bhwi-cli: command to enumerate devices by trevarj · Pull Request #18 · wizardsardine/bhwi

trevarj · 2026-03-10T16:09:09Z

The main goal of this change is to add a command to bhwi-cli that enumerates all
supported devices and displays information (ex. fingerprint) about them so that
the user can interact with a device.

Lots of tangential changes have also been made as I was trying to build out a
decent foundation for adding device enumeration.

Large refactors were done in order to make it possible to have heterogeneous
vectors of devices. It's not a major benefit but it makes the cli code nicer.
This includes the creation of the HWIDevice trait and making error types use
thiserror to implement std::error::Error. This also required the bhwi-wasm
errors to be changed from JsValue to a wrapper error, WasmError, which converts
the errors to strings. Not ideal, but could not find a better solution right
now.

Lots of the Transports from the e2e crates were moved around and changed for
re-usability between the cli crate e2e tests.

The rest of the changes were simply writing the Transports (HID and serial
backends) used for the currently supported devices.

…port On Guix System, using guix shell, the C_INCLUDE_PATH remains set but is missing the 32 bit headers. Unsetting it somehow allows cargo/cc to resolve things.

trevarj · 2026-03-11T09:34:37Z

+// generate this trait by putting it over HWI's definition and then also
+// generate the blanket impl which will map the errors to HWIDeviceError
+#[async_trait(?Send)]
+pub trait HWIDevice {


Trying to come up with a better solution to this before attempting to write a proc macro. Duplicating all the methods isn't so awful, but not very nice IMO

edouardparis · 2026-03-11T10:10:32Z

while compiling:

error[E0599]: no function or associated item named `from_utf8` found for type `str` in the current scope
   --> bhwi/src/jade/mod.rs:172:35
    |
172 |                     message: str::from_utf8(message)
    |                                   ^^^^^^^^^ function or associated item not found in `str`
    |
help: you are looking for the module in `std`, not the primitive type
    |
172 |                     message: std::str::from_utf8(message)
    |                              +++++

and after fixing, I tried running the list without any device.

cargo run -- device list
thread 'main' panicked at bhwi-cli/src/coldcard.rs:112:18:
couldn't connect to socket: Os { code: 2, kind: NotFound, message: "No such file or directory" }
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

trevarj · 2026-03-11T17:48:29Z

@edouardparis Thanks for trying it and reviewing! I got a bit lost in trying to create fancy futures stuff, that's why it crashes when it can't find a coldcard emulator.

message: str::from_utf8(message)

This is probably a rust version issue? It compiles for me on nightly. Perhaps I should start developing with stable.

Anyway, I resolved these issues so you can try again! bfbfe39

I think the flow of unlocking and setting the fingerprint is a little rough right now. Not sure on the best way to design that yet.

edouardparis · 2026-03-13T11:30:27Z

Some trait do not need to have more rigid requirement to work with you implementation. I have bhwi-cli compiling with this diff:

git diff
diff --git a/bhwi-async/src/lib.rs b/bhwi-async/src/lib.rs
index 082762f..b5e33d2 100644
--- a/bhwi-async/src/lib.rs
+++ b/bhwi-async/src/lib.rs
@@ -20,19 +20,19 @@ pub use ledger::Ledger;

 #[async_trait(?Send)]
 pub trait Transport {
-    type Error: StdError + Send + Sync + Debug + 'static;
+    type Error;
     async fn exchange(&mut self, command: &[u8], encrypted: bool) -> Result<Vec<u8>, Self::Error>;
 }

 #[async_trait(?Send)]
 pub trait HttpClient {
-    type Error: StdError + Send + Sync + Debug + 'static;
+    type Error;
     async fn request(&self, url: &str, request: &[u8]) -> Result<Vec<u8>, Self::Error>;
 }

 #[async_trait(?Send)]
 pub trait HWI {
-    type Error: StdError + Send + Sync + Debug + 'static;
+    type Error;
     async fn unlock(&mut self, network: Network) -> Result<(), Self::Error>;
     async fn get_master_fingerprint(&mut self) -> Result<Fingerprint, Self::Error>;
     async fn get_extended_pubkey(
@@ -77,11 +77,7 @@ impl HWIDeviceError {
 }

 #[derive(Debug, thiserror::Error)]
-pub enum Error<E, F>
-where
-    E: StdError + Send + Sync + 'static,
-    F: StdError + Send + Sync + 'static,
-{
+pub enum Error<E, F> {
     #[error("transport error: {0}")]
     Transport(E),

