Validate FFI primitive inputs + refresh bindings/tests

[chavic]

This PR addresses #1262 and hardens the FFI boundary by validating primitive inputs and surfacing explicit errors. Updates integration tests/bindings to match the new error shapes and removes silent acceptance of invalid fee rates/amounts.

Follow-ups

Decide final script/witness size caps and document rationale.
Add Dart integration fix for large fee rate parsing (separate PR).

[coveralls]

Pull Request Test Coverage Report for Build 21952519211

Details

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 83.263%

---

Totals
Change from base Build 21922394027:	0.0%
Covered Lines:	10238
Relevant Lines:	12296

---

💛 - Coveralls

[benalleng]

Instead of updating the msrv for dart I think we can simply use cargo install --locked wasm-bindgen-cli on the js workflow to prevent the install step from trying to update the deps

[chavic]

Instead of updating the msrv for dart I think we can simply use cargo install --locked wasm-bindgen-cli on the js workflow to prevent the install step from trying to update the

Thanks... applying that now

[benalleng]

CACK f331ef3 got the commits nicely reviewable again thanks for that.

[benalleng]

ACK f331ef3 I have just a few nits and a broader question but overall I think it looks good! Thanks again for getting these commits easier to read 👍

[benalleng]

A question for a followup perhaps but should we consider making a primitives module to keep this kind of file in case we end up needing more like it?

[chavic]

@benalleng Not bad actually, can move the shared validation helpers into a primitives module, only question is should this be in this PR or is it better to make that a separate PR

[benalleng]

this should be separate, and only probably once we have a lot more primitives to work with

[benalleng]

Jumped the gun, I think I would like this changed moved and explained into a new commit. I totally didn't realise it was a readme change..

[benalleng]

we sneaked 12d2d33 in #1301 so we can safely remove 050bd88 from this PR

[benalleng]

I think you may have misunderstood my comments on the review, the psbt consts can be simply imported and not set as strings in the tests. I appreciate seeing the test_utils.dart but it was not really my intention to need to add this.

[benalleng]

you've kind of reversed what you did here that you did in the dart file

[chavic]

This fell back during the rebase; I had a backup branch, so I'll bring it back. Side note, I've noticed only the Python test fails if the poll approach isn't used... will try to find out why (not the top task)

[chavic]

sidenote, should we consider asserting a failure when attempts exceed 3, unless the intent is explicitly “smoke test only”

[benalleng]

The can actually just be

final psbt = test_utils.originalPsbt();

you will need to add it to the test_utils.dart import though

[spacebear21]

Related note (but not a blocker for this PR), we should really not expose test utils in the main payjoin namespace for language bindings. This test_utils approach is a nice helper but the test utils are still exposed in the main namespace. I tried my hand at this briefly and couldn't find a good approach #1199 (comment)