Add hooks to enable TX debugging by djolertrk · Pull Request #2574 · 0xMiden/protocol

djolertrk · 2026-03-10T08:24:22Z

This PR adds a small hook to the transaction execution path so that the program runner used by midentx::TransactionExecutor can be swapped out while preserving the existing transaction preparation/execution pipeline.

The motivation is debugger support and better developer tooling. We want to be able to debug programs executed under the transaction kernel, instead of only debugging standalone VM programs. In practice, this lets downstream tooling swap in a custom executor that:

starts a DAP server
enables tracing/debugging
or wraps execution with richer diagnostics

while still reusing the existing TransactionExecutor logic for transaction preparation, host setup, note handling, foreign account loading, etc.

The default behavior is unchanged: TransactionExecutor still uses FastProcessor by default.

djolertrk · 2026-03-10T08:25:04Z

cc @bitwalker

PhilippGackstatter · 2026-03-10T13:35:18Z

Could you add a description or link an issue where we can find information about what the goal of the PR is? It is hard to review the changes or judge alternatives without this information.

bitwalker · 2026-03-10T13:36:37Z

@PhilippGackstatter @bobbinth This PR makes it possible for us to debug programs executed by the transaction kernel, which is a big improvement in DX. I'd like to try and get this released as part of the v0.22 release cycle/upcoming testnet - just let me know if you need anything on our end to make that go smoothly.

bitwalker · 2026-03-10T13:37:07Z

@djolertrk Looks like a couple lint checks failed

bobbinth

Thank you! Not a review yet - but a couple of high-level comments:

It seems like this PR includes a lot of extraneous changes (e.g., from the AggLayer crate). Could we remove them? (these changes will land soon as a part of #2546)
It would be great to provide a bit more detail about the goal here. After a brief look, my understanding is that we are trying to control how FastProcessor is instantiated, but I don't yet have full understanding of how this will be used in the compiler. I'm mostly trying to understand if the TransactionProgramFactory pattern is really needed here or if we can do something simpler (e.g., maybe adding a generic type to the TransactionExecutor would suffice).

Also, I wonder if there may be some overlap with #2293 (though, maybe not).

crates/miden-protocol/src/lib.rs

djolertrk · 2026-03-11T06:50:14Z

Thank you! Not a review yet - but a couple of high-level comments:

It seems like this PR includes a lot of extraneous changes (e.g., from the AggLayer crate). Could we remove them? (these changes will land soon as a part of feat: migrate miden-agglayer to VM 0.21 and crypto 0.22 #2546)

It would be great to provide a bit more detail about the goal here. After a brief look, my understanding is that we are trying to control how FastProcessor is instantiated, but I don't yet have full understanding of how this will be used in the compiler. I'm mostly trying to understand if the TransactionProgramFactory pattern is really needed here or if we can do something simpler (e.g., maybe adding a generic type to the TransactionExecutor would suffice).

Also, I wonder if there may be some overlap with #2293 (though, maybe not).

@bobbinth Thanks for the comments! Yes, it does overlap with the #2546 (the AggLayer changes for example), but I needed those so I can test the feature end to end. But yes, if the PR will land soon, I can just rebase on top of it -- so I am doing it.

The immediate consumer here is the debugger, not the compiler. The goal is to let miden-tx::TransactionExecutor reuse its existing transaction preparation / host setup logic, but swap out the program runner used underneath it. The default remains FastProcessor; the debugger plugs in an alternative executor that enables tracing / DAP-driven execution instead of always running straight to completion.

@PhilippGackstatter I've updated PR description as well.

djolertrk · 2026-03-11T06:50:54Z

note: Rebased on top of #2546.

crates/miden-protocol/src/transaction/executor.rs

crates/miden-tx/src/executor/mod.rs

djolertrk · 2026-03-11T14:28:17Z

note: now rebased on top of next since the #2546 is merged.

bitwalker · 2026-03-13T17:27:44Z

@djolertrk I think we've landed on our plan for how to proceed with this. If you can rebase on the latest changes in next, and make the switch to a single trait vs factory + executor traits, I believe this should be basically ready to merge.

djolertrk · 2026-03-13T19:31:31Z

@djolertrk I think we've landed on our plan for how to proceed with this. If you can rebase on the latest changes in next, and make the switch to a single trait vs factory + executor traits, I believe this should be basically ready to merge.

