Merge pull request #57 from huntc/key-state

Introduces key state for compaction

Author: huntc

Cargo.toml

@@ -53,7 +53,7 @@ serde = "1.0.151"
 serde_json = "1.0.90"
 serde_with = "2.0.1"
 sha2 = "0.10.6"
-smol_str = "0.2.0"
+smol_str = "0.3.2"
 test-log = "0.2.11"
 tokio = "1.23.0"
 tokio-stream = "0.1.11"

README.md

@@ -10,6 +10,9 @@ A commit log modelled on [Apache Kafka](https://kafka.apache.org/) is provided,
 
 A secret store modelled on [Hashicorp Vault](https://www.vaultproject.io/) is provided along with partial support for the Vault HTTP API.
 
+File-based implementations of the commit log and secret store suitable for use on embedded Linux are provided with 
+[logged](streambed-logged/README.md) and [confidant](streambed-confidant/README.md) respectively.
+
 Production services using the commit log and secret store have been shown to use less than 3MiB of resident memory
 while also offering good performance.
 

streambed-logged/README.md

@@ -137,7 +137,7 @@ thousands of records.
 
 The compaction of topics is an application concern and as it is able
 to consider the contents of a record. Logged provides functions to atomically "split" 
-an existing topic log at its head and yield a new topic to compact to. The 
+an existing topic log at its head and yield a new topic to compact to. An 
 application-based compactor can then consume the existing topic, retain the records 
 it needs in memory. Having consumed the topic, the application-based compactor will 
 append the retained records to the topic to compact to. Once appended, logged can be