Faku
ltät I
nfor
mat
ik, I
nstit
ut fü
r Sof
twar
e- u
nd M
ultim
edia
tech
nik,
Leh
rstu
hl fü
r Sof
twar
etec
hnol
ogie
20
Des
ign
Met
hod
s -
An
Ove
rvie
w
Ø Pr
of.
Dr.
U.
Aßm
ann
Ø Te
chni
sche
Uni
vers
ität
Dre
sden
Ø In
stitu
t fü
r So
ftw
are-
und
Mul
timed
iate
chni
k Ø Gr
uppe
Sof
twar
etec
hnol
ogie
Ø ht
tp:/
/st.
inf.tu
-dre
sden
.de/
teac
hing
/sw
t2
Ø Ve
rsio
n 12
-1.1
, 15
.12.
12
1. F
rom
Req
uir
emen
ts t
o D
esig
n
2. W
hat
is a
Des
ign
Met
hod
? 3
. Ove
rvie
w o
f D
esig
n M
eth
ods
1. F
un
ctio
nal
Dev
elop
men
t 2
. Act
ion
-Bas
ed D
evel
opm
ent
3. C
omp
onen
t-B
ased
Dev
elop
men
t 4
. Dat
a-O
rien
ted
Dev
elop
men
t 5
. Ob
ject
-ori
ente
d D
evel
opm
ent
6. T
ran
sfor
mat
ive
Dev
elop
men
t 7
. Gen
erat
ive
Dev
elop
men
t 8
. Mod
el-D
rive
n S
oftw
are
Dev
elop
men
t 9
. For
mal
Met
hod
s 1
0. A
spec
t-or
ien
ted
Dev
elop
men
t 4
. Oth
er A
rch
itec
tura
l Sty
les
5. D
esig
n H
euri
stic
s an
d B
est
Pra
ctic
es
Obl
igat
ory
Rea
ding
s
Ø P
fleeg
er C
hapt
er 5
Ø G
hezz
i Cha
pter
3
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
2
Sec
onda
ry L
itera
ture
Ø [
Thay
er]
Ric
hard
Tha
yer.
Sof
twar
e En
gine
erin
g. A
cur
ricu
lum
boo
k.
IEEE
Pre
ss
Ø [
Bud
gen]
Dav
id B
udge
n. S
oftw
are
Des
ign:
An
Intr
oduc
tion.
In
[Tha
yer]
Ø [
Thay
er&
McG
ettr
ick]
Ric
hard
Tha
yer,
And
rew
McG
ettr
ick.
Sof
twar
e En
gine
erin
g -
A E
urop
ean
Pers
pect
ive.
IEE
E Pr
ess
Ø [
Parn
as]
Dav
id P
arna
s. O
n th
e Crite
ria
To B
e U
sed
in D
ecom
posi
ng
Sys
tem
s in
to M
odul
es.
Com
mun
icat
ions
of th
e ACM
Dec
. 19
72.
The
clas
sic
artic
le o
n m
odul
arity
Ø [
Bro
oks]
Fre
derick
P.
Bro
oks
jr.
No
Silv
er B
ulle
t. E
ssen
ce a
nd
Acc
iden
ts o
f Sof
twar
e En
gine
erin
g. I
n [T
haye
r].
Won
derf
ul a
rtic
le
on w
hat
soft
war
e en
gine
erin
g is
all
abou
t Ø H
eise
Dev
elop
er P
odca
st h
ttp:
//w
ww
.hei
se.d
e/de
velo
per/
podc
ast/
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
3
Lite
ratu
re
Ø [
Bud
gen]
Dav
id B
udge
n. S
oftw
are
Des
ign.
Add
ison
-Wes
ley.
Ex
pand
s on
the
Bud
gen
pape
r. P
rett
y sy
stem
atic
. Ø [
Sha
w/G
arla
n] S
oftw
are
Arc
hite
ctur
e. 1
996.
Pre
ntic
e-H
all.
Gre
at
book
for
arc
hite
cts.
Ø [
Sha
w/C
lem
ents
] M
. Sha
w,
P. C
lem
ents
. A F
ield
Gui
de t
o Box
olog
y.
Ø [
Endr
es/R
omba
ch]
A.
Endr
es,
D.
Rom
bach
. A H
andb
ook
of s
oftw
are
and
syst
ems
engi
neer
ing.
Em
pirica
l obs
erva
tions
, la
ws
and
theo
ries
. Add
ison
-Wes
ley.
Ver
y go
od c
olle
ctio
n of
sof
twar
e la
ws.
N
ice!
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
4
Goa
ls
Ø G
et a
n ov
ervi
ew o
n th
e av
aila
ble
met
hods
to
com
e fr
om a
re
quirem
ents
spe
cific
atio
n to
the
des
ign
Ø U
nder
stan
d th
at s
oftw
are
engi
neer
s sh
ould
n't
get
stuc
k by
a
spec
ific
desi
gn m
etho
d
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
5
Sce
nario
Ø Y
ou a
re a
pro
ject
man
ager
in M
iller
Car
Rad
ios,
Inc
Ø Y
our
boss
com
es in
to y
our
offic
e an
d sa
ys:
“Our
com
petit
or S
mith
Car
Rad
ios
has
a ne
w s
atel
lite
radi
o. T
heir
sale
s ar
e gr
owin
g, a
nd o
ur c
usto
mer
s de
man
d it,
too
. H
ow q
uick
ly
can
you
deliv
er m
e a
sate
llite
rad
io?”
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
6
Sce
nario
Ø Y
ou a
re a
pro
ject
man
ager
in M
iller
Car
Rad
ios,
Inc
Ø Y
our
boss
com
es in
to y
our
offic
e an
d sa
ys:
Ø “O
ur c
ompe
titor
Sm
ith C
ar R
adio
s ha
s a
new
sa
telli
te r
adio
. Th
eir
sale
s ar
e gr
owin
g, a
nd o
ur
cust
omer
s de
man
d it,
too
. H
ow q
uick
ly c
an y
ou
deliv
er m
e a
sate
llite
rad
io?”
