Fix TestCompletion.py failure on Darwin after r246639

[chying]

Summary

• send empty line for multiple commands to complete

[tberghammer]

I don't understand why it is necessary on OSX but I am fine with this change if it fixes the issue (at the moment I am struggling with building lldb on OSX so I can't test it).

One question: How many empty lines should we send? Is it always enough to send 2, or should we send the same number of empty lines then the number of lines in the original command?

[chying]

Based on experiment I had, we need to send empty lines the same times as the number of lines in original commands on Darwin.
But on Linux, only one empty line is needed for multiple lines of commands.
It seems on Darwin, lldb doesn't automatically execute commands separated by new line.

lldb/trunk/test/functionalities/completion/TestCompletion.py

def complete_from_to(self, str_input, patterns, turn_off_re_match=False):
                 child.logfile_read = f_read
 
                 child.expect_exact(prompt)
                 child.setecho(True)
                 # Sends str_input and a Tab to invoke the completion machinery.
                 child.send("%s\t" % str_input)
                 child.sendline('')
                 child.expect_exact(prompt)
+                child.sendline('')
+                child.expect_exact(prompt)
 
         # Now that the necessary logging is done, restore logfile to None to
         # stop further logging.
         child.logfile_send = None
         child.logfile_read = None
         
         with open('child_send.txt', 'r') as fs:
             if self.TraceOn():