Hi, I think python / chapter-2 / hello_world_producer_pubconfirm.py / Line 47 is a bug although my python knowledge isn't great so Ive not added a pull request until this is confirmed?
If the msg_ids set is empty i.e. all previous messages have been ack'ed msg_ids.append(len(msg_ids) + 1) looks to append 1 to the set when a new message is published. Later at line 31 if a successful ack is received we try and remove the delivery_tag from the set which i believe wont exist as delivery_tag is a incrementing seq no and therefore not equal to 1 as appended above. I think the src should be calling channel.next_publish_seqno (or whatever the python equivalent is) and appending this to the set instead. The blog example here http://www.rabbitmq.com/blog/2011/02/10/introducing-publisher-confirms/ uses such an approach.
:-)
Hi, I think python / chapter-2 / hello_world_producer_pubconfirm.py / Line 47 is a bug although my python knowledge isn't great so Ive not added a pull request until this is confirmed?
If the msg_ids set is empty i.e. all previous messages have been ack'ed msg_ids.append(len(msg_ids) + 1) looks to append 1 to the set when a new message is published. Later at line 31 if a successful ack is received we try and remove the delivery_tag from the set which i believe wont exist as delivery_tag is a incrementing seq no and therefore not equal to 1 as appended above. I think the src should be calling channel.next_publish_seqno (or whatever the python equivalent is) and appending this to the set instead. The blog example here http://www.rabbitmq.com/blog/2011/02/10/introducing-publisher-confirms/ uses such an approach.
:-)