In https://github.com/estuary/connectors/pull/1563 and https://github.com/estuary/connectors/pull/1572, support for base64-encoded strings to be materialized as binary columns was added to materialize-databricks, for non-key fields. This required some wrangling of the staged textual data into a binary column, and managing pre-existing materializations which have always materialized these fields as strings previously.
At some point we should do the same for other materializations. This is not likely to be a quick thing for any of them, so it will take some time, and for now I am just logging this issue as something to follow up on. It would be significantly easier if we had a good strategy for changing materialized column types in a backward-compatible way, which may eventually emerge.
In https://github.com/estuary/connectors/pull/1563 and https://github.com/estuary/connectors/pull/1572, support for base64-encoded strings to be materialized as binary columns was added to
materialize-databricks
, for non-key fields. This required some wrangling of the staged textual data into a binary column, and managing pre-existing materializations which have always materialized these fields as strings previously.At some point we should do the same for other materializations. This is not likely to be a quick thing for any of them, so it will take some time, and for now I am just logging this issue as something to follow up on. It would be significantly easier if we had a good strategy for changing materialized column types in a backward-compatible way, which may eventually emerge.