Another Scheduling Problem

[stanley]

I read that other stuff about the chiropractor (or whatever) and have a similar situation. I'm automating a silk-screening shop, where they've got 8 presses. We've got a db for production orders which holds two items relative to the printing process - number of items to be printed and the type of item (so, say, 1200 t-shirts or 500 pillowcases). I've got another db for the presses themselves, the key info for which is the machine's capacity.

What I now need to do is make a system where the manager of the shop can schedule his jobs. The problem I'm having is how to make sure that he can't overschedule a machine. Until I solve that one, I can't really wrap my head around the thing at all.

Also (and this is where it gets really wierd) I'm not dealing with static time slots (i.e. hours or whatever, as in the chiropractor example) but rather with total capacity of a machine - each machine being different. But that's another problem.

[LiveOak]

Unfortunately the solution is the same. You must pick a time slot, even if it is one second and you have to use 86,400 of them for the day (actually you can't do this, the text field limitatin is 64K characters!). Using relationships I don't know of a way to perform other than an exact match. If you need to, you can use 1,440 time slots of one minute each for the day. A block of scheduled time for a machine would be represented by a list of the minutes it uses on a given day, for example:

A machine used on December 1st from 1200 hrs to 1208 hours would have the entrys:

120101-0721

120101-0722

120101-0723

120101-0724

120101-0725

120101-0726

120101-0727

120101-0728

You can leave out the "-" and use a straight number for date (days from 1/1/0001) or use days from some arbitrary date (1/1/2000) to save characters. Using this method you can compare schedules using relationship to find conflicts.

-bd

[ December 04, 2001: Message edited by: LiveOak ]

[Vaughan]

Forget FMP, set something up on paper first. Once you have that system up and running you've done most of the data design work to convert it into a database.