🔖 zbus 5.2.0

• ⚡️ Optimizations:
  • Don't use proxies in connection code. This allows LTO to remove the proxy-side code from the
    binary if the users isn't directly using proxy APIs. On busd, this reduces the binary size by
    200KB.
  • Disable docs in fdo interface introspection. These interfaces are well-known and documented (in
    the spec itself), so we don't need to provide the docs in the introspection data at runtime.
  • interface now allows disabling docs in introspection. Introspection strings are part of the
    binary and can have an impact on the binary size. This change allows to disable the addition of
    docs in the introspection for cases where there interface is well-known/well-documented. For
    example, combined with the following change of using this in the zbus::fdo interfaces, this
    will allow busd binary to shed 26 KB.
  • Avoid some formatting at runtime in interface macro. Add values already known to the
    introspection XML at compile time. This also removes about 500 bytes from busd binary size.
• 🚩 Add gvariant feature flag to zbus_macros. It's just a proxy feature for zvariant and
  only needed to fix the build in the root workspace when building with --features gvariant.
  #1125.
• ⬆️ Update tokio-vsock to 0.6.
• 🔊 Skip self in an instrumented method.
• 🔥 Drop now unused sink feature of futures-util.

🔖 zbus 5.1.1

• ➖ Tie async-process dependency to async-io feature, instead of the target OS. This fixes
  tokio users (a big majority of our users) needlessly depending on it along with a bunch of its
  dependencies.
• 🔊 Don't log error on broadcasting on generic message channel. Otherwise, we just end up with a
  lot of noise in the logs.

zvariant_utils 3.0.2

• 📝 Fix documentation of signature::Error.
• ⬇️ Reduce rust-version to 1.80. #1107

🔖 zvariant 5.1.0

• 🏷️ Implement Clone for *Value. #1078
• 🚑️ Fix a few potential panics. #1094
• 🐛 Handle all variants in Error's PartialEq impl.
• ♻️ Use winnow for validation of ObjectPath. #1070
• 🥅 Add Error::InvalidObjectPath variant.
• ➕ Add direct dep on winnow.
• ⬇️ Reduce rust-version to 1.80. #1107

🔖 zbus_xmlgen 5.0.1

• ⬇️ Reduce rust-version to 1.80. #1107

🔖 zbus_xml 5.0.1

• ⬇️ Reduce rust-version to 1.80. #1107

🔖 zbus_names 4.1.0

• ⚡️ Use winnow for validation of all names. This greatly simplifies the code and in most
  cases, also much faster. #1070.
• 🥅 More appropriate error on bus name parsing failure.
• 🚸 More specific error on invalid name conversion.
• 🏷️ Add Error::InvalidNameConversion.
• 🗑️ Deprecate specific invalid name variants of Error.
• 🚸 Use Error::InvalidName variant for invalid name everywhere.
• 🏷️ Add Error::InvalidName variant and all name parsing failures now return this variant.
• ⬇️ Reduce rust-version to 1.80. #1107
• ➕ Add direct dep on winnow.

🔖 zbus 5.1.0

• ⬇️ Reduce rust-version to 1.80. #1107
• 🦺 Verify validity of names and object paths pased to macros at compile time.
• 🐛 Generate interface properties signal methods only when needed.
• ♻️ Use winnow for parsing GUID and D-Bus address strings. #1070
• ✨ Add support for unixexec transport.
• 🔌 Add Command stdio socket type. This socket communicates with a spawned child process via
  its standard input and output streams.
• ➕ Add direct dep on winnow.

🔖 zvariant_utils 3.0.1

• ⚡️ Switch from nom to winnow for signature parsing. This brings in 2 benefits:
  • The signature parsing code is now up to 30% faster. This directly translates to
    encoding/decoding speed of variants (used extensively in the D-Bus/gvariant world). This is the
    path that saw the least optimization in 5.0.0.
  • We'll now be able to remove nom and minimal-lexical deps (recently introduced in 5.0.0).
    Apparently winnow was already our indirect dependency so no new deps needed.
• ➕ Add a direct dep on winnow.
• ➖ Drop now unneeded nom dependency.

🔖 zvariant 5.0.1

• 📌 Don't pin zvariant_utils dep. We don't always release zvariant_utils together with other
  crates, so unlike the macros crates, the versions don't need to be tied together. This will also
  allow us to bring zvariant_utils improvements to users without having to release all dependent
  packages as well.