This is not an ideal solution, but it makes sure sending a batch of notifications will not fail halfway through, because of 1 illegal device token.
This PR is based on https://github.com/nomad/houston/pull/73, so it will generate some overhead while waiting for the read_socket.
When an error is encountered, the push method will be invoked again, with push notifications that have not yet been processed. (Everything after the index of the failed notification)
This is not an ideal solution, but it makes sure sending a batch of notifications will not fail halfway through, because of 1 illegal device token. This PR is based on https://github.com/nomad/houston/pull/73, so it will generate some overhead while waiting for the
read_socket
.When an error is encountered, the
push
method will be invoked again, with push notifications that have not yet been processed. (Everything after the index of the failed notification)