The data transmission solution is not for error handling
The code review
“Bro, can you help to review this link again? I have fixed your comment ‘add some error handling here’” A teammate asked.
“Sure, let me take a look. could you explain why ‘smart_sleep’ could help handle the error”? I asked.
def smart_sleep(error_code):
tried = 1
max_try = 1024 # try 10 times
sleep_time = 500
while tried < max_try:
error_code = service_call(...)
if error_code == 0: # all good
return
if error_code == 500:
sleep(sleep_time*tried)
...
tried *= 2
....“Okay. As you can see I added retry when calling the service, also I added sleep time based on the error code, if all good then no sleep otherwise I will retry longer and longer time,” he said.
“Okay but why do you sleep longer and longer time for the error?” I asked.
“Because this idea is from ‘tcp congestion control’ — slow start, the error should be due to xxx which is something known and it could happen so I retry a few times it may work and I do not want to load the server so I put sleep longer and longer,” he said.
“Well bro, that is for a different issue. but let’s clarify error handling first. when an error happens, if it is something not expected or unknown to this code layer, just raise it up; if it is expected like you mentioned ‘common error’…