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
cost
cal
cula
tion tim
e pl
anni
ng
requ
irem
ents
an
alys
is
syst
em d
esig
n
proj
ect m
anag
emen
t so
ftwar
e pr
oces
s
7
The
Idea
l Des
ign
Proc
ess
Ø "
Des
ign
prod
uces
a w
orka
ble
solu
tion
to a
giv
en
prob
lem
" [B
udge
n]
Ø "
Des
ign
is t
he d
escr
iptio
n of
a s
olut
ion"
[Pf
leeg
er]
Ø "
The
Des
ign
Proc
ess
is t
he c
reat
ive
proc
ess
of t
rans
form
ing
the
prob
lem
into
a s
olut
ion"
[Pf
leeg
er]
Ø G
oal:
Thi
s le
ctur
e pr
esen
ts s
ome
syst
emat
ic w
ays
how
to
com
e to
a
wor
kabl
e so
lutio
n fo
r a
give
n pr
oble
m
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
8
20.1
Fro
m R
equi
rem
ents
to
Des
ign
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Syst
em in
tera
ctio
n an
alys
is
(sys
tem
con
text
ana
lysi
s)
Arc
hite
ctur
al S
tyle
A
naly
sis
Func
tiona
l Des
ign
Requ
irem
ent
Ana
lysi
s
OO
Des
ign
Tran
sfor
mat
ive
Des
ign
Syst
em in
terf
aces
(Con
text
mod
el)
Arc
hite
ctur
al D
esig
n
Arc
hite
ctur
e (S
AS)
Func
tiona
l Des
ign
OO
Des
ign
Tran
sfor
mat
ive
Des
ign
Det
aile
d D
esig
n
SD
DS
...
...
Top-
leve
l arc
hite
ctur
e
9
Ø Co
nten
ts o
f th
e Sof
twar
e Req
uire
men
ts S
peci
ficat
ion
(SRS)
(rep
.)
Ø T
he S
oftw
are
Req
uire
men
t
Spe
cific
atio
n (S
RS)
cont
ains
a
list
of t
hing
s th
e sy
stem
has
to
fu
lfill
Ø E
xam
ple
[Ric
hard
Fai
rley
,
Sof
twar
e En
gine
erin
g]
Ø U
sual
ly,
spec
ifica
tion
lang
uage
s
are
the
sam
e or
sim
ilar
for
re
quirem
ents
an
d de
sign
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Ø O
verv
iew
of Pr
oduc
t
Ø B
ackg
roun
d, E
nviron
men
t
Ø I
nter
face
s of
the
Sys
tem
(c
onte
xt m
odel
) Ø I
/O in
terf
aces
, da
ta for
mat
s (s
cree
ns,
prot
ocol
s, e
tc.)
, Com
man
ds
Ø O
verv
iew
of da
ta flo
w t
hrou
gh
syst
em,
Dat
a di
ctio
nary
Ø F
unct
iona
l req
uire
men
ts
Ø N
on-f
unct
iona
l req
uire
men
ts
Ø E
rror
han
dlin
g
Ø P
rior
itiza
tion
Ø P
ossi
ble
exte
nsio
ns
Ø A
ccep
tanc
e te
st c
rite
ria
Ø D
ocum
enta
tion
guid
elin
e
Ø L
itera
ture
Ø G
loss
ary
10
Con
tent
s of
the
Sof
twar
e Arc
hite
ctur
al D
esig
n Spe
cific
atio
n (S
AD
, SAS)
Ø C
once
ptua
l abs
trac
tion
leve
l Ø C
once
ptua
l ins
tead
of te
chni
cal
Ø C
oars
e gr
ain
inst
ead
of d
etai
led
Ø D
esig
n di
men
sion
s Ø S
truc
ture
(pa
rt-o
f re
latio
ns,
is-a
rel
atio
ns)
Ø F
unct
ion
(typ
es,
inte
rfac
es)
Ø B
ehav
ior
Ø S
yste
m c
ompo
nent
s an
d th
eir
inte
rfac
es
Ø C
ontr
act
spec
ifica
tions
of m
odul
es:
how
to
use
a m
odul
e?
Ø W
hat
shou
ld it
tak
e, w
hat
deliv
er (
pre-
and
pos
tcon
ditio
ns)
Ø C
ompo
nent
rel
atio
ns
Ø U
ses,
is-a
, pa
rt-o
f, b
ehav
es-l
ike
Ø C
onne
ctio
ns
Ø A
rchi
tect
ural
sty
les
(arc
hite
ctur
al p
atte
rns)
Ø C
oars
e gr
ain
patt
erns
of th
e ar
chite
ctur
e in
ter
ms
of c
ontr
ol flo
w a
nd d
ata
flow
Ø C
onst
rain
ts o
f m
odul
es,
rela
tions
, an
d co
nnec
tions
Ø D
esig
n pa
tter
ns
Ø M
icro
-str
uctu
res
in t
he d
esig
n m
odel
, m
ostly
on
the
colla
bora
tion
of 2
-5
clas
ses
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
11
Con
tent
s of
Det
aile
d D
esig
n D
ocum
ent
(SD
DS)
Ø S
DD
S =
Sof
twar
e D
etai
led
Des
ign
Spe
cific
atio
n
Ø F
ine-
grai
ned
desi
gn
Ø T
echn
ical
inst
ead
of c
once
ptua
l Ø S
ketc
h of
the
impl
emen
tatio
n w
ith p
seud
o co
de,
stat
echa
rts,
pet
ri n
ets,
or
othe
r de
sign
not
atio
ns
Ø B
ehav
iora
l mod
el
Ø T
ells
mor
e ab
out
the
HO
W,
with
out
givi
ng t
he im
plem
enta
tion
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
12
20
.2 D
ESIG
N M
ETH
OD
S
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
13
A S
oftw
are
Des
ign
Met
hod
(aka
Dev
elop
men
t M
etho
d)
… h
as 3
com
pone
nts
[Bud
gen]
: 1. R
epre
sent
atio
n pa
rt (
nota
tion,
lang
uage
) Ø S
et o
f no
tatio
ns in
(in
form
al)
text
ual,
(sem
i-fo
rmal
) di
agra
mm
atic
, or
m
athe
mat
ic (
form
al)
form
2. P
roce
ss m
odel
(“V
orge
hens
mod
ell”
, “P
roze
ssm
odel
l”)
Ø D
esig
n st
rate
gy:
A b
asic
des
ign
ques
tion
(foc
us o
f re
finem
ent)
Ø R
estr
uctu
ring
met
hods
Ø C
onsi
sten
cy c
heck
ing
3. S
et o
f he
uris
tics
Ø G
ener
al r
ules
of th
umb
Ø P
roce
ss-s
peci
fic r
ules
Ø P
roce
ss p
atte
rns
Ø D
esig
n pa
tter
ns
Ø A
dapt
atio
n ru
les
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
14
20.2
.1 D
esig
n Rep
rese
ntat
ions
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
15
Des
ign
Lang
uage
s
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Pro
gram
min
g
Lang
uage
s
Spe
cific
atio
n
Lang
uage
s
Exe
cuta
ble
Spe
cific
atio
n
Lang
uage
s
Pap
er
Spe
cific
atio
n
Lang
uage
s
Text
D
iagr
ams
Mat
h
Par
seab
le n
atur
al
lang
uage
In
form
al
Nat
ural
lang
uage
P
seud
o-co
de
Flow
cha
rt D
ata-
flow
Dia
gram
E
ntity
-Rel
atio
nshi
p
Dia
gram
ER
U
ML
Stru
ctur
e D
iagr
am
Sta
tech
arts
W
orkf
low
lang
uage
s (B
PE
L, B
PM
N)..
