[facontidavide]

parser_data_tamer now reads recordings of data_tamer 2.x as well as 1.x (manifest 0.10.0). data_tamer 2.x is a new major version, developed at https://github.com/facontidavide/data_tamer; PlotJuggler 4 is the release that reads its files.

Changes

• Dependency: the CPM download of PickNikRobotics/data_tamer 1.0.3 becomes the data_tamer 2.x sources. They are pinned by commit and SHA256 (7b4dc82), since 2.0.0 is not tagged yet; a comment says the pin moves to the 2.0.0 tag once it is released. The include directory is data_tamer_cpp/include (data_tamer_v2/data_tamer_parser.hpp).
• Parser API, 2.x:
  • BuildSchemaFromText() replaces the misspelled BuilSchemaFromText().
  • SplitMcapMessage() replaces the plugin's hand-written MCAP body framing.
  • The [[nodiscard]] bool result of ParseSnapshot() is checked, and malformed input becomes an error status instead of a crash. Everything is caught as std::exception, since this is a C ABI boundary.
  • The new VarNumber alternatives map to their own types: bool stays boolean, and char becomes a uint8 byte.
• Schemas:
  • read: version 4 (data_tamer 1.x), 5 (2.x line text) and 6 (the YAML rendering used by the 2.x ROS 2 publisher);
  • rejected, with a clear error: version-less schema texts from releases before 0.7.0.
• Tests: 23 cases.
  • Golden tests against the data_tamer wire-format vectors, read from the pinned sources rather than copied: text and YAML, plain and nested, full and masked snapshots, every field checked against expected.json.
  • data_tamer 1.x schema texts.
  • Malformed-input tests: oversized mask size, truncation, oversized payload size, version-less text.
• Docs: CHANGELOG, README and manifest version.

Tested

Release build with 0 warnings; 23/23 gtest and 7/7 ctest, with the pinned archive downloaded and hash-checked by CPM; pre-commit clean.

Follow-ups

• PJ4's WASM build compiles these plugin sources with its own data_tamer 1.0.3 (cmake/PjWasmDependencies.cmake) and never runs this CMake. When PJ4 updates its pin of this repository, it has to move to the same data_tamer 2.x commit.
• Faster parser: data_tamer main has since made the parser 2 to 7.6 times faster, with the same API. Updating the pin to a newer commit, or to the 2.0.0 tag, picks that up.

🤖 Generated with Claude Code

https://claude.ai/code/session_01UHSVkk2paVU7LhxqzUoHAH