Repository design pattern for database access

[NikolasHaimerl]

          What do you think about using a Repository design pattern, where we have a Repository interface providing the data access methods, and a Repository implementation that holds the reference to the pgPool instance?

See what I did in piece-indexer:

https://github.com/filecoin-station/piece-indexer/blob/main/repository/lib/redis-repository.js

There is nothing wrong with your functional-programming approach here, it's a valid direction too.

Originally posted by @bajtos in #39 (comment)

[NikolasHaimerl]

@bajtos In light of the recent discussion of moving the database access layer: Do we want to proceed with the repository design pattern. The way I see it, we would still need to either have very specific read functions in the repository class or we use a generic method (similar to loadDeals) and only provide constant strings that defined the specific query details in the data layer. Something like this:

export const QUERY_DEAL_WITH_HIGHEST_ACTIVATED_AT_EPOCH:string ="..."
export class PostgreSQLClass{
...
read(query:string,parameters:[]string) -> Promise<Array<ActiveDealDbEntry>>
...

and we then call it like so:

repository.read(QUERY_DEAL_WITH_HIGHEST_ACTIVATED_AT_EPOCH,[...])