Better behavior for "Invalid Descriptor Index" problem

[krlmlr]

I'm now convinced that the problem must be solved by using an appropriate driver, or putting long columns at the end of the query.

Still: Can we do better than throwing "Invalid Descriptor Index" at the user?

library(DBI)
library(tidyverse)

con <- dbConnect(odbc::odbc(), "mssql-test-ms", uid = "kirill", pwd = keyring::key_get("mssql", "kirill"))

data <- tibble(a = 1:3, b = "a", c = 1)

copy_to(con, data, name = "##data1", types = c("integer", "varchar(max)", "real")) %>% collect()
#> Error in result_fetch(res@ptr, n): nanodbc/nanodbc.cpp:2908: 00000: [Microsoft][ODBC Driver 17 for SQL Server]Invalid Descriptor Index
#> Warning in dbClearResult(res): Result already cleared

^{Created on 2020-04-28 by the reprex package (v0.3.0)}

Follow-up to #309.

[jimhester]

Should be fixed by #381