Currently when parsing files which use \r\n to separate lines on Windows, the query-result separator line looks like ----\r. This fails the if line == "----" { check, resulting in the whole file being swallowed as part of the query. The built-in lines method fixes this issue by splitting on both \n and \r\n.
The change in the result for the include_glob test is due to lines() trimming a trailing newline from one of the included files.
Should fix https://github.com/apache/arrow-datafusion/issues/4494
Currently when parsing files which use
\r\nto separate lines on Windows, the query-result separator line looks like----\r. This fails theif line == "----" {check, resulting in the whole file being swallowed as part of the query. The built-inlinesmethod fixes this issue by splitting on both\nand\r\n.The change in the result for the
include_globtest is due tolines()trimming a trailing newline from one of the included files.