. slab ii a B s , jel " ; DJ r] i ] - EJ r: KE 4 - Fe na l . | li da aH IBYv ima . Die di Na ! ije! zaj Hi ih s si - i k Hi: L: [ čah SE m Si JAM" i ad . 1 i 5 s REG "IB n.: j A : Hi | gg Li PU si z ta 1 Pi Pe ud di ji niji re mi do osa H: h Ši : i a ei še gi [ "a i tools life-cycle processes, per- A ČE - tribute to the | cesa of new The RAD | | E eti o dolg is ' based on the nine used by enterprises that, MARTIN have already achieved fast a pplicadons. development. RAD results are often im- pressive, with some achievements of more than 200 function points per per- son-month. The most successful tech- nigues used by these organizations were synthesized to form the basis for the RAD methodology, Reza in my pre-. j vious columns. Achieving heh speči one! is a complex process. Although many com- panies were very successful in achieving fast development, no enterprise was bori s | ing everything right. Many lacked criti- ' cal elements of the RAD life-cycle pro- cess. Some attempts at. fast development weren't productive because inadeguate h computer-aided software engineering (CASE) tools were used. And some mis- '" fired due to lack of user commitment. Still other projects didn't employ end .: users in the joint reguirements planning (JRP) and joint applications design (JAD) workshops. Some projects lacked a high-level executive owner to "grease . the wheels," while others didn't involve ' users in the evolution of the skem sea Some projects slipped because of inade- :"" guate cutover planning, others didn it gti emphasize guality. | One of the most powerful approaches | to fast development is to incorporate technigues for reusable design. The ma- Bi ; jority of the organizations studied, how- ever, weren't doing this. And some (hali were lacked an integrated code genera- tor; others didnt have a CASE reposi- tory. Reusability has a long-term payoff, giving steadily better results as more re- usable items are built. '' >": Information-systems (IS) managers '': can maximize productivity by identify- ' ing and combining all of the develop- ' ment, technigues that have demonstrat- ed high productivity. The RAD ioni ology attempts to do this. The figure ' summarizes the technigues that have iJ proven to be most effective. ' Dev Integrated CASE (I-CASE) tool sets are REV poeta OE RADJIKO S cycle pro ess to capture user specifica tlons, analyze them for k gical consistenc, npekane construction phase, a high - ly trained team of IS analysts uses I- ' CASE or o ovo te polti oe tem into a production application. To ' support these functions, the team re: iš guires a graphically orientec CASE '' tool set that integrates data- odeling, ' process-modeling, system-design, ti RE RAE II PMESa na HLA Tri PRE z čepe: PC WEEKNAPPLIGATION DEVELOPMENT pa H pobi, report generation, database- - Well managed 1»! generation, efficient code-generation and ' testing tools. | Specifications for an application are ) stored in an I-CASE repository. that s : available on-line to all developers. A de- € sign analyzer ensures that, specifications V tore in ine LCASE reposltory are ana " lyzed automatically for logical complete- ness, consistency and correctness.' ' ' —' CASE developers prosper most when [Ha K hey use I-CASE tool sets that generate "all the:code for an application from 9 s pecificadona stored in the I-CASE re- - pository. RAD encourages the use of a desktop development: environment, - which pves the fastest, zele za of APPLIED INTELLIGENCE en. All users must. know their role in the development life cycle. The most knowledgeable users must participate in the reguirements-planning and user-design workshops, which are conducted during the early phases of - the RAD life cycle; appropriate end us- ers should be selected to review the pro- totypes as the system evolves; and users and management must be highly moti- vated and enthusiastic about the devel- opment, process. Social disruption and the effects of cultural change can be minimized by ensuring that planning and education for job changes are done early in the life cycle. The IS project ua must; under- Factors Contributing to Success With Mk: KAR azhlčinublij Methodology optimizec for high-speed developmeni | Innovative management technigues Integrated CASE tools Reusable designs - Small, highly motivated teams Development within a time box Measurement of productivity End-user commitment Desktop development | environment John Avakian - In order for complex projects to succeed, development le o Le AUC One of the best approach- esto vive a EI is to reuse MJEEE Ni ma V ocelgh. ei and testing. Several %/ commercially available I-CGASE tool sets ' can generate code for an application on . — the desktop, eliminating the expense of J a mainframe code generator. A code op- : — timizer may also be used where oppro — prlate. Its important to have a high- level ex- ' 5 ecutive owner responsible for the sys- r tem who is financially committed to it - and is determined to move fast. The ex- ' ecutive owner must be briefed on the ' " events and timetable of the RAD life cy- ' -' cele. He or she must ensure that the us- — €rs want the system and are committed sto pbepiie establish its functions and de- Lee stand the RAD life cycle and, if possible, have prior experience with it. The project manager must also guide others " in the correct; use of the methodology. The JAD leader organizes and con- ducts the user JRP and JAD workshops. - This person, by doing these tasks contin- uously, will build up a high level of expertise,which is critical to consistent success with RAD. The construction, or SWAT (special. - ists with advanced tools), teams should be small, highly trained and skilled, with a powerful I-CASE tool set. They should be able to build systems rapidly from the red output of the JAD work- PPRIL 9, 1990 ization Key to High-Speed Development shops. A person skilled in data modeling can build or validate both the data mod- el used and the coordinating model for parallel development. Construction teams should be moti- vated by a sense of pride to achieve the RAD objectives. The RAD methodology, ideally repre- sented in hyperdocument form on each developer s PC, should give guidelines— and warnings of pitfalls—for each stage. The life cycle and methodology should be customized to adapt it to specific de- velopment practices and to make the best, use of the tool sets selected. The life cycle should have the follow- ing phases: reguirements planning, user design (these may be combined), con- struction and cutover. The reguirements-planning phase should incorporate a user JRP work- shop. The user design phase should in- clude one or more user JAD workshops, using I-CASE tool sets and prototyping technigues. The design specifications de- veloped in this phase should be stored in the repository of an I-CASE tool set so that these specifications can be used directly in the construction phase, where the design is defined in detail and used to drive a code generator. The planning and preparation for the cut: over phase should proceed in tandem with the other development phases. Obtaining High Ouality High guality can be obtained if all de- sign specifications and procedural code for the system are defined with the I- CASE tool and stored in the I-CASE re- pository, so the tool set can check their integrity and consistency. Hand-drawn diagrams and English text of traditional specifications should be avoided because a computer cannot, check them and they are usually full of errors. Specifications and design should be created in user workshops. This is faster than the traditional process of specifica- tion writing by analysts, and it meets users needs much better. Wherever pos- sible, documentation should be printed from the I-CASE tool set. Some tools generate development, workbooks. Standards and guidelines for usability are also necessary. A person with skill in this area should check the SANMI system for these gualities. The use of these technigues, within the framework of the RAD life cycle, will result in consistent, improvements in software-development produtctivity. Next week, Ill begin a series of arti- cles on reverse engineering, a new set of technigues that, attempts to convert data definitions and procedural code into higher-level design specifications. B The Loncepts embodied in RAD are de- zcribed in a new volume in the James Martin Report Series. For more infor- mation on this volume, call (800) 242: - 1240. For information on seminars, contact (in the United. States and Can-. ada) Technology Transfer Institute, 741 10th St., Santa Monica, Calif. 90402 (213) 394-8305. In Europe, con- paš Savant, 2 New St., Carnforth, nes., LA5 9BX United Kingdom (0524) 734 505