoch
    /// finalization, signed by that epoch's committee and reported by enough peers, resolves
    /// the floor (exercises the epoch-scoped scheme lookup and per-epoch sample size).
    #[test]
    fn test_resolves_floor_at_non_zero_epoch() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            // A committee for epoch 1 (f + 1 = 2), distinct from the harness's epoch-0 set.
            let mut rng = test_rng();
            let Fixture {
                schemes: epoch_one, ..
            } = scheme_mocks::fixture(&mut rng, b"_COMMONWARE_GLUE_FD_EPOCH_ONE", 4);
            let provider = epoch_provider([(Epoch::new(1), epoch_one[0].clone())]);

            let mut harness = Harness::setup_with(
                &context,
                4,
                NZDuration!(Duration::from_secs(3600)),
                Epoch::zero(),
                {
                    let provider = provider.clone();
                    move |_scheme| provider.clone()
                },
            )
            .await;
            harness.start_probes();
            let mut subscription = harness.nodes[0].probe.subscribe();

            // Two peers report the same epoch-1 finalization, signed by the epoch-1 committee.
            let (_, finalization) = build_finalization_at(&epoch_one, Epoch::new(1), 1, 7);
            harness.send_raw(1, 0, finalization_bytes(finalization.clone()));
            harness.send_raw(2, 0, finalization_bytes(finalization.clone()));

            context.sleep(Duration::from_millis(100)).await;
            let floor = subscription.try_recv().expect("floor resolved");
            assert_eq!(floor, finalization);
        });
    }

    /// Finalizations below the configured minimum epoch are ignored without blocking and do not
    /// prevent the same peers from contributing accepted finalizations later in the round.
    #[test]
    fn test_ignores_finalization_below_minimum_epoch() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let provider = EpochProvider::default();
            let mut harness = Harness::setup_with(
                &context,
                4,
                NZDuration!(Duration::from_secs(3600)),
                Epoch::new(1),
                {
                    let provider = provider.clone();
                    move |_scheme| provider.clone()
                },
            )
            .await;
            provider.insert(Epoch::zero(), harness.schemes[0].clone());
            provider.insert(Epoch::new(1), harness.schemes[0].clone());
            harness.start_probes();
            let mut subscription = harness.nodes[0].probe.subscribe();

            // These finalizations are valid and decodable, but below the configured search floor.
            let (_, old) = build_finalization_at(&harness.schemes, Epoch::zero(), 1, 1);
            harness.send_raw(1, 0, finalization_bytes(old.clone()));
            harness.send_raw(2, 0, finalization_bytes(old));
            context.sleep(Duration::from_millis(50)).await;
            assert!(
                matches!(
                    subscription.try_recv(),
                    Err(oneshot::error::TryRecvError::Empty)
                ),
                "below-minimum finalizations must not resolve the floor"
            );

            // The same peers can still contribute once they answer with an accepted epoch.
            let (_, accepted) = build_finalization_at(&harness.schemes, Epoch::new(1), 2, 2);
            harness.send_raw(1, 0, finalization_bytes(accepted.clone()));
            harness.send_raw(2, 0, finalization_bytes(accepted.clone()));
            context.sleep(Duration::from_millis(100)).await;

            let floor = subscription.try_recv().expect("floor resolved");
            assert_eq!(floor, accepted);
            let blocked = harness.oracle.blocked().await.unwrap();
            assert!(blocked.is_empty(), "no peer should be blocked");
        });
    }

    /// An all-epoch verifier may be unable to enumerate participants. Probe must verify with it
    /// but size peer samples from the epoch-scoped committee.
    #[test]
    fn test_sample_size_uses_scoped_scheme_with_participantless_all_verifier() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let mut rng = test_rng();
            let Fixture {
                participants,
                schemes,
                verifier,
                ..
            } = scheme_mocks::fixture(&mut rng, b"_COMMONWARE_GLUE_FD_ALL_VERIFIER", 4);
            let schemes: Vec<_> = schemes
                .into_iter()
                .map(|scheme| MaybeEnumerableScheme::new(scheme, true))
                .collect();
            let provider = ParticipantlessAllProvider {
                verifier: Arc::new(MaybeEnumerableScheme::new(verifier.clone(), false)),
                scheme: Arc::new(MaybeEnumerableScheme::new(verifier, true)),
            };

            let (network, oracle) = Network::new_with_peers(
                context.child("network"),
                SimConfig {
                    max_size: 1024 * 1024,
                    disconnect_on_block: true,
                    tracked_peer_sets: NZUsize!(1),
                },
                participants.clone(),
            )
            .await;
            let _network = network.start();
            for a in &participants {
                for b in &participants {
                    if a != b {
                        oracle
                            .add_link(a.clone(), b.clone(), LINK)
                            .await
                            .expect("failed to add link");
                    }
                }
            }

            let probe_network = oracle
                .control(participants[0].clone())
                .register(PROBE_CHANNEL, TEST_QUOTA)
                .await
                .expect("failed to register probe channel");
            let (probe, probe_mailbox) = Probe::<
                _,
                MaybeEnumerableScheme,
                ParticipantlessAllProvider,
                Variant,
                _,
                ed25519::PublicKey,
                _,
            >::new(Config {
                context: context.child("probe"),
                provider,
                strategy: Sequential,
                capacity: NZUsize!(100),
                blocker: oracle.control(participants[0].clone()),
                minimum_epoch: Epoch::zero(),
                retry_timeout: NZDuration!(Duration::from_secs(3600)),
            });
            let _probe = probe.start(probe_network);
            let mut subscription = probe_mailbox.subscribe();

