Open tnull opened 1 month ago
Please note how ridiculous the dependency tree (which in a Bitcoin context should always also be read as 'increased security risk') is that async_std
introduced -- all for a single method:
esplora-client v0.10.0 (/Users/erohrer/workspace/rust-esplora-client)
├── async-std v1.13.0
│ ├── async-channel v1.9.0
│ │ ├── concurrent-queue v2.5.0
│ │ │ └── crossbeam-utils v0.8.19
│ │ ├── event-listener v2.5.3
│ │ └── futures-core v0.3.30
│ ├── async-global-executor v2.4.1
│ │ ├── async-channel v2.3.1
│ │ │ ├── concurrent-queue v2.5.0 (*)
│ │ │ ├── event-listener-strategy v0.5.2
│ │ │ │ ├── event-listener v5.3.1
│ │ │ │ │ ├── concurrent-queue v2.5.0 (*)
│ │ │ │ │ ├── parking v2.2.1
│ │ │ │ │ └── pin-project-lite v0.2.13
│ │ │ │ └── pin-project-lite v0.2.13
│ │ │ ├── futures-core v0.3.30
│ │ │ └── pin-project-lite v0.2.13
│ │ ├── async-executor v1.13.1
│ │ │ ├── async-task v4.7.1
│ │ │ ├── concurrent-queue v2.5.0 (*)
│ │ │ ├── fastrand v2.0.1
│ │ │ ├── futures-lite v2.3.0
│ │ │ │ ├── fastrand v2.0.1
│ │ │ │ ├── futures-core v0.3.30
│ │ │ │ ├── futures-io v0.3.30
│ │ │ │ ├── parking v2.2.1
│ │ │ │ └── pin-project-lite v0.2.13
│ │ │ └── slab v0.4.9
│ │ │ [build-dependencies]
│ │ │ └── autocfg v1.1.0
│ │ ├── async-io v2.3.4
│ │ │ ├── async-lock v3.4.0
│ │ │ │ ├── event-listener v5.3.1 (*)
│ │ │ │ ├── event-listener-strategy v0.5.2 (*)
│ │ │ │ └── pin-project-lite v0.2.13
│ │ │ ├── cfg-if v1.0.0
│ │ │ ├── concurrent-queue v2.5.0 (*)
│ │ │ ├── futures-io v0.3.30
│ │ │ ├── futures-lite v2.3.0 (*)
│ │ │ ├── parking v2.2.1
│ │ │ ├── polling v3.7.3
│ │ │ │ ├── cfg-if v1.0.0
│ │ │ │ ├── rustix v0.38.31
│ │ │ │ │ ├── bitflags v2.4.2
│ │ │ │ │ ├── errno v0.3.8
│ │ │ │ │ │ └── libc v0.2.152
│ │ │ │ │ └── libc v0.2.152
│ │ │ │ └── tracing v0.1.40
│ │ │ │ ├── pin-project-lite v0.2.13
│ │ │ │ └── tracing-core v0.1.32
│ │ │ │ └── once_cell v1.19.0
│ │ │ ├── rustix v0.38.31 (*)
│ │ │ ├── slab v0.4.9 (*)
│ │ │ └── tracing v0.1.40 (*)
│ │ ├── async-lock v3.4.0 (*)
│ │ ├── blocking v1.6.1
│ │ │ ├── async-channel v2.3.1 (*)
│ │ │ ├── async-task v4.7.1
│ │ │ ├── futures-io v0.3.30
│ │ │ ├── futures-lite v2.3.0 (*)
│ │ │ └── piper v0.2.4
│ │ │ ├── atomic-waker v1.1.2
│ │ │ ├── fastrand v2.0.1
│ │ │ └── futures-io v0.3.30
│ │ ├── futures-lite v2.3.0 (*)
│ │ └── once_cell v1.19.0
│ ├── async-io v2.3.4 (*)
│ ├── async-lock v3.4.0 (*)
│ ├── crossbeam-utils v0.8.19
│ ├── futures-core v0.3.30
│ ├── futures-io v0.3.30
│ ├── futures-lite v2.3.0 (*)
│ ├── kv-log-macro v1.0.7
│ │ └── log v0.4.20
│ │ └── value-bag v1.9.0
│ ├── log v0.4.20 (*)
│ ├── memchr v2.7.1
│ ├── once_cell v1.19.0
│ ├── pin-project-lite v0.2.13
│ ├── pin-utils v0.1.0
│ └── slab v0.4.9 (*)
Thanks for raising the issue @tnull
In #98, a dependency on the
async_std
runtime was introduced. IIUC, this was done only to use itssleep
method.I don't think
rust-esplora-client
should make choices regarding which async runtime users should use. If anything, it should be using the by now defaulttokio
rather thanasync_std
.I believe I already brought this up in the discussion of the precursor #71, which however seems now gone for some reason.
An alternative to pulling in an extra dependency for this would simply be to have the user specify a
sleep
callback function that would use whatever sleep functionality they have available in their runtime.(cc @notmandatory @ValuedMammal)