.
Col
ored
Pet
ri ne
ts
Sta
te m
achi
nes
Vien
na D
evel
opm
ent
Lang
uage
VD
L/V
DM
Z B
ELA
N
SE
TL
Java
S
cala
C
#
Larc
h (w
ith p
rove
r)
CS
P
CC
S 16
Mod
elic
a M
etam
odel
ica
Mat
lab
Sim
ulin
k
20
.2.2
DES
IGN
P
RO
CES
SES
Gen
eric
ste
ps
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
17
20.2
.3 D
esig
n Pr
oces
s
Ø A
des
ign
pro
cess
is a
str
uctu
red
algo
rith
m (
or w
orkf
low
) to
ac
hiev
e a
desi
gn m
odel
fro
m a
req
uire
men
t sp
ecifi
catio
n Ø A
seq
uenc
e of
ste
ps
Ø A
set
of m
ilest
ones
Ø T
he d
esig
n pr
oces
s st
arts
fro
m t
he s
yste
m’s
inte
rfac
es (
cont
ext
code
l) a
nd r
efin
es it
s in
tern
als
Ø E
very
des
ign
proc
ess
Ø C
onta
ins
seve
ral c
entr
al g
ener
ic s
teps
Ø U
ses
gene
ral d
esig
n st
rate
gies
Ø E
nds
up in
a s
peci
fic a
rchi
tect
ural
sty
le
Ø D
esig
n pr
oces
ses
belo
ng t
o so
ftw
are
deve
lopm
ent
met
hods
/pr
oces
ses
• To
geth
er w
ith r
equi
rem
ents
, te
stin
g et
c.
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
18
Rep
etiti
on:
Gen
eric
Ste
ps in
Des
ign
Proc
esse
s
Ever
y de
sign
pro
cess
con
tain
s so
me
gene
ric
step
s Ø E
labo
ratio
n Ø
Wor
k ou
t a
cert
ain
aspe
ct o
f th
e de
sign
mod
el,
usin
g an
app
ropr
iate
des
ign
nota
tion
Ø R
efin
emen
t Ø
Ref
ine
an e
xist
ing
spec
ifica
tion/
mod
el,
repl
acin
g ab
stra
ct p
arts
by
deta
ils,
e.g.
, ad
d pl
atfo
rm-s
peci
fic d
etai
ls
Ø
Ret
ain
refin
emen
t co
nditi
ons
such
as
embe
ddin
g Ø A
bstr
actio
n is
the
opp
osite
of re
finem
ent
Ø C
heck
ing
Con
sist
ency
Ø C
heck
ing
busi
ness
rul
es a
nd c
onte
xt c
onst
rain
ts
Ø R
estr
uctu
re (
mor
e st
ruct
ure,
but
kee
p se
man
tics)
Ø S
plit
(dec
ompo
se,
intr
oduc
e hi
erar
chie
s, la
yers
, re
duci
bilit
y)
Ø C
oale
sce
(rea
rran
ge)
Ø S
ymm
etry
ope
ratio
ns (
sem
antic
s-pr
eser
ving
, re
stru
ctur
ing)
: Ø S
eman
tic r
efin
emen
t Ø R
efac
toring
Ø C
hang
e Rep
rese
ntat
ion
(Not
atio
n):
Ø S
impl
ifica
tion
(fac
toring
, tr
ansi
tive
redu
ctio
n, fac
adin
g)
Ø C
hang
e re
pres
enta
tion,
but
kee
p se
man
tics
Ø T
rans
form
a c
erta
in r
epre
sent
atio
n of
the
mod
el in
to a
noth
er o
ne
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
19
Dev
elop
men
t O
pera
tions
of D
esig
n M
etho
ds
Ø E
very
not
atio
n ha
s el
abor
atio
n, r
efin
emen
t, c
heck
ing,
and
st
ruct
urin
g op
erat
ions
Ø H
and
oper
atio
ns
Ø S
plit
(dec
ompo
se,
intr
oduc
e hi
erar
chie
s, la
yers
, re
duci
bilit
y Ø C
oale
sce
(rea
rran
ge)
Ø A
utom
atic
ope
ratio
ns
Ø G
raph
ana
lysi
s m
etho
ds,
such
as
cons
trai
nts
Ø G
raph
str
uctu
ring
met
hods
, e.
g.,
grap
h an
alys
is o
r tr
ansf
orm
atio
ns
Ø T
ext-
base
d sp
ecifi
catio
ns c
an b
e tr
ansf
orm
ed in
to A
SG
s an
d th
en s
truc
ture
d by
gra
ph s
truc
turing
met
hods
Ø S
ome
nota
tions
hav
e sp
ecifi
c au
tom
atic
met
hods
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
20
20.2
.2.1
Arc
hite
ctur
al S
tyle
s as
Res
ults
of D
esig
n Pr
oces
ses
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
21
Den
ert's
Law
on
Arc
hite
ctur
al S
tyle
s
Ø E
rnst
Den
ert.
Sof
twar
e En
gine
erin
g. S
prin
ger,
199
1.
Ø C
onse
quen
ce o
f D
ener
t's la
w:
Ø i
f w
e ca
n sp
lit o
ff a
con
cern
in a
n ap
plic
atio
n do
mai
n, w
e ar
rive
at
a ne
w
stan
dard
arc
hite
ctur
e (a
rchi
tect
ural
sty
le)
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Sep
arat
ion
of c
once
rns
lead
s to
sta
ndar
d ar
chit
ectu
res.
E
. Den
ert,
199
1
22
Arc
hite
ctur
al S
tyle
s
Ø A
n ar
chit
ectu
re s
tyle
em
ploy
s ce
rtai
n ty
pes
of c
once
pts
Ø C
erta
in t
ypes
of co
mpo
nent
s w
ith
Ø C
erta
in t
ypes
of co
nnec
tions
/con
nect
ors
Ø A
nd a
cer
tain
rel
atio
n be
twee
n co
ntro
l and
dat
a flo
w
Ø A
rchi
tect
ural
sty
les
enab
le u
s to
tal
k ab
out
the
coar
se-g
rain
st
ruct
ure
of a
sys
tem
Ø G
ood
for
docu
men
tatio
n an
d co
mpr
ehen
sion
Ø G
ood
for
mai
nten
ance
Ø A
rchi
tect
ural
Sty
les
vs D
esig
n Pa
tter
ns
Ø D
esig
n pa
tter
ns h
ave
been
cal
led
mic
roar
chite
ctur
es
Ø T
hey
gras
p a
rela
tions
hip
betw
een
seve
ral c
lass
es o
f an
app
licat
ion,
but
not
of
the
entir
e ar
chite
ctur
e or
sub
syst
em
Ø A
rchi
tect
ural
sty
les
are
coar
se-g
rain
des
ign
patt
erns
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
23
Wha
t Is
In
a Sty
le ?
