We have old Rust 1.29 error handling code still in `hashes`. Implement
`std::error::Error` for the `hex::Error` and `error::Error` types in
line with "modern" Rust 1.41.1 error handling.
ed6f6d11dd Implement fmt traits for ScriptBuf (Tobin C. Harding)
Pull request description:
We can improve ergonomics of the `script` module by implementing the `fmt` traits on `ScriptBuf`, trivial because we can call through to the `Script` implementations.
Fix: #1585
ACKs for top commit:
Kixunil:
ACK ed6f6d11dd
apoelstra:
ACK ed6f6d11dd
Tree-SHA512: 878a1522af4ed1e10d1d8d60d150e6571008c008b5e5c662c67462f9e09075b4f1fe4e399ed50e98cd7253b6815937c6732cd1ce02b74a5be017d5b8fcdbbd2f
7dde3b3b22 Make max/min_value functions const (Tobin C. Harding)
Pull request description:
The `max_value` and `min_value` functions only exist to be compatible/uniform with Rust 1.41.1 they will never change and they just return a constant value. They can therefore be made const functions.
ACKs for top commit:
apoelstra:
ACK 7dde3b3b22
Kixunil:
ACK 7dde3b3b22
Tree-SHA512: a998dcc68d409ee01188f1e1e157182b93fae17ad41c6b7440b69dcea367c6f65d7ecbdcdbbcdd6a0a8349e248ff13e06175ca1505df1e01be1935143bbffb3c
We can improve ergonomics of the `script` module by implementing the
`fmt` traits on `ScriptBuf`, trivial because we can call through to the
`Script` implementations.
The `max_value` and `min_value` functions only exist to be
compatible/uniform with Rust 1.41.1 they will never change and they just
return a constant value. They can therefore be made const functions.
The `rustfmt` config option `required_version` causes grief in CI
because everytime `rustfmt` is update our `nightly` job fails. Since we
require nightly anyways to run `rustfmt` the config option is basically
redundant.
Recently we merged `commit 53d4fe66b57c255086def2b5e47afaddee776b75` to
fix CI even though a better approach is to use `'_` because it assists
reading the code (shows that the bit stream writer is not writing from a
reference since its writing a `Copy` type `n`).
Add back in the `'_` (I forget what its called).
877f9af364 Add new hex parse error variant (Tobin C. Harding)
Pull request description:
Recently we used an error type that holds only one expected hex string length when parsing but for `PublicKey`s we have two (66 and 130). Add a new error variant to express the error. Requires adding a variant to `bip32` for the same thing.
Fix: #1281
ACKs for top commit:
Kixunil:
ACK 877f9af364
apoelstra:
ACK 877f9af364
Tree-SHA512: c1ca493ee30418bd82bc326b35c18731260e4217c371f37301a73c64f9a6631163801acc217c6c2c7b14f632a2ad5043174266c1b4fdce127698e68ab8494f20
3c0598b399 Add standard constants to lock times (Tobin C. Harding)
Pull request description:
Some of the lock time structs (`Height`, `Time` ect.) are missing standard constants for min, max ect.
Add standard constants taking into consideration the various locktime corner cases.
Add `max_value` and `min_value` to be consistent with Rust 1.41.1 (incl. `Sequence`).
Fix: #1451
This PR is not complex in itself but **locktimes are notoriously complex, please wait for 3 acks before merging** - and ack'ing makes no guarantee that reviewer got all corner cases :)
There is no rush on this one, apoelstra, Kixunil, sanket1729 please just review when your brain is fresh.
ACKs for top commit:
apoelstra:
ACK 3c0598b399
Kixunil:
ACK 3c0598b399
Tree-SHA512: aa3d112db83b4785edb0a7a517cc335ded59f5967eb39b8979a6d68f9bba4644a27e5ca400fcabf368a1f8c0eecdef0b87b1011933ac7fd96b467b8501533203
53d4fe66b5 Remove unnecessary lifetime (Tobin C. Harding)
Pull request description:
Clippy emits:
warning: the following explicit lifetimes could be elided: 'a
As suggested, remove the unnecessary lifetime.
ACKs for top commit:
apoelstra:
ACK 53d4fe66b5
Kixunil:
ACK 53d4fe66b5
Tree-SHA512: cc881164f319c6876cea9284aad5c613d3928864fe41e22f8c9c23a65aa6334e4c9fa6037f4ca7bed117e7915a389f152de2411f0b470975dd7e871e48de79c7
Recently we used an error type that holds only one expected hex string
length when parsing but for `PublicKey`s we have two (66 and 130). Add a
new error variant to express the error. Requires adding a variant to
`bip32` for the same thing.
