[rules-users] Need help in formulating rule (using list)

[Wolfgang Laun]

Expecting just what you wrote, I *did not* suggest changing TimeEntry.

Notice that I suggested using extended copies of TimeEntry, i.e., TimeEntryPlus,
just for reasoning. They could be removed after doing these checks, or
continue to exist in parallel with the Schedule fact.

(Although, the additions for TimeEntryPlus could be easily achieved by
subclassing, which is always cited as one of the major benefits of OO; it
would indeed not require any changes in your app.)

-W

2009/11/11 Brody Bach <brodybach at yahoo.com>:
> Hi Wolfgang,
>
> thanks for your reply.
> I did have the idea about having the concepts of first, last,predecessor,
> successor etc. in TimeEntry. Unfortunately, it really isn't possible because
> this class (i.e TimeEntry) is a main part of a huge application model, which
> has been existing since long, has many dependencies to other classes, and
> shouldn't be changed ("only" for fulfilling the requirement of rules-engine)
> . So unfortunately I can only use those which are available to me. That's
> why I was hoping if there's really pure rule-based solution to this.
>
> Regards
> ________________________________
> From: Wolfgang Laun <wolfgang.laun at gmail.com>
> To: Rules Users List <rules-users at lists.jboss.org>
> Sent: Wed, November 11, 2009 11:29:29 AM
> Subject: Re: [rules-users] Need help in formulating rule (using list)
>
> Implied by the List representation are concepts such as "first",
> "last", "predecessor"
> and "successor". All of these may or may not be represented by the fields in
> TimeEntry. This would be so if it is *guaranteed* that one TimeEntry's
> start time is another one's end time; (and then you wouldn't even need them
> being sorted). Failing this, an explicit representation by adding an ordinal
> number or (as I prefer doing) a reference to the previous and next TimeEntry
> might be considered.
>
> But, either way, reasoning over pairs of TimeEntry objects (of one
> Schedule) requires that they are all (at least temporarily) present as facts
> in
> WM. Then, to be identified as TimeEntrys of a specific Schedule, they also
> could have a reference to that.
>
> It's a simple task to write a rule that fires for a new Schedule and creates
> the extended TimeEntryPlus facts from its List<TimeEntry>. To avoid NPEs,
> I'd add a dummy TimeEntryPlus object (type=="OFF") before and after. Then,
> you
> could write rules
>
> rule checkValidBreak1
> when
>     $break : TimeEntryPlus( type == "BREAK" )
>     TimeEntryPlus( type == "WORK", succ == $break || pred == $break,
> length <= 10 )
> then
>   System.out.println( "Short work b/a break " + ... $break ...,
> $break.getSchedule() ... )
> end
>
> rule checkValidBreak2
> when
>     $break : TimeEntryPlus( type == "BREAK" )
>     TimeEntryPlus( type == "WORK", succ == $break )
> then
>   System.out.println( "Can't start with a break " + ... $break ...,
> $break.getSchedule() ... )
> end
>
> -W
>
>
> On Wed, Nov 11, 2009 at 10:46 AM, psentosa <psentosa at yahoo.com> wrote:
>>
>> Hi Brody,
>>
>> I have had similar situation several times, and each time I had to modify
>> my
>> objects to avoid this, and it was really tedious.
>> I'd also would like to know whether it is possible to have a "real
>> solution"
>> in this case without needing to change the base objects, but rather use
>> the
>> rule language to formulate this kind of business rule
>>
>> *why didn't I receive this mail through the mailing list? maybe you
>> haven't
>> subscribed yet?*
>>
>>
>>
>>
>> Brody bach wrote:
>>>
>>> Hi all,
>>>
>>> I have an object which I use as main fact object, let's say "Schedule",
>>> which has a list of objects called "TimeEntry"  --->
>>>
>>> class Schedule(){
>>>      List<TimeEntry> entries;
>>> }
>>>
>>> class TimeEntry(){
>>>      String type;
>>>      Date start;
>>>      Date end;
>>>      Integer length;
>>> }
>>>
>>>
>>> Now, if I insert Schedule into the WM, and would like to iterate over the
>>> list of TimeEntry, due to the following business rule :
>>>
>>> For all time entries, if it is of type BREAK:
>>> check whether a time entry of type WORK exist,
>>> before and/or after this BREAK, which has only length of 10 minutes
>>>
>>> The content of the time entry list has been sorted before so they are all
>>> in correct time order, e.g
>>>
>>> 09:00 - 09:10  WORK (10 min)
>>> 09:10 - 09:30  BREAK (20 min)
>>> 09:30 - 10:30  WORK  (60 min)
>>> --> in this case, the rule should fire
>>>
>>> How can I define this rule? Some questions which I have:
>>> - how do I check previous and next element of the current element?
>>> - the length of the list isn't fix; it may contain 1, 2, or more entries.
>>> So somehow in an attempt to access next element, it should take care of
>>> the bound of the list. But should this be really part of the rule
>>> definition?
>>>
>>> Many thanks in advance!
>>>
>>> Regards
>>>
>>>
>>>
>>>
>>
>> --
>> View this message in context:
>> http://old.nabble.com/Need-help-in-formulating-rule-%28using-list%29-tp26284362p26298480.html
>> Sent from the drools - user mailing list archive at Nabble.com.
>>
>> _______________________________________________
>> rules-users mailing list
>> rules-users at lists.jboss.org
>> https://lists.jboss.org/mailman/listinfo/rules-users
>>
>
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>
>
> _______________________________________________
> rules-users mailing list
> rules-users at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-users
>
>