The first part of my project using decision tables was a lot of fun -- especially writing rudelo.
The second part was not much fun at all. Debugging a series of decision tables was much more difficult than I anticipated (the reason I used a series of tables was so I could use a few tables with several rows each rather than one table with hundreds of rows).
I realized sadly at the end there was little chance of my tool being re-used by non-technical people, or even by technical people for that matter.
What would have changed things greatly is instrumentation -- for rufus-decision to return a structure which showed whether each cell matched in each row (and possibly also which matcher), and the intermediate output each row added (important when using accumulate)
In an app, this structure could then be replayed in a jquery grid view to allow visualization of a dataset through a series of decision tables.
This is something I'd like to work on. I'll start in a fork per usual.
Do you have any thoughts on this? Is it something that would be useful for you as well? I welcome any feedback on ensuring it adds minimal overhead when not being used.
Proposal: Instrumentation of the Matching Process
The first part of my project using decision tables was a lot of fun -- especially writing rudelo.
The second part was not much fun at all. Debugging a series of decision tables was much more difficult than I anticipated (the reason I used a series of tables was so I could use a few tables with several rows each rather than one table with hundreds of rows).
I realized sadly at the end there was little chance of my tool being re-used by non-technical people, or even by technical people for that matter.
What would have changed things greatly is instrumentation -- for rufus-decision to return a structure which showed whether each cell matched in each row (and possibly also which matcher), and the intermediate output each row added (important when using accumulate)
In an app, this structure could then be replayed in a jquery grid view to allow visualization of a dataset through a series of decision tables.
This is something I'd like to work on. I'll start in a fork per usual.
Do you have any thoughts on this? Is it something that would be useful for you as well? I welcome any feedback on ensuring it adds minimal overhead when not being used.