Add a public bundle-builder function as an alternative to the mutable builder. #404
Conversation
The term `recipient` is commonly used to refer to the address to which a note is sent; however, a bundle may include multiple outputs to the same recipient. This change is intended to clarify this usage. Co-authored-by: Daira Emma Hopwood <[email protected]>
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## main #404 +/- ##
==========================================
- Coverage 81.07% 80.99% -0.08%
==========================================
Files 31 31
Lines 3112 3141 +29
==========================================
+ Hits 2523 2544 +21
- Misses 589 597 +8 ☔ View full report in Codecov by Sentry. |
nuttycom
force-pushed
the
builder_functions
branch
from
c6d49ed to
2e2c161
Compare
| /// Defined in [Zcash Protocol Spec § 4.8.3: Dummy Notes (Orchard)][orcharddummynotes]. | ||
| /// | ||
| /// [orcharddummynotes]: https://zips.z.cash/protocol/nu5.pdf#orcharddummynotes | ||
| fn dummy(rng: &mut impl RngCore) -> Self { | ||
| pub fn dummy(rng: &mut impl RngCore) -> Self { |
There was a problem hiding this comment.
Why does this need to be public? SpendInfo::dummy is not public (so there's an API asymmetry now), and the builder is responsible for applying padding internally.
| pub fn dummy(rng: &mut impl RngCore) -> Self { | |
| fn dummy(rng: &mut impl RngCore) -> Self { |
There was a problem hiding this comment.
I want it to be possible for callers to add dummy actions beyond the two required padding; they can already do so by manually sending zero value to a random address, so exposing this convenience method doesn't add any capability that they don't already have, and I don't see there as being a substantive maintenance burden.
nuttycom
force-pushed
the
builder_functions
branch
from
c482781 to
422c5e3
Compare
| let num_actions = bundle_type | ||
| .num_actions(num_requested_spends, num_requested_outputs) | ||
| .map_err(|_| BuildError::BundleTypeNotSatisfiable)?; |
There was a problem hiding this comment.
This is now a bit unintuitive, because it can potentially include both real spends and outputs, and dummies beyond what the bundle type's padding requires. This opens up the API to user-defined extra padding that can potentially make their transactions distinguishable from network transactions.
That being said, the way that BundleType::num_actions works means that this doesn't prevent the BundleType's padding rule from being applied, so this is a hazard that should be documented, not a bug per se.
There was a problem hiding this comment.
The users can already add padding by just sending zeros to arbitrary addresses, so it doesn't change anything.
nuttycom
force-pushed
the
builder_functions
branch
from
e85e415 to
74ba740
Compare
Co-authored-by: str4d <[email protected]>
nuttycom
force-pushed
the
builder_functions
branch
from
74ba740 to
938b203
Compare
An alternative/extension to #403