aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rwxr-xr-xtests/runner.py3
-rw-r--r--tools/js_optimizer.py25
2 files changed, 24 insertions, 4 deletions
diff --git a/tests/runner.py b/tests/runner.py
index 864cde3d..3d5b7dc9 100755
--- a/tests/runner.py
+++ b/tests/runner.py
@@ -5934,6 +5934,9 @@ def process(filename):
del os.environ['EMCC_DEBUG']
shutil.copyfile('src.c.o.js', 'debug.js')
self.assertIdentical(open('release.js').read().replace('\n\n', '\n').replace('\n\n', '\n'), open('debug.js').read().replace('\n\n', '\n').replace('\n\n', '\n')) # EMCC_DEBUG=1 mode must not generate different code!
+ print >> sys.stderr, 'debug check passed too'
+ else:
+ print >> sys.stderr, 'not doing debug check because already in debug'
def test_python(self):
if Settings.QUANTUM_SIZE == 1: return self.skip('TODO: make this work')
diff --git a/tools/js_optimizer.py b/tools/js_optimizer.py
index 3fdabfa8..0c4eb7be 100644
--- a/tools/js_optimizer.py
+++ b/tools/js_optimizer.py
@@ -9,6 +9,8 @@ JS_OPTIMIZER = path_from_root('tools', 'js-optimizer.js')
BEST_JS_PROCESS_SIZE = 1024*1024
+WINDOWS = sys.platform.startswith('win')
+
def run_on_chunk(command):
filename = command[2] # XXX hackish
output = subprocess.Popen(command, stdout=subprocess.PIPE).communicate()[0]
@@ -59,12 +61,27 @@ def run(filename, passes, js_engine):
commands = map(lambda chunk: [js_engine, JS_OPTIMIZER, chunk] + passes, chunks)
if len(chunks) > 1:
- cores = min(multiprocessing.cpu_count(), chunks)
- if os.environ.get('EMCC_DEBUG'): print >> sys.stderr, 'splitting up js optimization into %d chunks, using %d cores' % (len(chunks), cores)
- pool = multiprocessing.Pool(processes=cores)
+ # We are splitting into chunks. Hopefully we can do that in parallel
commands = map(lambda command: command + ['noPrintMetadata'], commands)
- filenames = pool.map(run_on_chunk, commands, chunksize=1)
filename += '.jo.js'
+
+ fail = None
+ cores = min(multiprocessing.cpu_count(), chunks)
+ if cores < 2:
+ fail = 'python reports you have %d cores' % cores
+ elif WINDOWS:
+ fail = 'windows (see issue 663)'
+
+ if not fail:
+ # We can parallelize
+ if os.environ.get('EMCC_DEBUG'): print >> sys.stderr, 'splitting up js optimization into %d chunks, using %d cores' % (len(chunks), cores)
+ pool = multiprocessing.Pool(processes=cores)
+ filenames = pool.map(run_on_chunk, commands, chunksize=1)
+ else:
+ # We can't parallize, but still break into chunks to avoid uglify/node memory issues
+ if os.environ.get('EMCC_DEBUG'): print >> sys.stderr, 'splitting up js optimization into %d chunks (not in parallel because %s)' % (len(chunks), fail)
+ filenames = [run_on_chunk(command) for command in commands]
+
f = open(filename, 'w')
for out_file in filenames:
f.write(open(out_file).read())