I’m confused by the output of the checkpoint tool. The documentation explains that it outputs a file in properties file format, but when I run it, I get log4j output, including lots of diagnostic details. And even the final line that includes the checkpoint property has logging timestamp/classname prefixes. And then when I try to edit and re-import the file, the parser seems to give up half way when it reads the ProducerConfig values from the previous output. Is this what everyone else sees? I would expect that it would simply output one line per topic and partition. While I’m new to scala, when I look at the source, it appears to use logging both for diagnostics and for the checkpoint output. Is this by design? If so, perhaps we could suppress all of the diagnostic logging and only output with a flag? I see that if log4j.configuration isn’t specified via JAVA_OPTS environment variable, then the default log4j-console.xml is used which sends all INFO to stdout. Perhaps we can give CheckpointTool its own log4j config file so only the CheckpointTool class output is sent to stdout?
I’m using scala 0.9.1 -Jeff
