esperienze, curiosità, immagini dal mondo dell'informatica

Metrics and Function Point in Italian

Function points measure the size of an application's functionality, defined as a cohesive collection of procedures and data supporting a business objective. Each application is separated from the others and from the user since it is marked by a boundary. The boundary acts as a "membrane" through which the data processed by the Input, Output, and Inquiry transactions pass. The boundary should be seen from the business's point of view, and is not based on technical or physical considerations. Functional size is a standard conventional measurement regardless of the technology used to develop the software.
Functional size is one of the main variables of the software's price, which is also affected by the complexity and quality of the product, of the technology used for development, of reuse that is made, etc. Therefore, by taking specific productivities into account, function points are an aid for estimating, a priori, the effort needed to develop a project and for calculating, a posteriori, the value of the product and of the software.

The concepts related to functional size measurement (FSM) are defined and the principles for applying a FSM method are described in the ISO/IEC 14143-1 standard of 1997 entitled "Information Technology - Software measurement - Functional size measurement - Definition of concepts."

A FSM method must have the following features:

• it is based on a user requirements representation seen from the user's point of view;

• it may be applied immediately, as soon as the user's functional requirements have been defined and are available;

•  it derives functional size from the functional requirements, regardless of the technical requirements or the quality.

Functional size is also independent of the effort needed for development or maintenance, from the methods employed, from the physical supports used, and from the technological components. A FSM method should indicate the degree of convertibility with other sizing methods.
Other parts of the standard are being produced with regard to: compliance, checks, reference model, software domains.

Function points take concrete form in a set of scores assigned in accordance with the counting rules to Inputs (EI), Inquiries (EQ), Outputs (EO), and Logical files (ILF and EIF) evident from examining the application and its documentation.

In the world of Information Technology, function point metrics are gradually replacing LOC (Lines Of Code, formerly called "cards", derived from punched cards, read by a special hardware). Although the functions under consideration are both equal, LOC vary depending on the type of language used and are harder to quantify.

Although the method was defined by Allan Albrecht at IBM in the late 1970s, interest in function points began in the United States in the 1980s and made inroads into Italy in the 1990s. The most-used LOC measurement is the number of lines minus the comments, but an international standard has never been achieved. Function points, on the other hand, thanks to associations like IFPUG (International Function Point User Group) and ISO (International Organization for Standardization), are a standardized size measurement. GUFPI (Gruppo Utenti Function Point Italia) is Italy's reference for function point users. IFPUG 4.1 Unadjusted Function Point Method was approved by the ISO/IEC Joint Technical Committee 1 (JTC1) and became PAS (Publicly Available Specification) in 2001. The International Standards issued by JTC1, after a series of approvals, are considered the most authoritative standards in the field of Information Technology.

Capers Jones defined the average correspondence between LOC and FPs as "backfire". For example, according to Jones or David Consulting Group, 1 FP is equal to:
320 Assembler LOC or 575
128 C or 225
107 Cobol or 220
91 Cobol II or 175
55 C++ or 80
35 Java and Visual Basic-4 or 80
15 HTML-3.

It is evident that there is a difference between different sources. (For B. Boehm 1 FP is equal to 53 LOC C++ and Java). For publishing-type web sites developed with a large amount of text written within HTML instructions, we may approach the average correspondence of 15 LOC for 1 FP, without considering the LOC of the texts. An appropriate benchmarking should be used to check the "backfire" values in their own environments.

"Forward" function point counting (and not in "reverse" using the "backfire" technique), assumes the knowledge of precise rules, and is the object of new professions (CFPS - Certified Function Point Specialists). One can become certified as a counting specialist through GUFPI and IFPUG.

In truth, Function Points are more than a counting technique, since they help to thoroughly examine the functionalities, thereby improving analysis of - and making it possible to quantify - requirements. They produce better estimates, support acceptance testing, and improve the general documentation of the software. Function points also provide an excellent opportunity for perform a variety of censuses about the applications existing in the information system.

They may therefore be used to support the development process, and make it possible to manage some important level 2 key activities of the CMM (Capability Maturity Model) published in 1991 by SEI (Software Engineering Institute, Carnegie Mellon University), which became in 2000 CMMI (Capability Maturity Model Integration), in a highly organized manner and integrated with certain ISO international standards and industry documents.The model has 5 levels of process maturity:

1. Initial (performed informally)

2. Repeatable (managed in terms of requirements, planning, control, measurement, quality assurance, data, and configuration)

3. Defined (with full consideration of the user requirements, integrated, organized, and managed taking the risks into account)

4. Managed (quantitatively managed in all the aspects of quality and process)

5. Optimized (with cause-and-effect analyses and problem resolution, highly innovative)

The counting method applied with the required criteria has the following characteristics:
- repeatability;
- being sensitive to the magnitude perceived by the user by means of elementary process and complexity;
- comparability with other methods;
- accuracy;
- learnability (with about 1 month of theory and 2 of practise);
- usability (it takes from a few hours to 2-3 days to size an average application);
- documentability;
- sharing with the user.

In very simple terms, the technique may be said to provide a quantification of the information, from a logical point of view, that enters, leaves, and is stored in a computer by running a software application. By using the terms "internal" and "external" we assume that a limit has been determined - a boundary distinguishing the application to be quantified from other applications of the system. The elements being counted relate to one another. In brief, the files inside or outside the application (ILFs and EIFs) are referenced differently from the elementary activities of Input (EI), Inquiry (EQ), and Output (EO); the following are their primary purposes (the non-primary purposes are in parentheses):

An example of Input (EI) is provided by data acquisition, while an Inquiry (EQ) is provided by simple answer to a question, and an example of Output (EO) is achieved by printing out the calculated data; these activities make up the elementary processes - the smallest units of action meaningful to the user.

Some verbs that indicate the elementary processes for an EI and generally connected with one or more ILFs are: acquire, enter, add, send (by the user), set up, import, populate, assign, plan, schedule, delete, eliminate, modify, update, alter, accept, vary, review, check, validate, store, maintain.

The following list of verbs identify an EO and reference ILFs or EIFs: provide (*), print (*), display (*), produce (*), transmit (*), export (*), send (by the computer) (*), present (*), publish (*), calculate; the verbs marked by asterisks (*) may also be EQs if they do not contain calculations.

Other verbs that indicate an EQ and reference ILFs or EIFs are: query, search, decode, read, access, list.

Functional scores