Add a name parameter to allow for different fixtures

[Kjir]

It would be quite handy to be able to test the behaviour of the code with different results from the query: what if the table is empty? What if the table has more (or less) rows than the pagination limit?

I would expect to be able to provide a name to the asset, so that the same query can be registered with different outcomes. This would probably indirectly solve #159 as well.

Example code:

start_db_capturing("no-airlines")

con <- DBI::dbConnect(
  RPostgres::Postgres(),
  dbname = "nycflights"
)

get_an_airline(con)
DBI::dbDisconnect(con)

stop_db_capturing()

And test:

with_mock_db({
  con <- DBI::dbConnect(
    RPostgres::Postgres(),
    dbname = "nycflights"
  )
  
  test_that("We get no airline", {
    one_airline <- get_an_airline(con)
    expect_s3_class(one_airline, "data.frame")
    expect_equal(nrow(one_airline), 0)
  })
}, case = "no-airlines")

This would allow to use dittodb to run integration tests as well, instead of being able to test only the DB-to-R portion of the code.

In theory this could be achieved using with_mock_path, but it does feel clunkier

[jonkeane]

Thanks for the issue + the great reprex.

I think this is actually quite close to being a reality. start_db_capturing() already takes a path that it will save to, so that should be good. We could then add the case argument¹ to with_mock_db() that then uses with_mock_path() internally when it is called.

Might you be interested in sending a PR that does that?

Footnotes

1. At the risk of bikeshedding: I totally see where case comes from, but I wonder if it might be better to call this path to match that use elsewhere in the package? ↩