Normalize line endings in SkSL error tests

On Windows, the process output contains CRLF sequences. Calling
splitlines() recognizes those (Python universal line endings), so we get
a list of the output lines. Joining those with '\n' produces a string
that just has LF. (We add another one, so we still end the file with a
newline). Because we've opened the file in binary mode, the resulting
files now look the same, regardless of OS.

Change-Id: I1f032aa6e0f82057f593c25ce0676983733b9e56
Reviewed-by: John Stiles <>
Commit-Queue: Brian Osman <>
1 file changed