JIT compilation time: CSV tools
The major issue affecting the utility of the Parabix framework and Parabix applications generally is the high cost of JIT compilation.
Consider csv2json. A distinct pipeline is generated for each different "type" of CSV input file. At present, the type of the CSV file is essentially determined by the set of column names. Future developments of the CSV tools are likely to introduce additional attributes to the CSV file "type", such as the type of delimiter (e.g., semicolon or tab instead of comma), data type constraints for each field (e.g., a regular expression constraint for each field), potential options for white space trimming of selected fields, and JSON data type interpretation of each field (generate JSON bool, null, numeric values, rather than strings).
Although caching helps for the second and subsequent run involving any particular CSV file "type", most potential users of these tools will find the high JIT compilation time of the first run unacceptable.