Ø A
sty
le h
as 5
maj
or c
once
rns,
in w
hich
it c
an v
ary
[Sha
w/G
arla
n]
Ø S
tru
ctu
ral P
arts
: co
mpo
nent
s, in
terf
aces
(po
rts)
, co
nnec
tors
, cl
asse
s,
obje
cts,
mod
ules
Ø C
ontr
ol f
low
Ø T
opol
ogy
(in
whi
ch for
m c
oord
inat
ion
take
n pl
ace?
) Ø S
ynch
roni
satio
n (s
ynch
rono
us,
asyn
chro
nous
) Ø B
indi
ng t
ime
(Whe
n ar
e th
e co
mpo
nent
s or
gani
zed?
) Ø D
ata
flow
Ø T
opol
ogy
(How
doe
s th
e da
ta flo
w?)
Ø C
ontin
uity
(si
ngul
ar,
spor
adic
, co
ntin
uous
, st
rong
, w
eak)
Ø M
odus
(sh
ared
mem
ory,
mes
sage
s, .
.)
Ø I
nte
ract
ion
bet
wee
n co
ntro
l- a
nd d
ata
flow
Ø Is
omor
phic
sim
ilar
to a
dat
a st
ruct
ure
Ø D
irec
tion
(par
alle
l, an
tipar
alle
l)
Ø I
nva
rian
ts
Ø F
eatu
res
that
nev
er c
hang
e
Ø A
nal
ysis
fea
ture
s Ø H
ow c
an b
e ar
chite
ctur
e be
ana
lyze
d?
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
24
The
Des
ign
Prob
lem
Ø H
ow d
o I
derive
at
a de
sign
for
the
sys
tem
?
Ø H
ow d
o I
derive
at
an a
rchi
tect
ural
sty
le for
the
sys
tem
? Ø H
ow d
o I
derive
a d
etai
led
desi
gn?
Ø M
ost
ofte
n, a
fter
rea
ding
the
req
uire
men
ts,
the
syst
em lo
oks
like
in m
ist
Ø D
evel
oper
s ha
ve a
bad
fee
ling
in t
heir s
tom
ach
Ø T
hey
feel
the
ir w
ay for
war
d Ø
Impo
rtan
t is
: w
hich
que
stio
ns a
re a
sked
? Ø I
n de
sign
mee
tings
, th
e ba
sic
desi
gn q
uest
ions
are
pos
ed o
ver
and
over
aga
in,
until
a d
esig
n is
fou
nd
Ø S
elec
t a
desi
gn m
etho
d Ø P
ose
the
desi
gn m
etho
d's
basi
c qu
estio
n Ø P
erfo
rm t
he d
esig
n m
etho
d's
proc
ess
Ø Pe
rfor
m t
he d
esig
n m
etho
d's
step
s Ø I
f pr
oces
s ge
ts s
tuck
, ch
ange
des
ign
met
hod
and
try
anot
her
one
Ø H
owev
er,
be a
war
e, w
hich
des
ign
met
hod
and
proc
ess
you
use
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
25
Des
ign
Proc
esse
s ha
ve a
Foc
us o
f El
abor
atio
n an
d Ref
inem
ent
Ø A
cent
ral v
iew
poin
t w
ith a
set
of
conc
erns
, ac
cord
ing
to w
hich
the
sys
tem
is
elab
orat
ed
• Dec
ompo
sed
• Ref
ined
• c
ompo
sed
Ø An
ela
bora
tion
stra
tegy
Ø Th
e ce
ntra
l que
stio
n
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
26
20.3
Ove
rvie
w o
f El
abor
atio
n Str
ateg
ies
Ø A
des
ign
met
hod
relie
s on
a e
lab
orat
ion
str
ateg
y, in
clud
ing
a ba
sic
ques
tion
the
deve
lope
r ha
s to
pos
e hi
mse
lf, o
r th
e te
am a
sks
itsel
f Ø A
diff
eren
t qu
estio
n gi
ves
a di
ffer
ent
desi
gn m
etho
d Ø M
etho
ds c
an b
e gr
oupe
d ac
cord
ing
to t
heir foc
us o
f de
com
posi
tion
and
the
desi
gn n
otat
ion
they
use
. Ø F
un
ctio
n-o
rien
ted:
func
tion
in foc
us
Ø A
ctio
n-o
rien
ted
, ev
ent-
acti
on-o
rien
ted
: Act
ion
in foc
us
Ø D
ata-
orie
nte
d:
A d
ata
stru
ctur
e is
in foc
us
Ø C
omp
onen
t-or
ien
ted
(st
ruct
ure
-ori
ente
d):
par
ts in
foc
us
Ø O
bje
ct-o
rien
ted:
obje
cts
(dat
a an
d co
rres
p. a
ctio
ns)
in foc
us
Ø T
ran
sfor
mat
ion
al:
basi
c ac
tion
is t
he t
rans
form
atio
n Ø G
ener
ativ
e: b
asic
act
ion
is a
spe
cial
for
m o
f t
rans
form
atio
n, t
he
gene
ratio
n. A
lso
usin
g pl
anni
ng
Ø F
orm
al m
eth
ods:
cor
rect
ref
inem
ent
and
form
al p
roof
s in
foc
us
Ø A
spec
t-or
ien
ted
met
hod
s: r
efin
emen
t ac
cord
ing
to v
iew
poin
ts a
nd
conc
erns
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
27
20.3
.1)
Func
tion-
Orien
ted
Des
ign
(Ope
ratio
n-or
ient
ed,
Mod
ular
Des
ign)
Ø D
esig
n w
ith fun
ctio
nal u
nits
whi
ch t
rans
form
inpu
ts t
o ou
tput
s Ø M
inim
al s
yste
m s
tate
Ø I
nfor
mat
ion
is t
ypic
ally
com
mun
icat
ed v
ia p
aram
eter
s or
sha
red
mem
ory
Ø N
o te
mpo
ral a
spec
t to
fun
ctio
ns
Ø F
unct
ions
/ope
ratio
ns a
re g
roup
ed t
o m
odul
es o
r co
mpo
nent
s Ø D
ivid
e: fin
ding
sub
func
tions
Ø C
onqu
er:
grou
ping
to
mod
ules
Ø E
xam
ples
Ø P
arna
s' c
hang
e-or
ient
ed d
esig
n (i
nfor
mat
ion-
hidi
ng b
ased
des
ign,
see
ST-
1)
Ø L
ayer
ed a
bstr
act
mac
hine
s (s
ee S
T-1)
Ø U
se:
whe
n th
e sy
stem
has
a lo
t of
diff
eren
t fu
nctio
ns
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Wha
t ar
e th
e fu
ncti
ons
of t
he s
oftw
are
and
thei
r su
bfun
ctio
ns?
