z La ZIZ — O NE NEM Hmd | ii m zr ani —- —
zs
s — gas že v, TI m ana — 5 se mah i l H
a za ie smem lj '

ie i
Tam |

J'

i

AA NA Pe WEEKVAPPLICATION DEVELOPMENT

FEBRUARY 13, 1989

APPLIED INTELLIGENCE

Modify Your Methods To Take 'Advantage of ICASE Tools

grammers. As show in the figure, con- components should be catalogued in a reguired. When changes are made to the
In this, the :—— ventional technology often reguires CASE encyclopedia so that they can be design, the encyclopedia is automatically

last of a series " large development teams and long devel- selected when needed and modified as updated.

of columns on I- .: opment cycles. In contrast, many pro- reguired. The use of I-CASE tools avoids the
CASE technolo- ". jects or subprojects developed with I-.« ,. A large enterprise should employ the spaghettilike mess of the past and pro-
gy, we discuss CASE tools can be completed by one same I-CASE tool set at all locations motes cleanly structured engineering

the improve: —.. person. The brilliant, fast or hard work- where systems are built so that common with relatively fast and easy technigues
ment in produc- ing individual then has the opportunity data models, designs and program com- for maintenance.
tivity that can to excel. Management should encourage ponents can be used. Telecommunica- ' Although in some corporations the
be realized with | the most capable implementors to learn tions access to mainframe encyclopedias main thrust. with CASE tools is to in-
the use of CASE the full power of an I-CASE tool set in aids in the sharing of applications, docu-  crease the productivity of application

|

order to maximize productivity. /''. ment design, accounting procedures ske building, in others the goal is to improve
4. Toachieve ' 'x' Big projects should be subdivided into so on. | the guality of systems and achieve coor-
high productivity small projects, each of which can be in systems developed by traditional dination across a complex enterprise.
MARTIN in building com- ,'. completed.relatively guickly by one, two : manual technigues, maintenance is a The most, impressive system is not

puter applica- 4 or, at most, three implementors. The: major problem. Systems are often diffi- created with a single design and imple-

tions, it's neces- — CASE tool should make it possible to de- cult and Ee to change. Af-  , mentation. It evolves, being improved in
sary not only to select the best CASE '::.. | many steps at different times and

and I-CASE (integrated computer. aided. places.
software engineering) tools, but also to - The future will bring impressive soft-

adapt your organization and methods to Prod u ctiv ity Economy with ware and corporate computer systems,
take full advantage of these tools. : |. C ASE Tech nology and these will also be grown over many

The simplest, CASE tools are little | years Mari a ROE and organiza-
'more than diagramming aids. They sim- tions adding to them.
plify the drawing of asane and en- ii: It's difficult or impossible to grow
able them to be kept tidy and modified ;:;; C nv x Hi Tole al software that's a mess. To achieve long:
guickly. ; onve term evolution of software, we need

Code generators enable anlenentori! si Mu Iti- ( rso h | s acnca akce of data and struc-
to produce a working program guickly.: 4 eti tured models of processes.
However, if the generator is not linked P ole at Designs too complex for one person to

to a dictionary, data model or design know all the details of must be repre-
tools, the programs generated may be sented in an orderly fashion in an ency-

incompatible fragmen ilidesi ed and, i clopedia so that many people in many
not linked together. s ; a: uš he PRMAR RE places can add to the design. The design
To achieve high a dUCUVItY, the tools - h go Mo uovinsik needs standards and reusable compo-
for design need to be tightly coupled to jh Sai IRUEEVARAIA, nents and an architecture that facili-
the code generator. The tools should em-:.
ploy a data model and should enable sse
the design to be represented in a power: <;; iM |
ful, Viatal. easy- ti modify form (togi RS PMEKnG RSI: of computers that automatically place
which code is generated directly. . ns Mm telic mes! NO NE a EEA  orders, select suppliers, make trades and
The programs should be guickly exe-''" (HE si n G le pe rso z ENE EM NE zebra | so On, the behavior of these systems
cutable so that the designer can ObeEIVEJ B nie ASE P pioječi Rio O A sali ii ib ji db sis should be expressible in rules and dia-
what they do, adjust or add to the de-.;::;: tum ji SADNI di ske daro ana grams that the executives understand.
sign, rerun the programs, enhance the % ii

design and so on, until a autesnjj E When It Works, It Really Works

system is created. Some corporations have impressive

The principle of "what. you see is" A v computer systems designed to give them
what you get" should apply to-the com-

DE O V 39P (a competitive advantage: Take Ameri:

' bination of visual design tool and code "' ji 3 sh o k K AE ek JO le ena He » Can Airlines with its on-line terminals im
generator. The need for manual coding ' KON Na MVM, travel agents offices, or Benetton with
of procedures should be removed to the ;: KEM UI: its worldwide information system that

