dl_files, dl_database: avoid using multiple processes

The standard multiprocessing module is used to distribute a task to
multiple processes, which is useful when doing heavy computation due
to the limitations of CPython; however, making this work is dependent
on the ability to fork processes or else to kludgily emulate forking
on systems that don't support it.  In particular, it tends to cause
problems on Windows unless you are very scrupulous about how you write
your program.

Therefore, as a rule, the multiprocessing module shouldn't be used by
general-purpose libraries, and should only be invoked by application
programmers themselves (who are in a position to guarantee that
imports have no side effects, the main script uses 'if __name__ ==
"__main__"', etc.)

However, downloading a file isn't a CPU-bound task, it's an I/O-bound
task, and therefore for this purpose, parallel threads should work as
well or even better than parallel processes.  The
multiprocessing.dummy module provides the same API as the
multiprocessing module, but uses threads instead of processes, so it
should be safe to use in a general-purpose library.

Author: Benjamin Moody

wfdb/io/download.py

@@ -1,4 +1,4 @@
-import multiprocessing
+import multiprocessing.dummy
 import numpy as np
 import re
 import os
@@ -539,7 +539,7 @@ def dl_files(db, dl_dir, files, keep_subdirs=True, overwrite=False):
     print("Downloading files...")
     # Create multiple processes to download files.
     # Limit to 2 connections to avoid overloading the server
-    pool = multiprocessing.Pool(processes=2)
+    pool = multiprocessing.dummy.Pool(processes=2)
     pool.map(dl_pn_file, dl_inputs)
     print("Finished downloading files")
 

wfdb/io/record.py

@@ -1,5 +1,5 @@
 import datetime
-import multiprocessing
+import multiprocessing.dummy
 import posixpath
 import re
 
@@ -5279,7 +5279,7 @@ def dl_database(
     print("Downloading files...")
     # Create multiple processes to download files.
     # Limit to 2 connections to avoid overloading the server
-    pool = multiprocessing.Pool(processes=2)
+    pool = multiprocessing.dummy.Pool(processes=2)
     pool.map(download.dl_pn_file, dl_inputs)
     print("Finished downloading files")