Variable CATALOG_OBJECT_COLUMNSConst
CATALOG_OBJECT_COLUMNS: readonly [
"ID",
"IntegrationID",
"Name",
"DisplayName",
"Description",
"Category",
"APIPath",
"ResponseDataKey",
"DefaultPageSize",
"SupportsPagination",
"PaginationType",
"SupportsIncrementalSync",
"SupportsWrite",
"DefaultQueryParams",
"Configuration",
"Sequence",
"Status",
"WriteAPIPath",
"WriteMethod",
"DeleteMethod",
"IsCustom",
"CreateAPIPath",
"CreateMethod",
"CreateBodyShape",
"CreateBodyKey",
"CreateIDLocation",
"UpdateAPIPath",
"UpdateMethod",
"UpdateBodyShape",
"UpdateBodyKey",
"UpdateIDLocation",
"DeleteAPIPath",
"DeleteIDLocation",
"IncrementalWatermarkField",
"MetadataSource",
"SupportsCreate",
"SupportsUpdate",
"SupportsDelete",
"SyncStrategy",
"ContentHashApplicable",
"StableOrderingKey",
"CompanyIntegrationID",
"IntegrationObjectID",
"Provenance",
"ProvenanceDetail",
"IsSelected",
"SelectedAt",
"FirstSeenAt",
"LastSeenAt",
"LastSampledAt",
] = ...
Columns the object projection reads. The first 43 are IntegrationObject's own, copied verbatim so a per-connection row answers every question a shared row answers; the rest are what per-connection rows add.
IntegrationIDis deliberately among the first group and denormalised onto the table: connectors read it off the object, and a join per read would be the wrong trade.