Github user markap14 commented on the issue: https://github.com/apache/nifi/pull/2475 Thinking about this a little more, I think that the DISK resource makes a lot of sense to have but I think we should document as to when to use it - that being, it should be used if the Processor would use the disk in a way that may not be intuitive. For example, ConvertRecord perhaps does not need it, given that it reads the records once and writes them once, which is what would be expected for converting from one format to another. However, QueryRecord is a great example of where this annotation would make sense. This is because QueryRecord will read the data up to N number of times, where N is the number of SQL statements supplied. DetectMimeType is also an interesting example, because I would expect it to read through all of the FlowFile content, but in some cases it is able to read only a few bytes, I believe, to determine the content's mime type. Perhaps we should treat the NETWORK one the same way? Or potentially drop it? I don't know of any cases off the top of my head that would use the network in any unexpected way.
---