Hi,
I'm wondering about the usage of InputFormats repository object.
What kind of functionality does the InputFormats provide? Just object caching?
Since jobtracks is may be running in a different JVM as the JobClient this kind of cache makes no sense from my point of view.
If I would love to have a own InputFormat in my Job I need to hard code it in the InputFormats and submit the new InputFormats object within my job jar.
Isn't it or do I miss something in the puzzle?
What do you think about using the InputFormat class name as key and in case it is not available in the InputFormats cache we simply load the object via dynamic class loading from the job jar?
Makes that sense?
If you like I can create a patch for illustration.
Stefan
