Glossary
| Term | Meaning |
|---|---|
| Synapse | This platform: console, API, workers, and docs |
| Datasource / config | Saved source + sink pair (configs table) |
| Source | Where raw items are discovered (poll/consume) |
| Sink | Where successful records are written |
| Item | One discovered file, object, message, or HTTP body |
| Schema / subject / version | JSON Schema family; new register = next version |
| Classifier | Step that validates a record against a schema id |
| Transformer | Step that maps, projects, decodes, or expands records |
| Reader | Turns an item’s bytes into records (NMEA or JSON) |
| Pipeline | Datasource + reader + ordered classifier/transformer chain |
| Custom transformer | Saved Python transform(record) referenced by id |
| Execution | One scheduled run of a pipeline |
| Playground | Inline test without datasource I/O |
| Stage error | Per-record failure that does not abort the whole run |
| Worker | Process that claims and runs executions |
| Job bundle | Self-contained claimed job (pipeline + schemas + decrypted config) |
| Lease | Time-bounded ownership of a claimed/running execution |
| NMEA / VDM / VDO | Maritime sentence formats; AIS decode needs VDM |
Not product entities: cron schedules, schema discovery, visualization, generic “connectors” beyond source/sink types.