feat!: CLAUDE-GENERATED use class ID to deduplicate bytecode hashing and class ID derivation, use address for address derivation by dbanks12 · Pull Request #15683 · AztecProtocol/aztec-packages

dbanks12 · 2025-07-13T18:38:33Z

Changes bytecode ID to just be the same as class ID, unless bytecode retrieval fails in which case execution just uses bytecode ID of 0.

Adds deduplication to address derivation, class ID derivation and bytecode hashing.

dbanks12 · 2025-07-13T18:38:50Z

This stack of pull requests is managed by Graphite. Learn more about stacking.

fcarreiro

BTW, this is why I was asking if deduplication was done in your PR, I thought you would: https://docs.google.com/document/d/1CNHZrtBukR4CMRxF2vKnYiC77C-6J6ChOrqO9Gu_xr8/edit?disco=AAABnM22Gmw

fcarreiro · 2025-07-14T14:59:31Z

barretenberg/cpp/src/barretenberg/vm2/simulation/class_id_derivation.cpp

 void ClassIdDerivation::assert_derivation(const ContractClassId& class_id, const ContractClass& klass)
 {
-    // TODO: Cache and deduplicate.
+    if (derived_class_ids.contains(class_id)) {


When do you add the id?

I think here you could just make the event emitter a deduplicated event emitter and avoid all this. In non-assert mode, they will be equivalent.

Yikes yeah I didn't actually insert anywhere

fcarreiro · 2025-07-14T15:01:08Z

barretenberg/cpp/src/barretenberg/vm2/simulation/bytecode_hashing.cpp

 FF BytecodeHasher::compute_public_bytecode_commitment([[maybe_unused]] const BytecodeId bytecode_id,
                                                      const std::vector<uint8_t>& bytecode)
 {
+    if (bytecode_commitments.contains(bytecode_id)) {


Same, when do you add to the map?

fcarreiro · 2025-07-14T15:02:19Z

barretenberg/cpp/src/barretenberg/vm2/simulation/address_derivation.cpp


 void AddressDerivation::assert_derivation(const AztecAddress& address, const ContractInstance& instance)
 {
+    if (derived_addresses.contains(address)) {


This one has a more expensive computation, so you could chose to add a set (but you still need to insert at some point!). However, it might be simpler to just do it with a deduplicating event emitter.

fcarreiro · 2025-07-14T15:06:55Z

barretenberg/cpp/src/barretenberg/vm2/simulation/bytecode_manager.cpp

    }

    ContractClassId current_class_id = maybe_instance.value().current_class_id;
+    BytecodeId bytecode_id = BytecodeId(current_class_id);


I'd have to sit down quite some time to convince myself that using bytecode id works... in PIL. A few concerns
(1) in any given table that has bytecodeid how would you constrain (if needed) that there are no duplicates. Observe that even if two identical rows are not bad, you could claim 2 different commitments for the same bytecode id and that would be bad. With bytecode ids (and not class ids) you can ask for monotonically increasing numbers. Can we/should we do it for class id and how?
(2) It feels bytecode ids and class ids are being mixed. If you think that class id can be used, then maybe just keep/rename things to class id where it is what it represents.

I'd like to be sure that you think things will work before I spend a few hours myself looking in detail at all the connections. LMK!

I hadn't really thought 1 all the way through. I guess this was handled for bc_retrieval.pil just by enforcing that bytecode_id is doing ++ on each row.... Which I didn't realize and didn't unconstrain in this PR 😅

Yes, bytecode and class IDs are being mixed. Or well.... they're == except when it doesn't exist. I agree that maybe it makes sense to just use "class ID".

Observe that even if two identical rows are not bad, you could claim 2 different commitments for the same bytecode id and that would be bad.

I'm not convinced that claiming 2 different commitments for a bytecode ID would actually be bad if the bytecode ID == class ID. Because you'd have a constraint failure when one of your 2 different commitments doesn't match the commitment used when deriving the class ID. So I don't think we actually need to constrain that bytecode ID is deduplicated.

Although as discussed last week, we do need to constrain that bytecode ID is right

I'll write up a document with my thoughts and send it your way.

…ivation

dbanks12 · 2025-08-05T17:21:10Z

Replaced with #15977

dbanks12 mentioned this pull request Jul 13, 2025

feat!: constrain the GetContractInstance opcode in AVM #15430

Merged

dbanks12 mentioned this pull request Jul 13, 2025

feat!: constrain NullifierExists AVM opcode #15636

Merged

dbanks12 marked this pull request as ready for review July 13, 2025 18:40

dbanks12 requested review from IlyasRidhuan, Maddiaa0, fcarreiro and jeanmon as code owners July 13, 2025 18:40

dbanks12 removed request for IlyasRidhuan, Maddiaa0 and jeanmon July 13, 2025 18:41

dbanks12 force-pushed the db/gci branch from 985c7d2 to 60fdad4 Compare July 14, 2025 14:33

