Decision Table Patterns (2020) | Hacker News Reader

Decision Table Patterns (2020)

hillelwayne.com

73 pointsrramadass11 comments

Once you have a decision table, representing the logic conditions can be simplified by Karnaugh maps: https://en.m.wikipedia.org/wiki/Karnaugh_map

There are more expressive versions which are less formal but have the ability for more than one row to fire, which is then resolved with a "hit policy". Used in finance and process modelling

https://docs.camunda.io/docs/components/best-practices/model...

Nice! I didn't know of DMN (https://camunda.com/dmn/). Are there any detailed literature (articles/books/etc. on this and related decision techniques) that you can point us to ?

PS: I had posted another related article which you might find interesting - https://news.ycombinator.com/item?id=38822320

BPMN and DMN are mature and standardized technologies for enterprises. Support from IBM, SAP, Redhat etc. but Camunda is where I learnt it.

https://www.ibm.com/docs/en/odm/8.10?topic=center-when-use-d...

https://training.sap.com/course/sig141-dmn-12-basics-classro...

https://access.redhat.com/documentation/en-us/red_hat_decisi...

That's the decision engine part. For domain modeling (i.e. expressing logic) you have to search for fluff https://www.trisotech.com/dmn-hit-policy-explained/

I get a feeling that perhaps DMN is taught to business .ajora and never to CS majors, which is why it is this big thing no-one on HN knows about

https://mit-online.getsmarter.com/presentations/lp/mit-busin...

Truth tables were when the rubber really hit the road for me in understanding computers. Up until then I had understood transistor logic levels and the fundamentals of programming, but the two seemed miles apart.

Related: Decision Table based Methodology for Software Development - https://news.ycombinator.com/item?id=38821708

> What we can do is replace the number N with the equation N % 2. That has only a finite number of inputs and leads to a valid table.

Just so I'm understanding it correctly, it should say "finite number of OUTPUTS" there, right?

I think they mean "finite number of rows", which would be "finite number of distinct inputs" or "finite number of outputs". The context is the challenge of writing down an infinite table.

There's probably a recursive/inductive equivalent notation but maybe it's hard to render clearly as a table.

I just re-read the entire reasoning, and I got it all wrong I think. N%2 will only ever yield one of two inputs. In short, I shouldn't try to brain before the morning coffee ;D

There are of course ways to compress the information in these tables in many practical cases, e.g. using binary decision diagrams.

Any links/resources that you can share?

Reply on news.ycombinator.com