Fix: #1281
Implement `AsRef<[u8; X]>` for hash types including wrapped hash types.
Doing so means at times the compiler can no longer infer the type because we have
`AsRef<[u8]` implemented also but we can use `into_inner` and `as_inner`
to get the inner array if needed.
Add an implementation of `From<Address> for ScriptBuf` that calls
through to `address.script_pubkey` (which calls
`address.payload.script_pubkey()`).
Fix: #1457
e079524b2a Remove hashes::hex::HexWriter (Tobin C. Harding)
Pull request description:
The `HexWriter` is not used any more since we added the new hex code in internals for fast hex encoding.
While we are removing the benches for `HexWriter` also remove the last remaining bench for writing using `Display` because this is not the correct place for that code - its trivial to re add later in the correct module.
ACKs for top commit:
Kixunil:
ACK e079524b2a
apoelstra:
ACK e079524b2a
Tree-SHA512: 3896ad55e9e6f0e30d5494f008d942d7cb2ab264c97205d184abdb1f560704d721514ca9878af6012d1c10a2b8753e86542af7137726c1620b8c363ff49b0a61
Some of the lock time structs (`Height`, `Time` ect.) are missing
standard constants for min, max ect.
Add standard constants taking into consideration the various locktime
corner cases.
Add `max_value` and `min_value` to be consistent with Rust 1.41.1 (incl.
`Sequence`).
Fix: #1451
a762a89b48 Add documentation to Sequence::is_final (Tobin C. Harding)
b1490a26ea Move enables_absolute_lock_time method (Tobin C. Harding)
Pull request description:
The term "final" is an archaic Bitcoin term however it is well used, it exists in Bitcoin Core code as well as in various bips. To help folks new to Bitcoin add documentation to the `is_final` method including historical notes.
Note, this does _not_ deprecate `is_final` - while writing the notes I found the term "final" in enough official places that I think its fair game to keep the term, some things people just have to learn, we can definitely help with that learning though.
Fix: #1198
ACKs for top commit:
Kixunil:
ACK a762a89b48
apoelstra:
ACK a762a89b48
Tree-SHA512: 895fbdce90223d90c0a68fb1e3d6b7aada4a3606d1294ea4df1f4194681a79d970b0434e7bb078f6d5cbf413b3550e72560d6d5cf811a5a959adf53f7f778ab2
8c0e5213d3 Delegate debug for ScriptBuf to Script (Tobin C. Harding)
Pull request description:
Currently the derived implementation of `Debug` for `ScriptBuf` prints the inner vector of u8s as integers, this is ugly and hard to read. The `Script` implementation of `Debug` prints the script opcodes and data as hex, we can just delegate to it.
With this applied we get debug output of form:
Script(OP_DUP OP_HASH160 OP_PUSHBYTES_20 3bde42dbee7e4dbe6a21b2d50ce2f0167faa8159 OP_EQUALVERIFY OP_CHECKSIG)
Fix: #1516
ACKs for top commit:
Kixunil:
ACK 8c0e5213d3
apoelstra:
ACK 8c0e5213d3
Tree-SHA512: ca07d9fb191f4e0379cbd96b2944e6881094a8334d39b97209b6bf452a3c15d4aede53b9c88176b9b7667b7a539d47897940bc561dc9f8cd83ce1990a08047e1
1d3d5a9c5b Take Into<secp256k1::PublicKey> in PublicKey constructors (Tobin C. Harding)
b13a76407b keys: Clean up test imports (Tobin C. Harding)
Pull request description:
We can make the API more ergonomic by taking a generic argument that implements `Into<secp256k1::PublicKey>` in the `bitcoin::PublicKey` constructors.
The only thing than this is useful for is passing in `KeyPair` and the `From` implementation already exists. Add a unit test to verify.
