That's fair, and Mark's comment about the drools manual is encouraging as
well. Maybe I just haven't found really illustrative reading on declarative
programming yet. I've looked into it from time to time, but never found
enough guidance to figure out how it applies to real-world use cases.
I feel like a cookbook would be very useful though. Some common patterns
in procedural programming, why they are good candidates for accomplishing
with declarative programming, how they translate into rules, patterns, anti-patterns...practical
stuff. Seems like much of the lit on declarative programming is very academic...hard
to boil that down to tools I can use in my work. The anti-patterns would
be a big help actually, both high-level "this problem isn't suited to rules"
and low-level "don't write rules like this" stuff.
I'm really surprised there isn't more of this out there, or at least that
I haven't found it yet.
Hello Dave,
Before you try to tackle a rules engine, you should definitely have a
good understanding of declarative programming languages. Declarative
programming is very different from functional programming a la Java,
C++, C#, Visual Basic, Fortran, Cobol, etc. To build your
fundamentals in declarative programming, you should attempt to master
a language like Prolog or Lisp. These languages were the inspiration
for many of the rules engines that we have today. These were the
tools that Charles Forgy and others experimented with before they
developed the technologies that we leverage today in the rules
community. I would be hesitant to put a warning label on a Rules
Engine as we would on cigarettes. However, let me say this:
"WARNING: The surgeon general cautions users of rules engines that
dabbling with rules engines without a core understanding of
declarative programming may be hazardous to your sanity."
Best Regards,
Dave
-----Original Message-----
From: news [mailto:[EMAIL PROTECTED] On Behalf Of Russ Egan
Sent: Tuesday, April 11, 2006 8:25 AM
To: [email protected]
Subject: Re: [drools-user] How to express conditional logic via DRL -
?
Speaking of which, I've tried to find good, in-depth guides for rules
engine usage. Best practices, rules writing guidelines, cookbooks,
advanced use cases, patterns, etc... All I've found are very shallow
"introductions"
to rules engines. Does anyone know of good, educational materials of
this nature on the web? Or good books?
It's tough to avoid using rules engines "the wrong way" when this is
little guidance for using them "right way", and when rules engines
are so conceptually different from procedural programming.
Hello Mark,
Unfortunately I'm finding that many people just cannot adjust to the
declarative and coupletely decoupled nature of a rule engine. The end
result is that Drools is used as a script execution framework.
Felipe Piccolini wrote:
Mark,
I could vote on this, I would vote no to the 'else'. The declarative
thing about rule engines is that: no procedural code. So if you wanna
do an 'else' condition you should re-think your rule and chack if its
well writen. Of course this is just MHO.
:)
Tuesday, April 11, 2006, 6:09:22 AM, you wrote:
I thought about adding 'else' but I couldn't decide on the best way
to implement it, as there are several possabilities. Also 'else' is
not considered declarative, so its a kinda of code smell. I'll llook
into this again in 3.1
Mark
Geoffrey Wiseman wrote:
On 4/10/06, Dmitry Goldenberg <[EMAIL PROTECTED]>
wrote:
What is the timeframe for 3.1? Else/else if are very important to
what we're trying to do.
I can see how I can to the following with the method described by
Geoffrey:
if (A) then B else if (C) then D endif
I also want to be able to do the following:
if (A) then B else D endif
I imagine that the rule would have to be written as a combo of
if (A) then B endif
if (!A) then D endif
so that the conditions are complimentary...
Indeed, yes, that's how you'd handle it.
If it does make it into 3.1, that's a good thing in terms of
supporting the way people think, talk, and work in other languages,
but fundamentally, the same capabilities are there now -- they just
require more typing.
- Geoffrey
--
Geoffrey Wiseman
--------------------------
Felipe Piccolini
[EMAIL PROTECTED]
