35 fix periodic disc #47
Conversation
Let the variable really express what it means. It's current only reader can't be meaningfully invoked before a CONNECT packet was sent so there is no need to update the reader's code.
We don't want to let other objects besides the sender to set last_packet_sent_at.
…fine. Let the variable really express what it means. It's current only reader can't be meaningfully invoked before a CONNACK packet was received so there is no need to update the reader's code.
We don't want to let other objects besides the handler to set last_packet_received_at.
…e in the return value (currently not used in any caller) if we've actually not sent the packet.
Situation before this change: If you keep publishing (only) messages with QoS 0 in the time frame of keep_alive interval, then you update sender.last_packet_sent_at, but you'll get no response so you'll never update handler.last_packet_received_at. So you send no PINGREQ packets in connection_helper.check_keep_alive and eventually the check "disconnect_timeout_at <= now" in that method triggers a disconnect. Fix this by sending PINGREQ packet also if the client haven't recently received any packet from the server (besides sending PINGREQ packet if the client haven't recently sent any packet to the server). Avoid sending more PINGREQ packets before the client receives a PINGRESP to the first one.
p-goudet
force-pushed
the
35_fix_periodic_disc
branch
from
3949cef
to
694e03b
Compare
|
Any updates on this? :) |
|
I am sorry I have been busy with other project lately. I would try to go back on this as soon as possible and finally merge it. 🙇 |
|
Any news? :P |
|
I will to try to merge it by the end of this week. I have to review some local changes and check if I include them or not before that 🙇 |
|
Thank you, that would be great, will test as soon as it's released! :) |
|
The fix seems to be working, however I got some performances issues while the client is disconnected from the server. Edit: The performance issue seems to be related with other part. I merge this change |
|
Thank you :) - is there an ETA for releasing paho-mqtt-1.0.13? |
|
I don't know yet, very soon I hope :). |
|
Thank you, looking forward :) |
|
Friendly follow up :P |
|
Any update? :) |
|
This branch has been already merged on the master branch. Other change would be add to the master branch before the release of 1.0.13 version. You may download the source and install the gems by hand to get the latest update. |
Fix the check alive process. The current version has a bug in handling of ping packet in case of sending qos = 0 package.