Sequence question

Question

I'm curious what others are doing with sequences in their AppEx applications. I know this isn't strictly an AppEx question - but I recently ran into serious problems because of my choice of how I'm using sequences.
The problem goes something like this: I create a new table and while I'm doing so I go ahead and create a new sequence for that table. Seems innocent enough. Now, I just happen to use this new table as a lookup table for some foreign key values in one of my main data tables. Not a problem. Works like a charm.
I start populating data into the main table and the lookup tables. Again, not a problem. The users want a little query mecanism for the report so I decide that I'm going to implement it in the UI with a checkbox based on an LOV that allows users to find records based on checking values and clicking a 'Find' button. OK, create a checkbox item, a named LOV, assign it to the checkbox item. Now amend the query for the report region to use the checkbox via an instr(:P1_MY_SEARCH_CHKBX, column_name)>0 kind of approach.
Ooops! I'm not getting what I want. The changes to the query are clearly working for some cases but I'm seeing John Doe's projects (like I want) and Mary Jane's too (which I don't). Turns out John Doe's ID is '1' and Mary Jane's is '11' and '1' is definitely in '11', so things are working as they should based on my implementation, but I'm getting 'unexpected results' from the users point of view.
Obviously, the root cause of this comes down to my choice to create that new sequence for each table and increasing the likelihood that this situation would occur. If I had used a single sequence value for the entire schema, that started with a large value (maybe a guid) and built all of my triggers off that one sequence then this little UI anomaly would likely not have occured.
Sorry for the long, boring story - but I'm curious what others are doing and how many people have run into this same kind of problem?
Earl

Answer 1

Personally I would do it in this sort of way. It may add an extra page or two, but I think in the long run it becomes easier.
Here is my sample app - click a couple, hit the search button, and it will show you your results.
http://htmldb.oracle.com/pls/otn/f?p=20233:1:17770724214406136839::NO:::
What I would do in your case is to do something similar - list out the project managers using htmldb_item.checkbox(1, manager_id). This will tie the exact id to that box. Create a before header process that creates a collection - this collection will hold your checked boxes. Add an after submit process to add items to that collection. On your next page, you would do something like this to restrict the rows:
select pro.project_number, pro.description
  from project pro
where exists (select 1
               from htmldb_collections col
               where pro.manager_id = col.c001 and
         col.collection_name = 'PROJECT_MANAGER')Personally I like collections a lot - that way someone can go back to that page, add a couple extra checks and it repopulates the collection. If you need to see the exact code, let me know and i'll post it on my application page.
As far as your 2nd question, its a matter of preference. There is nothing wrong or right about creating one sequence - yes you're right - the user never sees it, so it really doesnt matter. But I guess I am a creature of habit, and dont see what creating one sequence really saves you. Its sometimes nice to be able to see the next sequence value and have an approximate guess that you are creating the 243rd new employee instead of the 125,335th new object in the entire database.
To each their own!