PDK Async Dev Updates - Dec 1 2025

[spacebear21]

Contributors please post what you have been busy doing, what you doing this week and what are your constraints. Please post them as comments on this discussion by EOD.

[chavic]

Last Week

• Uniffi-Dart – landed PR #102 to teach the Dart bindings real async. The runtime now tracks Rust and Dart futures, generators emit async factory methods for constructors/free fns, and callback codegen can await Dart implementations. Cleaned up the dart_async fixture so it uses the new factories and tightened the timing assertions, closing out issue #66.

• Rust Payjoin FFI – followed up on rust-payjoin Uniformly use primitive types in FFI boundary, with runtime validation #738 by removing bitcoin_ffi and making the FFI boundary primitive-only. Address/PSBT/script/tx data are all plain strings/bytes with runtime validation before rewrapping into payjoin types, keeping validation intact while dropping the bitcoin_ffi dependency.

This week

• Address any issues with the PRs
• Define and add a dedicated primitive-conversion error category in the send/receive public error enums; wire all malformed base64/hex/txid/vout/script cases to it at the FFI boundary.
• Regenerate bindings (JS/Dart/Python) and update tests in each language to assert the new error on bad primitives.
  • update relevant documentation

[spacebear21]

Last week:

Focused primarily on language bindings, with the goal of ultimately publishing 1.0-rc packages to the pub.dev and npm libraries respectively for dart and JS.

• Rename payjoin bindings to just "payjoin" (removing _ffi and _dart suffixes)
• Reviewing and testing compatibility of @kumulynja's Shift to Native Assets + update to uniffi 0.30.0 Uniffi-Dart/uniffi-dart#99 PR which ugprades uniffi-dart to 0.30 and moves to "Dart native assets". This would allow us to ship the dart package with Rust code that compiles natively, instead of compiling and publishing binaries for every target architecture we want to support.
• To make this work, I also needed to fork uniffi-bindgen-react-native to support 0.30 and upgrade bitcoin-ffi as well.
• I discovered that unfortunately dart doesn't allow making blocking calls of async functions (like we do in payjoin-cli with tokio::task::block_in_place), so we will likely need some kind of async interface for the state machine callback functions and persister methods. We must figure this out to unblock Bull Bitcoin Mobile. I'm exploring doing this at the FFI layer only by exposing async duplicates of those functions behind a feature-flag, which use block_in_place to "convert" the provided async function to a sync one for the native payjoin library. This is contingent on @chavic's Add async constructor & callback support for Dart bindings Uniffi-Dart/uniffi-dart#102.

This week:

• Continue Switch dart bindings to use native assets #1208 for uniffi 0.30 + Dart native assets support
• Lots of PRs to review - at least these big ones:
  • Shift to Native Assets + update to uniffi 0.30.0 Uniffi-Dart/uniffi-dart#99
  • Add async constructor & callback support for Dart bindings Uniffi-Dart/uniffi-dart#102
  • feat: add non-persisted Async Payjoin support bitcoindevkit/bdk-cli#200
  • Move the Default Payjoin FFI API to primitives #1191

[nothingmuch]

last 3 weeks: mostly spent reviewing literature on several topics relevant to simulations and txn construction

as an example, the thing i spent the most time on was BFT consensus variations and relaxed models for agreement/broadcast, trying to reconcile some details imposed by the domain (for example the requirment for unanimity of signatures means safety is for ~free but liveness is much harder, so can we exploit this to improve liveness of the consensus layer), and especially how to make the semi-honest threat model evolve into the byzantine threat model, i.e. if we started with just trusted (for liveness) counterparties in multiparty payjoin, if done in a particular way can that improve the liveness guarantees we're able to later make in the byzantine setting (this is known as asymmetric byzantine fault tolerant consensus, where different parties can make different trust assumptions)...

a few more examples: some recent divisible e-cash work might allow us to bypass reissuance requests, different (reliable) broadcast primitives, possible transport layers for both settings (iroh or LN for semi honest, tor and i2p for bft, ...)

[benalleng]

Last week:

• Established better communications with some of the liana devs in their repo to hopefully get some more clear feedback in implementation.
• Moved our flake over to nixos-stable 25.05 to protect ourselves from unstable packages.

This week:

• Continue pushing liana integration forward, hopefully moving on to adding some PJ Receiver support alongside the msrv upgrade
• With nixos 25.11 released I plan to update our flake as mentioned in Use nixos stable for nix flake #1194 (review)
• Look at pulling out some of the various consts spread across the codebase used for tests and unify them into the test-utils module

[arminsabouri]

Last week was a short week. Mostly focused on code reviews, attempting to undust some old PRs and get them reviewed or assign a reviewer. Also focused on defining the cost function and strategies for unilateral spenders and payjoiners in the simulation framework. Strategies in this context define the possible actions an agent can take. And Cost functions define how they "score" them. With the aim of an agent being to reduce their cost given a list of actions they can take i.e their wallet state. Right now I am dealing with some bugs that I introduced pre-thanksgiving. By mid week I hope to define two strategies agents can adopt. To unilaterally spend once the deadline is approaching or optimistically do or respond to payjoin attempts. The batching anxiety should also increase as deadlines appraoch for a payment intent but should also be a function of how much fee saving you are gaining. And in the future a "privacy preserved" metric. After that the focus will be on multi party constructions. Lastly, today on a research call nothingmuch and I spent some time discussing BFT consensus variations (update above summarizes what we talked about https://github.com/orgs/payjoin/discussions/1209#discussioncomment-15127465).

[Mshehu5]

Last week:

• Pushed Derive P2TR input weight from witness data #1200 and addressed some reviews.
• Continued working on the BDK CLI persister, making changes to align with the current PR updates. While doing that, I found a minor issue and left a review: feat: add non-persisted Async Payjoin support bitcoindevkit/bdk-cli#200 (review).

This week :

• Continue working on Derive P2TR input weight from witness data #1200 and remove it from draft once I receive feedback from spacebear regarding the comments on Derive P2TR input weight from witness data #1200 (review).
• I’m also looking to pick up tracking: opt-out of v1 requests on the directory #1198 (comment) if no other contributors are currently working on it.
• Continue working on the BDK CLI persister and plan to push code to Efe’s PR branch once reviews on the his current PR are completed

[DanGould]

Limited activity for me this past week. Thanks everyone for keeping ahead in my relative absence.

• Finished up Post to test new RSS feed blog push
• Met with Brandon about RSS feed from blog. We're almost there!
• Got foundation banking, remote corporate operation in a happy place
• @zealsham and I caught up, discussed ABC conference attendance strategy