28
Ø "
Div
ide
and
Con
quer
” of
fun
ctio
n Ø D
ecom
pose
sys
tem
into
sm
alle
r an
d sm
alle
r pi
eces
•
Idea
lly,
each
pie
ce c
an b
e so
lved
sep
arat
ely
• Id
eally
, ea
ch p
iece
can
be
mod
ified
inde
pend
ent
of o
ther
pie
ces
Ø R
ealit
y: e
ach
piec
e m
ust
com
mun
icat
e w
ith o
ther
pie
ces
•
This
com
mun
icat
ion
impl
ies
a ce
rtai
n co
st
• At
som
e po
int
the
cost
is m
ore
than
the
ben
efit
prov
ided
by
the
indi
vidu
al
piec
es
• At
this
poi
nt,
the
deco
mpo
sitio
n pr
oces
s ca
n st
op
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
29
20.3
.2)
Act
ion-
Orien
ted
Des
ign
Ø A
ctio
n-or
ient
ed d
esig
n is
sim
ilar
to fun
ctio
n-or
ient
ed d
esig
n, b
ut
actio
ns r
equi
re s
tate
on
whi
ch t
hey
are
perf
orm
ed (
impe
rativ
e,
stat
e-or
ient
ed s
tyle
)
Ø D
ivid
e: fin
ding
sub
actio
ns
Ø C
onqu
er:
grou
ping
to
mod
ules
Ø E
xam
ples
: Ø P
etri N
ets
Ø U
se-c
ase-
base
d de
velo
pmen
t Ø D
ata-
flow
bas
ed d
evel
opm
ent
SA,
SAD
T
Ø U
se:
whe
n th
e sy
stem
map
s to
a s
tate
spa
ce,
in w
hich
act
ions
for
m
the
tran
sitio
ns
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Wha
t ar
e th
e ac
tion
s th
e sy
stem
sho
uld
perf
orm
?
30
Str
uctu
ral D
ecom
posi
tion
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
A y
A y
x x
A
x y
A
x x
inpu
t ou
tput
tra
nsfo
rm
Co-
ordi
nate
31
Res
ult
1: C
all-
Bas
ed A
rchi
tect
ural
Sty
le
Ø C
ompo
nent
s de
note
pro
cedu
res
that
cal
l eac
h ot
her
Ø C
ontr
ol flo
w is
sym
met
ric
(cal
ls a
nd s
ymm
etric
retu
rns)
Ø D
ata-
flow
can
be
Ø p
aral
lel t
he c
all (
push
-bas
ed s
yste
m):
cal
ler
push
es d
ata
into
cal
lee
Ø a
ntip
aral
lel,
i.e.,
par
alle
l to
the
retu
rn (
pull-
base
d sy
stem
): c
alle
r dr
ags
out
data
fro
m c
alle
e Ø A
ka “
Clie
nt-S
erve
r” in
loos
ely
coup
led
or d
istr
ibut
ed s
yste
ms
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Mod
ule
Mod
ule
Mod
ule
Syst
em
call
retu
rn
call
retu
rn
call
retu
rn
call
retu
rn
call
retu
rn
32
Res
ult
2: D
ata-
Flow
Bas
ed S
yste
ms
(Pip
e-an
d-Fi
lter,
Cha
nnel
s, S
trea
ms)
Ø I
f da
ta flo
ws
in s
trea
ms,
cal
l-ba
sed
syst
ems
are
exte
nded
to
stre
am-b
ased
sy
stem
s Ø C
ompo
nent
s: p
roce
sses
, co
nnec
tors
: st
ream
s Ø C
ontr
ol flo
w is
asy
nchr
onou
s, c
ontin
uous
Ø D
ata-
flow
gra
ph o
f co
nnec
tions
, st
atic
or
dyna
mic
bin
ding
Ø D
ata-
flow
can
be
para
llel t
o th
e co
ntro
l-flo
w (
push
-bas
ed s
yste
m)
or
antip
aral
lel (
pull-
base
d sy
stem
)
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
arch
itect
ural
glu
e co
de
Filte
r
Pipe
cat s
erve
r.log
| gr
ep "A
ddin
g st
uden
t"
33
Exam
ples
Dat
a-flo
w b
ased
sys
tem
s:
Ø I
mag
e pr
oces
sing
sys
tem
s Ø M
icro
scop
y, o
bjec
t re
cogn
ition
Ø D
igita
l sig
nal p
roce
ssin
g sy
stem
s Ø V
ideo
and
aud
io p
roce
ssin
g, e
.g.,
the
sat
ellit
e ra
dio
Ø C
onte
nt m
anag
emen
t sy
stem
s (C
MS)
Ø D
ata
is s
tore
d in
XM
L or
rel
atio
nal f
orm
at
Ø P
ipel
ines
pro
duce
dis
play
for
mat
Ø B
atch
-pro
cess
ing
syst
ems
Ø U
NIX
she
ll sc
ript
s pr
ovid
es u
ntyp
ed d
ata
flow
(te
xts)
Ø M
icro
soft
Pow
er S
hell
prov
ides
typ
ed d
ata-
flow
, ty
ped
in X
ML
Cal
l-ba
sed
syst
ems:
Ø O
bjec
t-or
ient
ed fra
mew
orks
Ø L
ayer
ed a
rchi
tect
ures
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
34
20.
3.2.
2) E
vent
-Con
ditio
n-Act
ion-
Orien
ted
Des
ign
Ø E
vent
-con
ditio
n-ac
tion
rule
s (E
CA r
ules
) Ø O
n w
hich
eve
nt,
unde
r w
hich
con
ditio
n, fol
low
s w
hich
act
ion?
Ø D
ivid
e: fin
ding
rul
es for
con
text
s Ø C
onqu
er:
grou
ping
of ru
les
to r
ule
mod
ules
Ø E
xam
ple:
Ø B
usin
ess-
rule
-bas
ed d
esig
n (S
BVR)
Ø U
se:
whe
n th
e sy
stem
map
s to
a s
tate
spa
ce,
in w
hich
act
ions
for
m
the
tran
sitio
ns a
nd t
he a
ctio
ns a
re g
uard
ed b
y ev
ents
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Wha
t ar
e th
e ev
ents
tha
t m
ay o
ccur
and
ho
w d
oes
my
soft
war
e re
act
on t
hem
?
35
Arc
h. S
tyle
: Ev
ent-
base
d Arc
hite
ctur
al S
tyle
(Im
plic
it In
voca
tion
Sty
le)
Ø C
ompo
nent
s: p
roce
sses
or
proc
edur
es
Ø C
onne
ctor
s: A
nony
mou
s co
mm
unic
atio
n by
eve
nts
Ø A
sync
hron
ous
com
mun
icat
ion
Ø D
ynam
ic t
opol
ogy:
Lis
tene
rs c
an d
ynam
ical
ly r
egis
ter
and
unre
gist
er
Ø L
iste
ners
are
impl
icitl
y in
voke
d by
eve
nts
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
On
Eve
nt
If C
ondi
tion
then
Act
ion
On
Eve
nt
If C
ondi
tion
then
Act
ion
On
Eve
nt
If C
ondi
tion
then
Act
ion
36
SBVR E
xam
ple
(OM
G B
usin
ess
Rul
e La
ngua
ge)
curr
ent
cont
act
deta
ils
C
once
pt T
ype:
rol
e
D
efin
ition
: c
onta
ct d
etai
ls o
f re
ntal
tha
t ha
ve b
een
conf
irm
ed b
y re
nter
of r
enta
l
rent
al
D
efin
ition
: c
ontr
act
that
is w
ith r
ente
r an
d sp
ecifi
es u
se o
f a
car
of c
ar g
roup
and
is for
re
ntal
per
iod
and
is for
ren
tal m
ovem
ent
optio
nal e
xtra
D
efin
ition
: I
tem
tha
t m
ay b
e ad
ded
to a
ren
tal a
t ex
tra
char
ge if
the
ren
ter
so c
hoos
es
E
xam
ple:
One
-way
ren
tal,
fuel
pre
-pay
men
t, a
dditi
onal
insu
ranc
es,
fittin
gs (
child
sea
ts,
sate
llite
nav
igat
ion
syst
em,
ski r
ack)
S
ourc
e: C
RIS
G [
“opt
iona
l ext
ra”]
rent
al a
ctua
l ret
urn
date
/ tim
e
C
once
pt T
ype:
rol
e
D
efin
ition
: d
ate/
time
whe
n re
nted
car
of re
ntal
is r
etur
ned
to E
U-R
ent
rent
al r
eque
sts
car
mod
el
S
ynon
ymou
s Fo
rm:
car
mod
el is
req
uest
ed for
ren
tal
N
eces
sity
: E
ach
rent
al r
eque
sts
at m
ost
one
car
mod
el.
P
ossi
bilit
y: T
he c
ar m
odel
req
uest
ed for
a r
enta
l cha
nges
bef
ore
the
act
ual p
ick-
up d
ate/
time
of t
he r
enta
l.
N
eces
sity
: N
o ca
r m
odel
req
uest
ed for
a r
enta
l cha
nges
aft
er t
he
act
ual p
ick-
up d
ate/
time
of t
he r
enta
l
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
37
JBos
s Rul
es
<ru
le n
ame=
"Fre
e Fi
sh F
ood
Sam
ple"
>
<
para
met
er id
entif
ier=
"car
t">
<
java
:cla
ss>
org.
droo
ls.e
xam
ples
.jav
a.pe
tsto
re.S
hopp
ingC
art<
/jav
a:cl
ass>
<
/par
amet
er>
<
para
met
er id
entif
ier=
"ite
m">
<
java
:cla
ss>
org.
droo
ls.e
xam
ples
.jav
a.pe
tsto
re.C
artI
tem
</j
ava:
clas
s>
<
/par
amet
er>
<
java
:con
ditio
n>ca
rt.g
etIt
ems(
"Fi
sh F
ood
Sam
ple"
).s
ize(
) =
= 0
</j
ava:
cond
ition
>
<
java
:con
ditio
n>ca
rt.g
etIt
ems(
"Fi
sh F
ood"
).s
ize(
) =
= 0
</j
ava:
cond
ition
>
<
java
:con
ditio
n>ite
m.g
etN
ame(
).eq
uals
( "G
old
Fish
" )<
/jav
a:co
nditi
on>
<
java
:con
sequ
ence
>
Sys
tem
.out
ln(
"Add
ing
free
Fis
h Fo
od S
ampl
e to
car
t" )
; ca
rt.a
ddIt
em(
new
org
.dro
ols.
exam
ples
.jav
a.pe
tsto
re.C
artI
tem
( "F
ish
Food
Sam
ple"
, 0.
00 )
);
dr
ools
.mod
ifyO
bjec
t( c
art
);
<
/jav
a:co
nseq
uenc
e>
</r
ule>
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
38
Even
t-Bus
Ø B
asis
of m
any
inte
ract
ive
appl
icat
ion
fram
ewor
ks (
XW
indo
ws,
Jav
a AW
T, J
ava
Info
Bus
, ..
..)
Ø S
ee d
esig
n pa
tter
n O
bser
ver
with
Cha
nge
Man
ager
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Eve
ntB
us (M
edia
tor)
Sub
ject
S
ubje
ct
Sub
ject
Obs
erve
r Jr
ules
-ba
sed
Obs
erve
r
EC
A-r
ule
base
d O
bser
ver
39
Arc
h. S
tyle
: W
orkf
low
-Bas
ed S
yste
ms
Ø A
wor
kflo
w d
escr
ibes
the
act
ions
on
cert
ain
even
ts a
nd c
ondi
tions
Ø F
orm
ed b
y a
deci
sion
ana
lysi
s, d
escr
ibed
by
ECA r
ules
Ø I
nste
ad o
f a
data
-flo
w g
raph
as
in p
ipe-
and-
filte
r sy
stem
s, o
r a
cont
rol-
flow
gra
ph a
s in
cal
l-ba
sed
syst
ems
Ø A
con
trol
-and
-dat
a flo
w g
raph
ste
ers
the
syst
em
Ø T
he d
ata-
flow
gra
ph c
onta
ins
cont
rol-
flow
inst
ruct
ions
(if,
whi
le,
..)
Ø T
his
wor
kflo
w g
raph
is s
imila
r to
a U
ML
activ
ity d
iagr
am,
with
pip
es a
nd
switc
h no
des
Ø O
ften
tra
nsac
tion-
orie
nted
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Wor
kflo
w
Filte
r
Pipe
? ?
40
App
licat
ion
Dom
ains
of W
orkf
low
Arc
hite
ctur
es
Ø B
usin
ess
soft
war
e Ø T
he b
ig fra
mew
orks
of SAP,
Peo
ples
oft,
etc
. al
l org
aniz
e w
orkf
low
s in
co
mpa
nies
Ø P
rodu
ctio
n pl
anni
ng s
oftw
are
Ø W
eb s
ervi
ces
are
desc
ribe
d by
wor
kflo
w la
ngua
ges
(BPE
L)
Ø M
ore
in c
ours
e “C
ompo
nent
-bas
ed S
oftw
are
Engi
neer
ing”
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
41
Arc
h. S
tyle
: Arc
hite
ctur
al S
tyle
of Com
mun
icat
ing
Sta
te
Mac
hine
s
Ø P
roce
sses
can
be
mod
eled
with
sta
te m
achi
nes
that
rea
ct o
n ev
ents
, pe
rfor
m a
ctio
ns,
and
com
mun
icat
e Ø M
odel
che
ckin
g ca
n be
use
d fo
r va
lidat
ion
of s
peci
ficat
ions
Ø L
angu
ages
:
Ø
Este
relle
, Lo
tos,
SD
L Ø
UM
L an
d its
sta
tech
arts
Ø
Het
ereg
enou
s Ric
h Com
pone
nts
(HRC)
Ø
EAST-
AD
L
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
42
App
licat
ions
Ø P
roto
col e
ngin
eering
Ø A
utom
atic
der
ivat
ion
of t
ests
for
sys
tem
s Ø T
elec
omm
unic
atio
n so
ftw
are
Ø E
mbe
dded
sof
twar
e
Ø
In c
ars
Ø
In p
lane
s Ø
In r
obot
s
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
43
20.3
.3)
Dat
a-O
rien
ted
Des
ign
Ø D
ata-
orie
nted
des
ign
is g
roup
ed a
roun
d a
inpu
t/ou
tput
/inn
er d
ata
stru
ctur
e Ø o
r a
lang
uage
for
a d
ata
stru
ctur
e (r
egul
ar e
xpre
ssio
ns,
finite
aut
omat
a,
cont
ext-
free
gra
mm
ars,
...
