Dealing With Generated Lines of Code

February 12, 2009 · Filed Under Software Estimating, Software Sizing  - 0 Comment(s)

We have gotten several questions recently on how to estimate using lines of code when the lines are generated, that is there is a preprocessor that allows specification of the solution and the code is automatically generated to implement the solution.

 Define Functionally: The ideal way to deal with generated code is to use function based sizing of some sort.  SEER offers everything from use cases, to IFPUG function points, to COSMIC function points, to SEER’s function based sizing which can handle both IT processing and highly algorithmic processing sizing.  Saying I will deliver 5 reports, 2 databases, and 7 algorithms, for example will generate both the cost, schedule, risk and an approximation of how many lines would be produced as found on the “software metrics report”

 If your process demands lines of code the question becomes “how do I enter the lines into SEER”

 Entering Adjusted  Generated Lines: Entering lines generated is somewhat like the old practice of estimating by the number of computer words or machine cycles.  It can certainly work.  But the generated lines do not capture the work performed to obtain them as well as hand generated lines do.  If you have your own ratios, use them. If you do not, SEER can approximate them for you.

  • 1. Set the Resulting programming language
  • 2. Enter the generated lines in the lines of code fields (For this example assume they are all new)
  • 3. Change the language to “generated code” or if the particular generation tool is listed choose that. (As an example if the target language is “C” it converts to a .29 factor
  • 4. SEER-SEM will provide a conversion factor for you and will automatically convert the code if you tell it to.

See the BLOG on Counting XML Lines for information regarding XML from a code generator

Entering the Total Generated Lines:  Build a knowledge base… Enter the anticipated lines output by the code generator and use the SEER-SEM calibration mode to determine appropriate calibration factors.  That information can reside in a knowledge base and can calibrate the effort, schedule, etc.  Additionally you should select the out-of-the-box “code generator” knowledge base to appropriately adjust phases, testing, etc. for the generated code environment.


Thank you for reading “Dan on Estimating”, if you would like more information about Galorath’s estimation models, please visit our contact page or call us at +1 310 414-3222.

Related posts:

  1. Lines of Code Versus Functon Points Versus Use Cases For Sizing
  2. Counting XML Source Lines of Code
  3. Software Code Counter Review and Recommendations
  4. New Code Counting Tool Made Available
  5. New Code Counter Update Available from USC

Comments

Leave a Reply