Fix: #1453
## Note
As per the discussion in #1453 I checked secp and bitcoin for all keys that can be converted using `From` and it turns out its only `KeyPair` which already has `From` impls - good rust-bitcoin devs :)
ACKs for top commit:
Kixunil:
ACK 1d3d5a9c5b
apoelstra:
ACK 1d3d5a9c5b
Tree-SHA512: b5e5272561de15cdcfb15913aa5d42ddc96bf2fd5835068a5a9aa0274074ffa698ec9e81707f102b7d1b244f1abd0fdbd0eb4b6b505c84c3d5719dcb01d46efb
49e8b8da32 Use write_all for sighash encoding (Tobin C. Harding)
Pull request description:
From BIP143:
> If sighash type is SINGLE and the input index is smaller than the number of outputs, hashOutputs is the double SHA256 of the output amount with scriptPubKey of the same index as the input;
Currently we are using a `Sighash` which wraps double sha256 so while technically correct this means we are relying on `Sighash` to implement `Encodable`. We can remove this requirement by directly using the `sha256d::Hash` type to hash the outputs data.
Fix: #1549
ACKs for top commit:
Kixunil:
ACK 49e8b8da32
apoelstra:
ACK 49e8b8da32
Tree-SHA512: 8dd0037245a7cf180ba8a6eceeadad912d4adc14fc3f49df9008856de262624666d7d575195eea4868b2a5252dc565590e6be78471053b5e6367f3d2363310e8
e7bbfd3913 Improve Psbt error handling (DanGould)
Pull request description:
## Separate `encode::Error` and `psbt::Error` recursive dependency
This initial work attempts to fix#837's first 2 points
> - The current psbt::serialize::Deserialize has an error type of consensus::encode::Error. I think we should cleanly separate consensus encoding errors from application-level encoding errors like psbt.
> - There is a recursive dependence between encode::Error and psbt::Error which would need to be cleanly dissected and separated so that there is no dependence or only one-way dependence.
## Better `ParseError(String)` types
arturomf94 how compatible do your #1310 changes look to address #837's third point with this design?
> - There are a lot ParseError(String) messages that could use a better type to downflow the information.
I think your prior art would completely address this issue now.
## On handling `io::Error` with an associated error
`encode::Error` has an `Io` variant. now that `Psbt::deserialize` returns `psbt::Error` and produces an `io::Error`, we need an `Io` variant on `psbt::Error`. Except that doing so breaks `#[derive(Eq)]` and lots of tests for `psbt::Error`.
Kixunil, I'm trying to understand your feedback regarding a solution to this problem.
> I believe that the best error untangling would be to make decodable error associated.
> I meant having associated `Error` type at `Decodable` trait. Encoding should only fail if the writer fails so we should have `io::Error` there (at least until we have something like `genio`).
>
> > [it] is a problem to instantiate consensus::encode::Error in [the psbt] module for `io::Error`?
>
> It certainly does look strange. Maybe we should have this shared type:
>
> ```rust
> /// Error used when reading or decoding fails.
> pub enum ReadError<Io, Decode> {
> /// Reading failed
> Io(Io),
> /// Decoding failed
> Decode(Decode), // consensus and PSBT error here
> }
> ```
>
> However this one will be annoying to use with `?` :( We could have `ResultExt` to provide `decode()` and `io()` methods to make it easier.
>
> If that's not acceptable then I think deduplicated IO error is better.
Kixunil didn't we just get rid of Psbt as `Decodable`? Would this make more sense to have as an error associated with `Deserialize`? Or did we do the opposite of what we should have by making Psbt only `Serialize`/`Deserialize` because of #934, where only consensus objects are allowed to be `Decodable`? I wonder if we prioritized that strict categorization and are stuck with worth machinery because of it. My goal with #988 was to get to a point where we could address #837 and ultimately implement PSBTv2.
ACKs for top commit:
tcharding:
ACK e7bbfd3913
apoelstra:
ACK e7bbfd3913
Tree-SHA512: 32975594fde42727ea9030f46570a1403ae1a108570ab115519ebeddc28938f141e2134b04d6b29ce94817ed776c13815dea5647c463e4a13b47ba55f4e7858a
44d3ec487d Rename Payload::as_bytes to inner_prog_as_bytes (sanket1729)
a446df583c Make Payload non-exhaustive (sanket1729)
6ebc9de252 Introduce WitnessProgram struct and cleanup Address validity invariants (sanket1729)
41652caf05 Introduce is_spend_standard method (sanket1729)
Pull request description:
Fixes#1561.
Highlights:
- Segwitv0 programs with lengths apart from 20 or 32 are invalid `Address` struct. Such Addresses are useless and we should not parse/create them.
