constraints - preferred one, preferred set of

[liquid]

Generally, constraints have the two forms:
item has preferred other item
item has preferred set of other items

I experience that during data preparation for a timetable the things are often changing or are caused by planning progress in the way it's necessary to add to a constraint of first form some other item, making the constraint of the second form. In this case, I must delete the former constraint and create a new one, to respect a set of other items. It would be simpler to just add other item to previously set up one, making therefore a set of preferred other items automatically. I'm not sure if the set of preferred other items is equivalent to many constraints of the first form (with one other item). If it were I think it would be possible to abandon this distinction in a future "big" update of FET program.
I wonder whether the reason for this rather technical (a matter of development) and difficult to change or may be not so difficult.

[Liviu Lalescu]

Are you referring to preferred times?

An -X- means rejected with the corresponding weight. If there are more constraints, the maximum weight wins. This constraint applies as rejecting a (some) slot(s) (rather than accepting some).

Preferred starting time is a locking constraint, you can play with it in the generated timetable.

I'll add this in the TODO, along with my answer.

[liquid]

Thanks for your answer. I'll try to be more specific below.
I used to work with schedules incrementally. This means for example that I don't specify at the beginning all existing rooms suitable for a subject as a set of preferred rooms. Only one at the beginning and then by the next group I'd like to add one or more rooms for the subject. I have two ways:

• Add the second constraint that assigns a subject for the new room
• Remove previously set constraint "one room for one subject" and introduce a new one "set of rooms for one subject"

That's what I meant starting this thread. I'm sorry if the first post was misunderstood. I guess it would be more convenient to simply add the second room to previously set constraint rather than use of two described ways above. It would be, however, possible if I could add a constraint of form "set of rooms for one subject" with only one room at the beginning, predicting that in the future I'll add some more rooms to that. FET disables this though.

[Liviu Lalescu]

Variant 1 is wrong. You should use 2. Yes, I'll add this in the TODO.

[Liviu Lalescu]

I did it, thank you for your suggestion! Please see http://lalescu.ro/liviu/fet/forum/index.php?topic=1817.0 and let me know your test results.