overwrite_missing_table spec should use iris_in instead of iris in dbWriteTable

[martinstuder]

DBItest/R/spec-sql-write-table.R

Lines 151 to 154 in 7bd0ae1

	overwrite_table_missing = function(ctx, con, table_name) {
	iris_in <- get_iris(ctx)
	expect_error(
	dbWriteTable(con, table_name, iris[1:10, ], overwrite = TRUE),

The call to dbWriteTable above should use iris_in instead of iris since the expect_equal_df call uses iris_in.

[krlmlr]

Thanks. What problems are you seeing with the current implementation?

[martinstuder]

@krlmlr I was investigating a failure in that test case (overwrite_table_missing) that I thought was related to a difference between datasets::iris and the result of DBItest:::get_iris. It turns out the test case actually passes and the issue was a different one unrelated to DBItest. If I understand the test case correctly, it would still seem safer to use iris_in in case the definition of DBItest:::get_iris should change.

Partially related to that may be that DBItest:::get_iris currently includes an as.character conversion on Species that doesn't have any effect since it performs the conversion on a local copy of iris instead of datasets_iris:

DBItest/R/utils.R

Line 121 in 3c37762

iris$Species <- as.character(iris$Species)

[krlmlr]

Good catch, thanks!

[github-actions]

This old thread has been automatically locked. If you think you have found something related to this, please open a new issue and link to this old issue if necessary.