Open kevin-dp opened 1 week ago
@alco @kevin-dp the postgres algorithm depends on the server encoding as well - should we grab the encoding in the metadata acquisition stage, or just default to UTF8 for start?
I've implemented a basic Electric.Postgres.Identifier.parse
in https://github.com/electric-sql/electric/pull/1829 to serve the immediate needs of that feature.
Postgres makes use of it's server encoding and max identifier character length configuration, so we'd need to retrieve those as well as part of our "prep" and pass them along to the parsing routine. I will address that in a separate PR, probably overlaps with @alco 's work on putting all prep queries in the replication connection.
@alco @kevin-dp the postgres algorithm depends on the server encoding as well - should we grab the encoding in the metadata acquisition stage, or just default to UTF8 for start?
In the spirit of "any complex system that works has evolved from a simpler system that worked" let's start with assuming UTF8 first.
As proposed by @alco in https://github.com/electric-sql/electric/issues/1764#issuecomment-2391485961: