I have a REST methodology question - if not appropriate for this forum,
please forgive and redirect me if you would.
(But I would love to hear the REST pros weigh-in!)
We are designing a financial forecasting Web Service using Jersey.
Clients would request a "CompanyScenario" resource, essentially a
container for about 50,000+ data points per company, calculated on the
server. (Example data points would be ROI at year 2007 and EBITDA at
year 2010.)
So for example, a client could call for
"/resources/companies/IBM/scenarios/default" which would return the
default scenario for IBM.
Most applications would not need all 50,000 data points for a scenario,
and those would take up over 5MB which would be overkill to return on
each call.
In fact, most applications would only need about 1000 data points from a
scenario, though we would never know which in advance.
So the problem is how should the client tell the server exactly which
data points it needs?
We were thinking of having the client pass in the names of desired data
points as query string parameters, but that URI could easily exceed 4KB,
which would exceed the limits on URI size.
On the other hand, using HTTP POST violates the contract of GET gets
stuff and POST creates stuff.
The O'Reilly RESTFul Web Services book touches on this briefly, and
suggests appending .1, .2, .3 to represent specific variations of a
resource. But this doesn't really work for us, because a client needs to
be able to request ad hoc sets of variables, many at a time.
I am wondering if there is a best practice suggestion for this common
use-case?
Much thanks, Victor
==============================================================================
Please access the attached hyperlink for an important electronic communications disclaimer:
http://www.credit-suisse.com/legal/en/disclaimer_email_ib.html
==============================================================================