Quote:
Originally posted by Dravyk@May 25 2004, 01:17 AM
Wish someone would come up with a proven formula of some kind to make things more accurate in forecasting, to benefit both the programmer and the client. It seems to be a bit of seat of the pants kind of profession that's hard to do it any other way.
|
There is one. Any of us that went to programming/analysis school would tell you that it works great in an academic environment where you have all sorts of time to play with and your "client" is able to give you their undivided attention for hours on end. Unfortunately, the methodology gets fastttracked in the real world.
Rox described it fairly well there. The process of getting a good definitions document can be a long hard process. To get to a good determination of the actual amount of time and resources required you have to go through a certain amount of your actual design process. To give an accurate estimate, you have to disect the project to what amounts to the "module" level.
Now, the rub here, from the programming perspective, is that if the client, after you go through this process, decides not to buy, then you have to eat that time somehow. If the client does buy, you try to include that time in the actual quote in order to recoup it.
The rub, from the client point of view, is that it takes an inordinate amount of their time for me to be able to provide them with a realistic estimate.
So, in an effort to not have to eat time and an effort to satisfy the actual client needs, the process becomes shortcircuited.
The process that we, in programming school
learned is basically:
Design: (this is anywhere from an 8 hour to multi month timeframe)
Code: (If the time spent in design is used properly, the coding period can be fairly rapid. Withouty a detailed design, this period can draw out and draw out and draw out)
Test and Debug: (Depending on the quality of the time spent in design and coding, this can be a short time frame or an extended period)
This results in any project being able to be quoted as "at least a year". 3 months design time, 3 months coding time, 3 months testing and debug
The actual process that ends up resulting in many cases is:
Design, code and test as we go and hope that we get it right the first time.
