Exceptions caused by batch_submit may not be handled properly.

[warownia1]

When the batch_run is called it calls batch submit and returns a generator object. Any errors raised while iterating over the generator are not caught by batch_run but are raised in the caller context.

Example:

def getval(x): raise RuntimeError

def gen():
    try:
         return (getval(x) for x in range(3))
     except:
         print('exception handled')

for _ in gen(): pass  # raises RuntimeError

Changing the signature of gen to:

def gen():
    for i in range(3):
         try:
             yield getval(i)
         except:
             print('exception handled')

handles the exceptions

[warownia1]

batch_submit runs the jobs immediately end returns an iterable of jobs statuses. Even if the iterable may defer status retrieval, the requirement for batch_run implementations is that it returns after the jobs are started.