This release changed the behavior of the following functions:
As of 0.8.430, each aws-api client uses a single shared http-client by default. Before this release, each aws-client got its own instance of http-client by default, which caused the number of threads consumed to increase linearly in relation to the number of aws-clients created. To reduce resource consumption in the case of many aws-clients, we recommended that you create a single instance of the http-client and explicitly share it across all aws-clients. This is no longer necessary.
With the introduction of a shared http-client, this function was
updated so that it has no effect when using the shared http-client,
but will continue to call cognitect.aws.http/stop on any other
http-client instance.
These changes have the following effects:
Programs that were creating multiple aws-clients without supplying an http-client, and without ever calling stop, will see a reduction in resource consumption.
Programs that were creating an instance of
cognitect.aws.client.api/default-http-client and sharing it across
aws-clients should see no change. You can, however, safely stop doing
that.
For programs that were using the default aws-client constructor and calling stop on each aws-client, the shared http-client will not be shut down. This should have no negative impact on resource consumption, as there is only one http-client in this case, and its resources are managed by aws-api.
For programs that were creating multiple aws-clients in order to get around an "Ops limit reached" error, this is a breaking change. For this case, we recommend, for now, that you supply a new http-client for each aws-client.