- Renamed `is_standard` to `is_spend_standard`.
ACKs for top commit:
apoelstra:
ACK 44d3ec487d
tcharding:
ACK 44d3ec487d
Tree-SHA512: 1ee36f7ea25c65619ddf7d643d025690096876843dbe6fbdf877ce23e88049d79b0bbd78cee6cf4b415bca028b3634bb70c9f52d1098bd90558e6ba7f8731332
Addresses with Segwitv0 not having len 20/32 are invalid and cannot be
constructed. Also cleans up a API bug in
ScriptBuf::new_witness_prog(ver, prog) allowing prog of invalid lenghts.
ebfbe74243 Implement `Debug` for generic `Address<V: NetworkValidation>` (Jiri Jakes)
Pull request description:
Previously `Debug` was implemented for both `Address<NetworkChecked>` and `Address<NetworkUnchecked>`, but not for cases when the `NetworkValidation` parameter was generic. This change adds this ability. Based on Kixunil's tip.
With previous implementation, the `test_address_debug()` resulted in error:
The added `Debug` on `NetworkChecked` and `NetworkUnchecked` are required by compiler.
---
While dealing with derives and impls, I also attempted to turn all the derives on `Address` into manual impls (see Kixunil's suggestion in https://github.com/rust-bitcoin/rust-bitcoin/pull/1489#discussion_r1052448057). The motivation behind this was the possibility to remove derives on `NetworkChecked` and `NetworkUnchecked`, too. However, even with manual impls, all the traits on `NetworkChecked` and `NetworkUnchecked` were still required by compiler in this sort of situations (see also the rest of the same discussion linked above). I do not fully understand why, perhaps limitation of this way of sealing traits?
It can be demonstrated by removing `Debug` derivation on `NetworkUnchecked` and `NetworkChecked` in this PR and running `test_address_debug()`.
Therefore, if we want to allow users of the library to define types generic in `NetworkValidation` and at the same time derive impls, it seems to me that `NetworkChecked` and `NetworkUnchecked` will have to have the same set of impls as `Address` itself.
ACKs for top commit:
Kixunil:
ACK ebfbe74243
tcharding:
ACK ebfbe74243
apoelstra:
ACK ebfbe74243
Tree-SHA512: 87f3fa4539602f31bf4513a29543b04e943c3899d8ece36d0d905c3b5a2d76e29eb86242694b5c494faa5e54bb8f69f5048849916c6438ddd35030368f710353
6e56feed57 bip158: Replace usage of HashSet with BTreeSet (Tobin C. Harding)
Pull request description:
The `bip158` module uses a `HashSet` and in order to do so requires the `hashbrown` dependency for "no-std" builds.
We can replace the usage of `HashSet` with a `BTreeSet` in `bip158` and remove the `hashbrown` dependency entirely.
This patch makes no claims about performance cost or benefit of this change. The patch also makes no claims about the validity of the current `HashSet` usage.
The `hashbrown` dependency and `HashSet` usage can be trivially added back in if someone comes up with perf data to back it up.
ACKs for top commit:
Kixunil:
ACK 6e56feed57
apoelstra:
ACK 6e56feed57
Tree-SHA512: 6e8d6af7ccf22031a22ea19b731fbe5f6cbfdfb1e510119a7dbdd2e8521eeb3b0d2c7b9e000641deae8960287099dbf58515341720cf7c41d1fcab08308b0b74
We can make the API more ergonomic by taking a generic argument that
implements `Into<secp256k1::PublicKey>` in the `bitcoin::PublicKey`
constructors.
The only thing than this is useful for is passing in `KeyPair` and the
`From` implementation already exists. Add a unit test to verify.
Fix: #1453
Currently we have an attempted tag ```compile_fail that seems to be
aiming at allowing code that does not build to exist in rustdoc. This is
causing an error when running tests.
No clue how this made it through CI.
Use ```ignore to prevent rustdoc code from being built.
Currently the derived implementation of `Debug` for `ScriptBuf` prints
the inner vector of u8s as integers, this is ugly and hard to read. The
`Script` implementation of `Debug` prints the script opcodes and data as
hex, we can just delegate to it.