@@ -202,8 +198,8 @@ pub trait OnUnlock {
 }

 pub trait CommonInterface<C, T, R, E> {
-    type TransportError: StdError + Send + Sync + Debug + 'static;
-    type HttpClientError: StdError + Send + Sync + Debug + 'static;
+    type TransportError;
+    type HttpClientError;

     #[allow(clippy::type_complexity)]
     fn components(
@@ -220,8 +216,6 @@ async fn run_command<D, C, E, F>(
     command: C,
 ) -> Result<common::Response, Error<E, F>>
 where
-    E: StdError + Send + Sync + std::fmt::Debug + 'static,
-    F: StdError + Send + Sync + std::fmt::Debug + 'static,
     D: CommonInterface<
             common::Command,
             common::Transmit,

For the implementation of HWIDevice we will see later how it goes, If you look at bhwi-wasm I directly used enum instead of trying to keep a special trait. Maybe this enum can be removed later in favor of the HWIDevice trait.

The main goal of this change is to add a command to bhwi-cli that enumerates all supported devices and displays information (ex. fingerprint) about them so that the user can interact with a device. Lots of tangential changes have also been made as I was trying to build out a decent foundation for adding device enumeration. Large refactors were done in order to make it possible to have heterogeneous vectors of devices. It's not a major benefit but it makes the cli code nicer. This includes the creation of the HWIDevice trait and making error types use thiserror to implement std::error::Error. This also required the bhwi-wasm errors to be changed from JsValue to a wrapper error, WasmError, which converts the errors to strings. Not ideal, but could not find a better solution right now. Lots of the Transports from the e2e crates were moved around and changed for re-usability between the cli crate e2e tests. The rest of the changes were simply writing the Transports (HID and serial backends) used for the currently supported devices.

trevarj · 2026-03-13T13:43:10Z

Some trait do not need to have more rigid requirement to work with you implementation

This is true. Fixed.

trevarj added 2 commits March 9, 2026 11:56

website: gitignore ts build info files

307a01f

bhwi-wasm: set env vars for when glibc wasn't build with multilib sup…

ea9b12b

…port On Guix System, using guix shell, the C_INCLUDE_PATH remains set but is missing the 32 bit headers. Unsetting it somehow allows cargo/cc to resolve things.

trevarj requested a review from edouardparis March 11, 2026 07:56

edouardparis marked this pull request as ready for review March 11, 2026 09:18

trevarj commented Mar 11, 2026

View reviewed changes

edouardparis reviewed Mar 11, 2026

View reviewed changes

Comment thread bhwi/src/common.rs Outdated

Comment thread bhwi-cli/src/bin/bhwi.rs Outdated

Comment thread bhwi-cli/src/lib.rs Outdated

trevarj force-pushed the cli-enumerate branch from b4c9dfc to 0f1ee80 Compare March 11, 2026 17:47

trevarj force-pushed the cli-enumerate branch from 0f1ee80 to bfbfe39 Compare March 11, 2026 17:50

edouardparis reviewed Mar 13, 2026

View reviewed changes

Comment thread bhwi-cli/src/bin/bhwi.rs Outdated

trevarj force-pushed the cli-enumerate branch from bfbfe39 to 00f5856 Compare March 13, 2026 13:04

trevarj requested a review from edouardparis March 13, 2026 13:43

edouardparis approved these changes Mar 13, 2026

View reviewed changes

edouardparis merged commit 1791cbc into wizardsardine:main Mar 13, 2026
7 checks passed

edouardparis mentioned this pull request Mar 13, 2026

bhwi-cli: list connected devices #12

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

bhwi-cli: command to enumerate devices#18

bhwi-cli: command to enumerate devices#18
edouardparis merged 3 commits intowizardsardine:mainfrom
trevarj:cli-enumerate

trevarj commented Mar 10, 2026

Uh oh!

trevarj Mar 11, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

edouardparis commented Mar 11, 2026

Uh oh!

trevarj commented Mar 11, 2026 •

edited

Loading

Uh oh!

edouardparis commented Mar 13, 2026

Uh oh!

Uh oh!

trevarj commented Mar 13, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

trevarj commented Mar 10, 2026

Uh oh!

trevarj Mar 11, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

edouardparis commented Mar 11, 2026

Uh oh!

trevarj commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

edouardparis commented Mar 13, 2026

Uh oh!

Uh oh!

trevarj commented Mar 13, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

trevarj commented Mar 11, 2026 •

edited

Loading