)
Ø T
he a
lgor
ithm
of th
e sy
stem
is is
omor
phic
to
the
data
and
can
be
derive
d fr
om t
he d
ata
Ø I
nput
dat
a (i
nput
-dat
a dr
iven
des
ign)
Ø O
utpu
t da
ta (
outp
ut-d
ata
driv
en d
esig
n)
Ø I
nner
dat
a
Ø D
ivid
e: fin
ding
sub
-dat
a st
ruct
ures
Ø C
onqu
er:
grou
ping
of da
ta a
nd a
lgor
ithm
s to
mod
ules
Ø E
xam
ple:
Ø J
acks
on S
truc
ture
d Pr
ogra
mm
ing
(JSP)
Ø E
TL p
roce
ssin
g in
info
rmat
ion
syst
ems
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Wha
t do
es t
he d
ata
look
like
?
44
Dat
a-Fl
ow S
tyle
: Reg
ular
Bat
ch P
roce
ssin
g (E
TL P
roce
ssin
g)
Ø R
egul
ar B
atch
Pro
cess
ing
is a
spe
cific
bat
ch-p
roce
ssin
g st
yle.
In
such
an
appl
icat
ion,
reg
ular
dom
ains
are
pro
cess
ed:
Ø R
egul
ar s
trin
g la
ngua
ges,
reg
ular
act
ion
lang
uage
s, o
r re
gula
r st
ate
spac
es
Ø T
he for
m o
f th
e da
ta c
an b
e de
scribe
d by
a
Ø R
egul
ar e
xpre
ssio
n, r
egul
ar g
ram
mar
, st
atec
hart
, or
JSP
diag
ram
tre
e
Ø O
ften
tra
nsac
tion-
orie
nted
Ø E
xam
ple:
Ø R
ecor
d pr
oces
sing
in b
ank
and
busi
ness
app
licat
ions
: Ø B
ank
tran
sact
ion
soft
war
e Ø D
atab
ase
tran
sact
ion
soft
war
e fo
r bu
sine
ss
Ø B
usin
ess
repo
rt g
ener
atio
n fo
r m
anag
ers
(con
trol
ling)
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
45
Arc
h. S
tyle
: Rep
osito
ry S
yste
ms
(Dat
a Bas
e Sys
tem
s)
Ø P
roce
ssin
g is
dat
a-or
ient
ed
Ø F
ree
coor
dina
tion
the
com
pone
nts,
can
be
com
bine
d w
ith c
all-
base
d st
yle
or p
roce
ss-s
tyle
Ø O
ften
als
o st
ate-
orie
nted
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Rep
osito
ry
Rea
d/w
rite
46
Exam
ple:
Rep
osito
ry S
tyle
in a
Com
pile
r
Ø T
he a
lgor
ithm
s ar
e st
ruct
ured
alo
ng t
he s
ynta
x of
the
pro
gram
s Ø T
he D
esig
n Pa
tter
n “V
isito
r” s
epar
ates
dat
a st
ruct
ures
fro
m
algo
rith
ms
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Lexi
cal
Ana
lyse
r
Pars
er
Sem
antic
A
naly
sis
Opt
imiz
er
Tran
sfor
mat
ion
Phas
e
Cod
e ge
nera
tor
Rep
osito
ry
47
Rep
osito
ry S
tyle
in a
Int
egra
ted
Dev
elop
men
t En
viro
nmen
t
Ø I
DE
stor
e pr
ogra
ms,
mod
els,
tes
ts in
the
ir r
epos
itory
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Lexi
cal
Ana
lyse
r
Pars
er
Sem
antic
A
naly
sis
Ref
acto
ring
Pret
ty
Prin
ter
Rep
osito
ry
Dia
gram
Vi
sual
izer
Uni
t Tes
ting
48
Info
rmat
ion
Sys
tem
s –
Que
ries
on
a Rep
osito
ry
Ø A
lgor
ithm
s ar
e st
ruct
ured
alo
ng t
he r
elat
iona
l dat
a Ø D
ata
war
ehou
se a
pplic
atio
ns p
rovi
de q
uery
ing
on m
ultid
imen
sion
al
data
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Que
ry 1
Q
uery
O
ptim
izer
Que
ry3
Rep
osito
ry
Que
ry 2
49
Bla
ckbo
ard
Sty
le
Ø T
he b
lack
boar
d is
an
activ
e re
posi
tory
(i.e
., a
n ac
tive
com
pone
nt)
and
coor
dina
tes
the
othe
r co
mpo
nent
s Ø b
y ev
ent
notif
icat
ion
or c
all
Ø D
omin
ant
styl
e in
exp
ert
syst
ems
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Bla
ckbo
ard
Rea
d/w
rite
50
Fire
/trig
ger
20.3
.4)
Com
pone
nt-B
ased
Des
ign
(Str
uctu
re-O
rien
ted
Des
ign)
Ø F
ocus
is o
n th
e H
AS-A
(PA
RT-
OF)
rel
atio
n Ø F
ocus
is o
n pa
rts,
i.e.
, on
an
hier
arch
ical
str
uctu
re o
f th
e sy
stem
Ø D
ivid
e: fin
ding
sub
com
pone
nts
(par
ts)
Ø C
onqu
er:
grou
ping
of co
mpo
nent
s to
larg
er c
ompo
nent
s
Ø E
xam
ple:
Ø D
esig
n w
ith a
rchi
tect
ural
lang
uage
s (s
uch
as E
AST-
AD
L)
Ø D
esig
n w
ith c
lass
ical
com
pone
nt s
yste
ms
(com
pone
nts-
off-
the-
shel
f, C
OTS
),
such
as
CO
RBA o
r Aut
oSAR
Ø H
owev
er,
man
y co
mpo
nent
mod
els
exis
t Ø S
epar
ate
cour
se “
Com
pone
nt-b
ased
sof
twar
e en
gine
erin
g (C
BSE)
”
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Wha
t ar
e th
e co
mpo
nent
s (p
arts
) of
the
sys
tem
, th
eir
stru
ctur
e, a
nd t
heir
rel
atio
ns?