With this applied we get debug output of form:
Script(OP_DUP OP_HASH160 OP_PUSHBYTES_20 \
3bde42dbee7e4dbe6a21b2d50ce2f0167faa8159 OP_EQUALVERIFY OP_CHECKSIG)
Fix: #1516
When we introduced the SPDX license blurb in [0] we incorrectly gave
attribution to Andrew when the original file author had the attribution
as "the rust-bitcoin developers". The original author [1] was Tamas
Blummer and he copied this code from code he wrote and explicitly
re-licenses it. In order to make the re-licensing comment a little
clearer and fix the mis-attribution use Tamas' name in the attribution.
[0] commit: `91ff2f628ce7db732d234a812e29fa8508f501a1 Introduce SPDX license identifiers`
[1] commit: `c93a70487f81a93c7d479ae046c75590d9fb7733 Add client side block filter (BIP158) (#281)`
The term "final" is an archaic Bitcoin term however it is well used, it
exists in Bitcoin Core code as well as in various bips. To help folks
new to Bitcoin add documentation to the `is_final` method including
historical notes.
The `bip158` module uses a `HashSet` and in order to do so requires the
`hashbrown` dependency for "no-std" builds.
We can replace the usage of `HashSet` with a `BTreeSet` in `bip158` and
remove the `hashbrown` dependency entirely.
This patch makes no claims about performance cost or benefit of this
change. The patch also makes no claims about the validity of the current
`HashSet` usage.
The `hashbrown` dependency and `HashSet` usage can be trivially added
back in if someone comes up with perf data to back it up.
Previously `Debug` was implemented for both `Address<NetworkChecked>`
and `Address<NetworkUnchecked>`, but not for cases when the
`NetworkValidation` parameter was generic. This change adds this
ability.
In preparation for deprecating the `is_final` method; move the
`enables_absolute_lock_time` method to be directly above the `is_final`
method.
Refactor only, no logic changes.
The `HexWriter` is not used any more since we added the new hex code in
internals for fast hex encoding.
While we are removing the benches for `HexWriter` also remove the last
remaining bench for writing using `Display` because this is not the
correct place for that code - its trivial to re add later in the correct
module.
`Sighash` does not need to implement `Encodable` because it is
claimed (I don't know exactly myself) that `Sighash` is never consensus
encode in Bitcoin.
We are currently relying on `Sighash` to implement `Encodable` when
encoding creating the segwit v0 sighash for a single input.
For reference, from BIP143:
If sighash type is SINGLE and the input index is smaller than the
number of outputs, hashOutputs is the double SHA256 of the output
amount with scriptPubKey of the same index as the input;
We can use `write_all` directly to write the hashed bytes and remove the
implementation of `Encodable` from the `Sighash` type.
While we are at it, use `write_all` to write the zero hash also to make
the code more uniform and understandable.
Fix: #1549
Recently while trying to fix CI I (Tobin) added a call to `cargo clean`
and a code comment justifying it. This was merged because while not
incorrect it is redundant since calling `cargo` with different
`RUSTFLAGS` triggers a rebuild.
53ee42d7c1 Fix ASAN in CI (Tobin C. Harding)
Pull request description:
When we merged `hashes` into `rust-bitcoin` we ran the test script a couple of times using `./hashes/contrib/test.sh` this causes the `cargo` commands in the CI script to be run from the crate root which is not what we want.
This showed up recently because `cargo test` was getting run in `bitcoin` after building with address/memory sanitazation configured into the build.
While we are at it run `cargo clean` after the last sanitizer build just to future proof the CI script in case we later accidentally re-use the same build with plain old `cargo` (without `-Zbuild-std etc.`).
ACKs for top commit:
apoelstra:
ACK 53ee42d7c1
sanket1729:
ACK 53ee42d7c1
Tree-SHA512: 3d0b9aa35be6290808e2d23144e4dce467ab433862b375e1f164dc41261919e8e27e4cab117bbe9bfe18814ec7bf468e1c03674d20e9f36b5319e986fc329e89
When we merged `hashes` into `rust-bitcoin` we ran the test script a
couple of times using `./hashes/contrib/test.sh` this causes the `cargo`
commands in the CI script to be run from the crate root which is not
what we want.
This showed up recently because `cargo test` was getting run in
`bitcoin` after building with address/memory sanitazation configured
into the build.
While we are at it run `cargo clean` after the last sanitizer build just
to future proof the CI script in case we later accidentally re-use the same
build with plain old `cargo` (without `-Zbuild-std etc.`).
Tobin C. Harding
Andrew Poelstra
Arturo Marquez
sanket1729
Jiri Jakes
bnabi