            let mut peer_channels = Vec::new();
            for public_key in participants.iter().take(3).skip(1) {
                peer_channels.push(
                    oracle
                        .control(public_key.clone())
                        .register(PROBE_CHANNEL, TEST_QUOTA)
                        .await
                        .expect("failed to register peer probe channel"),
                );
            }
            let (_, finalization) = build_finalization_at(&schemes, Epoch::zero(), 1, 1);
            for (sender, _) in &peer_channels {
                let mut sender = sender.clone();
                sender.send(
                    Recipients::One(participants[0].clone()),
                    finalization_bytes(finalization.clone()),
                    false,
                );
            }

            context.sleep(Duration::from_millis(100)).await;
            let floor = subscription.try_recv().expect("floor resolved");
            assert_eq!(floor, finalization);
        });
    }

    /// A finalization for an epoch the provider has no scheme for cannot be judged, so it is
    /// dropped without blocking the sender (and never reaches the floor).
    #[test]
    fn test_ignores_unknown_epoch_finalization() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            // The provider knows only epoch 1.
            let mut rng = test_rng();
            let Fixture {
                schemes: epoch_one, ..
            } = scheme_mocks::fixture(&mut rng, b"_COMMONWARE_GLUE_FD_EPOCH_ONE", 4);
            let provider = epoch_provider([(Epoch::new(1), epoch_one[0].clone())]);

            let mut harness = Harness::setup_with(
                &context,
                4,
                NZDuration!(Duration::from_secs(3600)),
                Epoch::zero(),
                {
                    let provider = provider.clone();
                    move |_scheme| provider.clone()
                },
            )
            .await;
            harness.start_probes();

            // A finalization for the unknown epoch 5: dropped before verification, so the sender
            // is not blocked.
            let (_, unknown) = build_finalization_at(&epoch_one, Epoch::new(5), 1, 1);
            harness.send_raw(1, 0, finalization_bytes(unknown));

            context.sleep(Duration::from_millis(100)).await;

            let blocked = harness.oracle.blocked().await.unwrap();
            assert!(
                !blocked.contains(&(
                    harness.participants[0].clone(),
                    harness.participants[1].clone(),
                )),
                "an unknown-epoch finalization must be ignored, not blocked"
            );
            let mut subscription = harness.nodes[0].probe.subscribe();
            context.sleep(Duration::from_millis(50)).await;
            assert!(
                matches!(
                    subscription.try_recv(),
                    Err(oneshot::error::TryRecvError::Empty)
                ),
                "an unknown-epoch finalization must not resolve the floor"
            );
        });
    }

    /// A finalization buffered while its epoch was known stays harmless if that epoch is later
    /// forgotten: subsequent selection passes treat it as unjudgeable (still reachable) rather
    /// than counting it, and the sender is not retroactively blocked.
    #[test]
    fn test_forgotten_epoch_finalization_is_not_counted() {
        let runner = deterministic::Runner::timed(Duration::from_secs(30));
        runner.start(|context| async move {
            let provider = EpochProvider::default();

            let mut harness = Harness::setup_with(
                &context,
                4,
                NZDuration!(Duration::from_secs(3600)),
                Epoch::zero(),
                {
                    let provider = provider.clone();
                    move |_scheme| provider.clone()
                },
            )
            .await;
            provider.insert(Epoch::new(1), harness.schemes[0].clone());
            provider.insert(Epoch::new(2), harness.schemes[0].clone());
            harness.start_probes();
            let mut subscription = harness.nodes[0].probe.subscribe();

            // Peer 1 reports an epoch-1 finalization; it verifies and is buffered below the
            // sample size.
            let (_, epoch_one_finalization) =
                build_finalization_at(&harness.schemes, Epoch::new(1), 1, 1);
            harness.send_raw(1, 0, finalization_bytes(epoch_one_finalization));
            context.sleep(Duration::from_millis(50)).await;

            // Forget epoch 1, so the buffered finalization's scheme is now unavailable.
            provider.forget(Epoch::new(1));

            // One peer reports an epoch-2 finalization; ingesting it re-runs selection over
            // the buffer, where the stale epoch-1 entry can no longer be judged or counted.
            let (_, epoch_two_finalization) =
                build_finalization_at(&harness.schemes, Epoch::new(2), 1, 2);
            harness.send_raw(2, 0, finalization_bytes(epoch_two_finalization));
            context.sleep(Duration::from_millis(50)).await;

            // No floor (one currently judgeable vote, below the sample size) and nothing is
            // blocked.
            assert!(
                matches!(
                    subscription.try_recv(),
                    Err(oneshot::error::TryRecvError::Empty)
                ),
                "a single judgeable vote must not resolve the floor"
            );
            let blocked = harness.oracle.blocked().await.unwrap();
            assert!(blocked.is_empty(), "no peer should be blocked");
        });
    }
}