Expand description
Shared dump-loading model behind both output formats.
A dump is decoded once into a Loaded — its instance_id, schema registry, intern table, and
the flat list of records recovered from every shard (sorted into the global
(ts_nanos, shard_id, local_seq) order). The Parquet (crate::convert) and Chrome-trace
(crate::trace) writers both build on &[Loaded], so multi-dump merge is uniform: load each
input (in parallel), then hand the slice to the writer.
Structs§
- Loaded
- A single decoded dump: its identity, registry, intern table, and recovered records.
- Rec
- One decoded record, attributed to a schema within its
Loadeddump.