View or Package, logic problem which is better

Question

I'm new to Oracle so I'm trying desperately to take what I know
from other systems and apply it here and I'm hitting a stumbling
block. I have a very complex view that I will spare you the
details to, but what it amounts to is:
create or replace view testview
as
select * from table_x where year_column=2001 ;
this is fine, but what do I do when I want to change the year to
something else? I don't want to have a view for every year,
that's a maintenance nightmare. I'm in the beginning stages
with packages, so I understand how they work and mostly how to
build one, and I'm halfway through building a procedure to
create the view, but in that case, I wonder what will happen
when the procedure is run a second time, will the person who ran
it the first time suddenly get the new view, will it invalidate
itself, or will it just crash? I am stuck with one schema with
a "global" type login, so it's not like each person can have
their own. I have thought of cursors, but I don't understand
how to use an explicit cursor as part of another SQL/cursor/view
because this view is only a small part of other views I am
building. I need to figure out how to pass the parameter from
an external system (non-Oracle) and handle most of the code in a
procedure/package or if I make this a view of all the available
data (with no year limit) and do the parameters on the other
side in the non-Oracle system it really slows things down to an
unacceptable level and doesn't teach me how to do anything new
or take advantage of the Oracle backend.
Any help will be deeply appreciated. thanks!

Answer 1

Hey!
thanks for the input. It's exactly what I'm looking for as a
newbie. To give you some background, the database isn't that
unusual, it's a government financial app (non Oracle but Oracle
db) with about a million tables which I'm connecting to through
a database link which I have in my own database space because we
(devs) aren't allowed to create packages/procs in their space.
Which makes sense from a security standpoint, but it requires
some creative thinking for other issues. I'm doing reports,
which means creating a lot of "denormalized" views and when we
start talking about an entire year's worth of data at the
transaction level, it slows way down. Luckily a lot of the
stuff I'm doing is actually summaries for the year, etc. My
front end at the moment is Access/Visual FoxPro, but I'm hoping
to move to more web based reporting (tool TBD later - likely ASP
or Java, any recommendations?) as I stuff as much as humanly
possible into the PL/SQL to reduce the network traffic. Access
handles views really decently/transparently, but there's some
ugliness with ref cursors. One of my cohorts has found an
example on the web somewhere and while passing the parameter in
isn't a big deal, it's an ugly bunch of VBA code just to process
the returned cursor, something about it having to load it row by
row. I've only glanced at it and need to look at it in more
detail. I was just hoping I'd missed something more
obvious/simpler through inexperience. I do admit Oracle is
capable of doing some of the most complicated SQL I've ever seen
and it's kinda fun to see exactly how many
joins/tables/subqueries I can throw into a single SQL. I'm just
not used to looking at it that way; I'm more used to writing
several sequential SQL statements that can build on or refer to
each other by a temporary name/alias. I either haven't figured
out the syntax/commands to map what I know how to do to the way
Oracle does it, or maybe it's just never done that way? Hard to
say. But then again I live to learn through trial by fire. :)
thanks much!!