Yes. I agree. Let's do it that way (I was already on it -- doing some final testing).
Thanks a lot!

bobbinth

Looks good! Thank you. I left some comments inline.

crates/miden-tx/src/executor/program_executor.rs

crates/miden-protocol/src/transaction/executor.rs

crates/miden-tx/src/lib.rs

crates/miden-tx/src/executor/mod.rs

crates/miden-tx/src/executor/notes_checker.rs

djolertrk · 2026-03-14T17:56:40Z

@bobbinth Thank you for the comments!

bobbinth

Looks good! Thank you!

crates/miden-tx/src/executor/program_executor.rs

bobbinth · 2026-03-14T18:45:43Z

crates/miden-tx/src/executor/notes_checker.rs

                .map_err(TransactionCheckerError::TransactionPreparation)?;

-        let processor = FastProcessor::new(stack_inputs).with_advice(advice_inputs);
+        let processor = EXEC::new(stack_inputs, advice_inputs, ExecutionOptions::default());


Not related to this PR, but I wonder if the options should be configured somehow (i.e., we shouldn't be using just the default options). cc @PhilippGackstatter.

bobbinth · 2026-03-14T18:48:21Z

crates/miden-tx/src/executor/mod.rs

        let (mut host, stack_inputs, advice_inputs) = self.prepare_transaction(&tx_inputs).await?;

-        let processor = FastProcessor::new(stack_inputs).with_advice(advice_inputs);
+        let processor = EXEC::new(stack_inputs, advice_inputs, ExecutionOptions::default());


Same comment as above.

bitwalker requested review from PhilippGackstatter and bobbinth March 10, 2026 13:33

bobbinth reviewed Mar 10, 2026

View reviewed changes

crates/miden-protocol/src/lib.rs Outdated Show resolved Hide resolved

djolertrk force-pushed the pr/enable-tx-debugging-clean branch from 917ffe3 to 6a5bc19 Compare March 11, 2026 06:50

bobbinth reviewed Mar 11, 2026

View reviewed changes

crates/miden-protocol/src/transaction/executor.rs Outdated Show resolved Hide resolved

bobbinth reviewed Mar 11, 2026

View reviewed changes

crates/miden-tx/src/executor/mod.rs Show resolved Hide resolved

djolertrk force-pushed the pr/enable-tx-debugging-clean branch from 6a5bc19 to 27c1fea Compare March 11, 2026 14:27

djolertrk mentioned this pull request Mar 11, 2026

Enable TX debugging 0xMiden/miden-client#1883

Draft

djolertrk added 2 commits March 13, 2026 21:31

feat: add transaction executor factory hook

ffaff59

fixup: simplify transaction program executor hook

d1e99cf

djolertrk force-pushed the pr/enable-tx-debugging-clean branch from 27c1fea to d1e99cf Compare March 13, 2026 21:04

bobbinth reviewed Mar 14, 2026

View reviewed changes

fixup: address comments

0ccc18d

bobbinth approved these changes Mar 14, 2026

View reviewed changes

fixup: add TODO and changelog

4ee94c5

Conversation

djolertrk commented Mar 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

djolertrk commented Mar 10, 2026

Uh oh!

PhilippGackstatter commented Mar 10, 2026

Uh oh!

bitwalker commented Mar 10, 2026

Uh oh!

bitwalker commented Mar 10, 2026

Uh oh!

bobbinth left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

djolertrk commented Mar 11, 2026

Uh oh!

djolertrk commented Mar 11, 2026

Uh oh!

Uh oh!

Uh oh!

djolertrk commented Mar 11, 2026

Uh oh!

bitwalker commented Mar 13, 2026

Uh oh!

djolertrk commented Mar 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

bobbinth left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

djolertrk commented Mar 14, 2026

Uh oh!

bobbinth left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

bobbinth Mar 14, 2026

Choose a reason for hiding this comment

Uh oh!

bobbinth Mar 14, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

djolertrk commented Mar 10, 2026 •

edited

Loading

djolertrk commented Mar 13, 2026 •

edited

Loading