Easier SQL backends

[ggpinto]

Here I changed cache_postgres to facilitate new SQL backends, what I have done is:

• In the test file I have only added a call to {memoise} to facilitate interactive tests
• As many checks and data were moved to private functions to facilitate the inclusion of new SQL backends. MS SQL Server can, for example, be added like so (I will do it and update the documentation if you are OK with this change):

cache_ms_sql <- R6::R6Class(
  "cache_ms_sql",
  inherit = cache_postgres,
  public = list(
    #' @description
    #' Start a new Microsoft SQL Server cache
    #' @param ... Parameters passes do DBI::dbConnect(odbc::odbc(), ...)
    #' @param cache_table On `initialize()`, the cache object will create a table
    #' to store the cache. Default name is `bankrcache`. Change it if you already
    #' have a table named `bankrcache` in your DB.
    #' @param algo for `{memoise}` compatibility, the `digest()` algorithm
    #' @param compress for `{memoise}` compatibility, should the data be compressed?
    #' @return A cache_ms_sql object
    initialize = function(...,
                          cache_table = "bankrcache",
                          algo = "sha512",
                          compress = FALSE) {

      private$check_dependencies("cache_ms_sql", "odbc")

      private$interface <- private$connect(odbc::odbc(), ...)

      private$sql_column_data = list(
            column_id = list(type = "varchar(max)", type_description = "varchar"),
            column_cache = list(type = "varbinary(max)", type_description = "varbinary")

      private$cache_table <- cache_table

      private$check_table()

      private$algo <- algo

      private$compress <- compress
    }
  )
)

[ColinFay]

Hey @ggpinto,

thanks for that. I'm on holidays rn so i'll have a look into that next week :)