MVSFORUMS.com

mital_amit · Beginner Joined: 10 Sep 2004 Posts: 8 Topics: 2

Hi All,

Could somebody point me to what estimation techniques are followed for S390 Assembler?

Eg: For coding and Unit Testing a 1000 line Assembler program how many PD (Person Days ) of effort will be required. Is there a LOC to PD mapping which can be done for S/390 Assembler or is it something which is done by gutfeel!

Would appreciate if folks can share their experience.

Thnx,
Amit Mital

kolusu · Posted: Thu Sep 01, 2005 10:39 am Post subject:

Dibakar · Posted: Thu Sep 01, 2005 11:01 pm Post subject:

semigeezer · Posted: Fri Sep 02, 2005 1:46 am Post subject:

Aaaahh! If your company is using Lines of Code as a measure of ANYTHING quit now!!! It means they don't know what they are doing and that management has no idea what they are managing. Please tell me the name of the company so that I can be sure not to do any business with them -- I wouldn't trust such a company with my money.

In all seriousness (actually, that was serious), LOC is an utterly pointless, time wasting, energy sapping, misleading and diversionary waste of effort for exactly the reasons that were just pointed out by everyone else. A competent programmer will be able to tell you how much time it will take to code an assignment and how many errors he or she expects to find in testing or maybe after release. What you want to measure is how much resource needs to be applied to solve a given assignment (programmer months for example). LOC will not lead that. Experience will.

mital_amit · Beginner Joined: 10 Sep 2004 Posts: 8 Topics: 2

Guys,

Let us quantify what an experienced programmer looks at when estimating. There has to be a systematic way we can estimate. I know in other languages we have something like a FP/OB (Function Point/Object Point) to LOC to PD mapping are we saying that this does not work for Assembler ?

Any guideline tips, pointers or metrics will be very helpful.

dtf · Posted: Fri Sep 02, 2005 11:39 am Post subject:

If you have a detailed design specification, you should be able to apply some metrics to this to come up with a time estimate for coding. LOC would be pretty meaningless since I think it is pretty difficult to estimate LOC based upon a spec for an assembler program. Even if you could, this does not speak to the complexity of the program.

If the person doing the code has a good understanding of the requirement and the programming language, coding in assembler should not be much more involved than any other language (COBOL for example). This is especially true if a decent set of macros have been developed for use with the system in question. Of course as with any programming ?plagiarism? should be encouraged. Find a program which most closely matches the requirement, and use this as a base. If no such program exist, it would beneficial to have a ?skeleton? program to be used as a starting point.

So, I guess what I would do if you feel you have a decent estimating technique for COBOL coding, I would use the same metrics, and then apply some padding factor, say 1.20 or something. So 10 hours (for COBOL) becomes 12 hours for assembler. So much depends on the competence of the programmer in question, but this is true regardless of language. This is what makes it more difficult to estimate programming than say building a house, where the skill level of a framer is more likely to be interchangeable.
________
mflb vaporizer