NO. OF PEOPLE B-

- tates the incremental addition of new
functions. — šel
For executives to control the behavior

maximum extent. ieaša irateia zaš Sušin Vie rusiji il Ni plani makes the world activities "transparent"
The generator may initially be used los a o RA SA MNE | pg SEPE Bavd kannum.:— to the decision makers near Rome.

produce structured code that relates to :' zr pi Mo piku KREN Mer osi jgra ni — reatej k | | | | Systems like these demonstrate how a

the design screens. This code may be ii A ČALI toke O ROI | corporation can pull ahead'of its compe-

used for prototyping and debugging. Be-  . 2.2, Sij) PRTI | | dia | tition by using information and automa-

cause structured code does not give opti-- ' K ; vi ntegrated CASE tools promote the use oj tion better. Efficient corporations will

uti roli tone be pode ME daš] ) highly productive small development teams and oridvride auto edine O MEDE,

into an optimizer, which creates code zi žeje | increase the Ure oj godopea applications. gle but nevertheless enable procedures to be

with optimal machine performance. This... ,,....,,, i | Sko slenih | adapted guickly to changing needs.

code will never be touched by mainte- si, EI Maki eki | UH ne JOK JE V ig ŠE nale To do this reguires engineering-style

rogrammers; all maintenance is pie |

| methodologies carried out with automat-
performed at the specification level. — .. fine with computer precision the ei ns - ter being modified many times, they of- : ed tools. It reguires encyclopedia-based

. The integrated designer-generator tool Wi faces between the comp ponents er. : ten become fragile; even minor changes I-CASE tools. Simple software engineer-
should aid in rapid construction and ed by separate teagnesi ši žaj Milek S result in bugs and breakdowns. ing is not enough—to build a computer-
modification of prototypes. It should | At the start ofa kje project, an enti- —' A goal of CASE tool sets is to pro- ized corporation we need information
generate test data and provide testing

ty-relationship diagram should be: creal, ' — duce systems that are guick and easy to engineering.
tools. It should generate database code /|/' ed for the data that will be used, the '

> change. Maintenance isn't done by dig: —
and job-control code, so that the pro- - data elements should be defined ind be? , $ing around in spaghetti code but by The James Martin Productivity Series,
gram can be executed guickly when de- — data correctly norma ' The same" dli) odificat. »sign specifications, an information service updated guar-
sign changes are made. | data model should be used for all sub- - followed by regeneration of code. terly, is available through High Pro-
Productivity in system development s" [ projesta. The now of data ane onimi Traditional maintenance is often | ductivity Software Inc., of Marble-
- strongly affected by the number of peo- | . among subr ts sl |. — made more difficult by lnedeguato docu: head, Mass. (800) 242-1240. For infor-
- ple in the development team. Large | mentati en. ince progr« mation on seminars, contact (in the
teams tend to give low development - United States and Canada) Technology

productivity, because the number of ne | HI B | zije NE Institute, 741 10th St., Santa
teractions between team upernbeka in- pna Calif. taci Sam ti 2la) a

4 i UI Me | PI |!
Ij ' 4 b. JU ns l zet | k. - 4 1 s j b ME ! HI h
i ke z kari R-a s i: upi m: Po ud 4 .e en z Mei J ie a sei
sa h o ge se si ae MM A v m Se pg PEN uri Ah —E0A?! ua nala, ei a | k - oč oz nm NUJO Že maje | ču
pare se —h.ciji: nala duda Ci aa Y za : Mae či: pa.E deb Ha "1
y se H h -— TH a e nič Uiea. —- (
k 1 il | HM adi mo | zet Masa ' | k 4,
| ' - sm: j 4 ra. Ii wi ae ti "4d | ! s aH o (5Š— ska 44 se- »4
| : "A, ti ge z lea, emi Pk — i k "RE . tiče asih . V 4 d
| i ris x ' sa Ni oi Ra zra aj Me eaj  EŠE audi ee NE Mi a ea s - Z Me Ve" ane ae Ja Pi ie ma ai jej h LJ te ga .i
| - Š) H Rea Beč ia f "a a H se Paš ; Z les ua ela pi ; Ure Lak H go tai sj pale po U Ra CE ti IH biča meji, rje ik ud s pare a a K
ma i ki neč "o nk - Šepet, du s Pe z nj IE K šir idi F Ki Par d m kaj K za in de Ža sa prs si a ga ia i
k, AT tre: 4 aro mika R | | ' aka ia
4 uč ši IM uf ud cb, d 1 b Liu j la | be kači kuri h 4, gati s | li mai | l i
- č i mat:. Pi umi MM oa - mn 6% li h h bi ža ma tore n ii sm eh.
h ka h - mag s za 7 "4 BP ral š nea Cal rj Hu di pisa
H ne Aja ma io aa o ea NOE a poo v zrn k sa magi epineia,.. z
a o 4 4 zj JI
1 Mičo pe k. Mo "S j h
s