roberts@studguppy.lanl.gov (Doug Roberts) (08/17/90)
I don't know whether to laugh or cry. We have a customer (who shall remain nameless) who has asked us to build a fairly large-scale simulation of their facility. The problem is that this customer just recently inherited the role of running the facility when they won the contract to do so. While many employees transferred, many left. Because of this, they aren't very familiar with the operational logic of their operation. The facility is largish (> 4000 employees), but the high employee turnover means that much experience has been lost. We have been told, in fact, that there simply isn't anyone left who can explain operations at a functional specification level of detail. To compound this problem, the customer has assigned a freshly-degreed (inexperienced) person as our primary point of contact. This person has no real-world software engineering experience, is completely unfamiliar with top-down design methodologies, object-oriented design, or software requirements documentation. Naturally, therefore, this person is attempting to impose a design methodolgy upon us that goes like this: (Customer speaking) "We don't know how our facility works, and so we can't supply you with requirements and functional specifications. Further, we don't know what data we will want the model to capture for us, so we want you [us] to design the model to capture 'everything'. And, since we don't know how to do anything else yet, we want to design the output data post-processor first." Now, lest you think I'm making all of this up, I'll include a paragraph from the customer's "requirements" document for the model's post processor: "With the current design, the analyst running the simulation is not required to foresee which information he/she will want to analyze. A detailed history of all material, processes, equipment, personnel, [...], and storage is kept during the simulation. The analyst, using the post processor, has the flexibility to manipulate and analyze any data from any of the history information kept from that simulation. If the analyst wants to produce a Master [...] Schedule (monthly totals of production), observe actual daily operations, as well as analyze bottlenecks, he/she has the flexibility and is only required to run one simulation. The post processor is intended to be user friendly as well as flexible, using pop-up menus and prompts to guide the user." Sigh. What we need are requirements and functional specifications documentation; what we are getting is a bottom up design specification. --Doug -- ================================================================ Douglas Roberts | Los Alamos National Laboratory |I can resist anything Box 1663, MS F-609 | except temptation. Los Alamos, New Mexico 87545 | ... (505)667-4569 |Oscar Wilde dzzr@lanl.gov | ================================================================
steve@hubcap.clemson.edu ("Steve" Stevenson) (08/20/90)
When I worked in industry, that was a fairly common problem. You will need to get someone to interface with the group and the liaison who can ask questions without getting a lot of religion in them. Real models are built ``middle out'' and don't come together easily. -- =============================================================================== Steve (really "D. E.") Stevenson steve@hubcap.clemson.edu Department of Computer Science, (803)656-5880.mabell Clemson University, Clemson, SC 29634-1906