Closed thelee closed 4 years ago
i think we'll just have to pull in the "last_updated" from schema-less and have to just specify a manual override for the initial table generation. the only reason why "last_updated" doesn't work is because we're working with past data. with regular pipeline runs, last_updated should exactly equal the freshness
no, that doesn't work, because due to latency of some data sources we wouldn't be able to just turn off the override when we launch the actual pipeline
some dates are being populated with speculative future dates and so are not a good signal of the last time a data source has been updated. spend some time thinking about/fixing this.