dbanks12 force-pushed the db/dedup branch 2 times, most recently from 0bc88e8 to b641799 Compare July 14, 2025 14:44

dbanks12 force-pushed the db/gci branch from 60fdad4 to dda8606 Compare July 14, 2025 14:44

dbanks12 force-pushed the db/dedup branch from b641799 to 8c1c41b Compare July 14, 2025 14:52

dbanks12 force-pushed the db/gci branch from dda8606 to 79a88b5 Compare July 14, 2025 14:52

fcarreiro reviewed Jul 14, 2025

View reviewed changes

dbanks12 force-pushed the db/dedup branch from 8c1c41b to 66e33c5 Compare July 14, 2025 18:35

dbanks12 force-pushed the db/gci branch from 79a88b5 to 2e4e394 Compare July 14, 2025 18:35

dbanks12 force-pushed the db/dedup branch from 66e33c5 to ba3be58 Compare July 14, 2025 18:38

dbanks12 force-pushed the db/gci branch 2 times, most recently from 03a4646 to 96f94f2 Compare July 15, 2025 15:02

dbanks12 force-pushed the db/dedup branch from ba3be58 to 583d039 Compare July 15, 2025 15:02

dbanks12 force-pushed the db/gci branch from 96f94f2 to dcf75af Compare July 15, 2025 17:39

dbanks12 force-pushed the db/dedup branch from 583d039 to 70e0f03 Compare July 15, 2025 17:39

dbanks12 marked this pull request as draft July 15, 2025 17:39

dbanks12 force-pushed the db/dedup branch from 70e0f03 to 6957632 Compare July 15, 2025 18:05

dbanks12 force-pushed the db/gci branch 2 times, most recently from 0ce85f7 to d6eb37a Compare July 15, 2025 20:09

dbanks12 force-pushed the db/emitnullifier branch from 8427cb6 to c920de5 Compare July 22, 2025 15:21

dbanks12 force-pushed the db/dedup branch from 0bc55d7 to 8383589 Compare July 22, 2025 15:21

Base automatically changed from db/emitnullifier to merge-train/avm July 22, 2025 15:22

dbanks12 force-pushed the db/dedup branch 2 times, most recently from 2f9bbfa to 8689ea0 Compare July 22, 2025 21:12

Base automatically changed from merge-train/avm to next July 23, 2025 09:07

AztecBot added 13 commits July 24, 2025 00:56

[empty] Start merge-train. Choo choo.

fb8ac86

Merge branch 'next' into merge-train/avm

469029b

Merge branch 'next' into merge-train/avm

5a71f33

Merge branch 'next' into merge-train/avm

18642e3

Merge branch 'next' into merge-train/avm

4c66e79

Merge branch 'next' into merge-train/avm

8d9f3d1

Merge branch 'next' into merge-train/avm

8014550

Merge branch 'next' into merge-train/avm

1c4bdc6

Merge branch 'next' into merge-train/avm

55618da

Merge branch 'next' into merge-train/avm

dfc188d

Merge branch 'next' into merge-train/avm

fb2fc89

Merge branch 'next' into merge-train/avm

e54c999

Merge branch 'next' into merge-train/avm

50e553c

dbanks12 changed the base branch from next to graphite-base/15683 July 24, 2025 15:14

dbanks12 force-pushed the db/dedup branch from 8689ea0 to 3b21cce Compare July 24, 2025 15:14

dbanks12 force-pushed the graphite-base/15683 branch from f52457d to 5f4f579 Compare July 24, 2025 15:14

dbanks12 changed the base branch from graphite-base/15683 to merge-train/avm July 24, 2025 15:14

dbanks12 changed the title ~~feat!: deduplicate bytecode hashing, class ID derivation, address derivation~~ feat!: CLAUDE-GENERATED deduplicate bytecode hashing, class ID derivation, address derivation Jul 24, 2025

feat!: deduplicate bytecode hashing, class ID derivation, address der…

fe65f04

…ivation

dbanks12 force-pushed the db/dedup branch from 3b21cce to fe65f04 Compare July 24, 2025 15:17

dbanks12 changed the title ~~feat!: CLAUDE-GENERATED deduplicate bytecode hashing, class ID derivation, address derivation~~ feat!: CLAUDE-GENERATED use class ID to deduplicate bytecode hashing and class ID derivation, use address for address derivation Jul 25, 2025

Base automatically changed from merge-train/avm to next July 25, 2025 18:50

dbanks12 closed this Aug 5, 2025

Conversation

dbanks12 commented Jul 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

dbanks12 commented Jul 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

fcarreiro left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dbanks12 Jul 22, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dbanks12 commented Aug 5, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

dbanks12 commented Jul 13, 2025 •

edited

Loading

dbanks12 commented Jul 13, 2025 •

edited

Loading

dbanks12 Jul 22, 2025 •

edited

Loading