51
20.3
.5)
Obj
ect-
Orien
ted
Des
ign
Ø D
ata
and
actio
ns a
re g
roup
ed in
to o
bjec
ts,
and
deve
lope
d to
geth
er
Ø F
ocus
is o
n th
e is
-a a
nd t
he b
ehav
es-l
ike
rela
tion
Ø A
par
t of
the
sys
tem
is li
ke o
r be
have
s lik
e an
othe
r pa
rt (
sim
ilarity
)
Ø D
ivid
e: fin
ding
act
ions
with
the
ir e
nclo
sing
obj
ects
Ø C
onqu
er:
grou
p ac
tions
to
obje
cts
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Wha
t ar
e th
e "o
bjec
ts"
of t
he s
yste
m?
Wha
t ar
e th
e ac
tion
s an
d at
trib
utes
of
the
obje
cts?
52
Obj
ect-
Orien
ted
Des
ign
Met
hods
Ø C
RC c
ards
(ST-
1)
Ø V
erb
subs
tant
ive
anal
ysis
(ST-
1)
Ø C
olla
bora
tion-
base
d de
sign
and
CRRC (
ST-
1)
Ø U
se-C
ase
Rea
lizat
ion
Ana
lysi
s
Ø B
ooch
met
hod
Ø R
umba
ugh
met
hod
(OM
T)
Ø (
Rat
iona
l) U
nifie
d Pr
oces
s (R
UP,
or
Uni
fied
Met
hod)
Ø u
ses
UM
L as
not
atio
n
Ø H
iera
rchi
cal O
O M
etho
d (H
OO
D)
Ø O
ften
, O
O is
use
d, w
hen
the
real
wor
ld s
houl
d be
sim
ulat
ed
(sim
ulat
ion
prog
ram
s)
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
53
Arc
h. S
tyle
: O
bjec
t-O
rien
ted
Cal
l-Bas
ed A
rchi
tect
ural
Sty
le
Ø C
ontr
ol flo
w is
sym
met
ric
(cal
ls a
nd r
etur
ns)
Ø C
ontr
ol flo
w is
not
fix
ed (
dyna
mic
arc
hite
ctur
e vi
a po
lym
orph
ism
) Ø C
ontr
ol-f
low
can
be
sequ
entia
l or
para
llel
Ø D
ata-
flow
can
be
para
llel t
he c
all (
push
-bas
ed s
yste
m)
or
antip
aral
lel,
i.e.,
par
alle
l to
the
retu
rn (
pull-
base
d sy
stem
)
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Cla
ss
Sub
clas
s
Sub
clas
s
Syst
em
call
retu
rn
call
retu
rn
call
retu
rn
disp
atch
54
Arc
h. S
tyle
: O
bjec
t-O
rien
ted
Proc
ess
Sys
tem
s (A
ctor
Sys
tem
s)
Ø O
bjec
t-or
ient
ed s
yste
ms
can
be p
aral
lel
Ø A
ctor
s ar
e pa
ralle
l com
mun
icat
ing
proc
esse
s Ø P
roce
sses
tal
k di
rect
ly t
o ea
ch o
ther
Ø U
nstr
uctu
red
com
mun
icat
ions
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
55
Arc
h. S
tyle
: Pr
oces
s Tr
ee S
yste
ms
(UN
IX-L
ike)
Ø P
roce
sses
(pa
ralle
l obj
ects
) ar
e or
gani
zed
in a
tre
e
Ø a
nd t
alk
only
to
thei
r de
scen
dant
s
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
56
20.3
.6)
Tran
sfor
mat
iona
l Des
ign
Ø W
e st
art
with
an
initi
al,
abst
ract
des
ign
that
mee
ts t
he
requ
irem
ents
Ø
The
cont
ext
mod
el a
nd t
he t
op-l
evel
arc
hite
ctur
e Ø T
he im
plem
enta
tion
is a
chie
ved
by a
n ite
rativ
e tr
ansf
orm
atio
n pr
oces
s, s
tart
ing
from
an
initi
al d
esig
n Ø R
efin
emen
t-ba
sed
deve
lopm
ent
Ø R
efac
toring
-bas
ed d
evel
opm
ent
uses
sym
met
ry o
pera
tions
(re
fact
orin
gs)
Ø S
emi-
auto
mat
ical
ly d
eriv
ing
a fin
al d
esig
n Ø D
ivid
e: fin
d st
eps
from
the
initi
al t
o th
e fin
al d
esig
n Ø C
onqu
er:
chai
n th
e st
eps
Ø N
ote:
thi
s de
sign
met
hod
is o
rtho
gona
l to
the
othe
rs,
beca
use
it ca
n be
com
bine
d w
ith a
ll of
the
m
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
How
sho
uld
I tr
ansf
orm
the
cur
rent
des
ign
to a
n be
tter
ver
sion
and
fi
nally
, the
impl
emen
tati
on?
57
Tran
sfor
mat
iona
l Ref
inem
ent-
Bas
ed D
esig
n
Ø W
ide
spec
tru
m la
ng
uag
es u
ses
rule
-bas
ed t
rans
form
atio
n sy
stem
s an
d tr
ansf
orm
atio
n pl
anne
rs
Ø T
his
star
ts a
t th
e re
quirem
ent
spec
ifica
tion
and
refin
es (
unde
r pr
oofs
of
corr
ectn
ess)
exp
ress
ive
expr
essi
ons
to e
xecu
tabl
e pr
ogra
ms
(sem
anti
c-p
rese
rvin
g r
efin
emen
t)
Ø T
he s
eman
tic
refi
nem
ents
are
ref
acto
ring
s w
hich
low
er e
xpre
ssiv
e ex
pres
sion
s to
low
-lev
el
Ø S
eman
tics
can
be p
rove
n in
diff
eren
t fo
rms,
e.g
., w
ith H
oare
logi
c, D
ynam
ic
logi
c, o
r de
nota
tiona
l sem
antic
s Ø S
eman
tic-p
rese
rvin
g re
finem
ent
does
not
nee
d te
stin
g, b
ecau
se a
ll de
rive
d pr
ogra
ms
are
corr
ect
by c
onst
ruct
ion.
The
met
hod
is a
lso
a fo
rmal
met
hod.
Ø E
xam
ples
•
CIP
-L (
Mun
ich
Uni
vers
ity)
•
F. L
. Bau
er,
M.
Bro
y, R
. G
natz
, W
. H
esse
, B.
Krieg
-Brü
ckne
r, H
. Pa
rtsc
h, P
. Pe
pper
, an
d H
. W
össn
er.
Tow
ards
a w
ide
spec
trum
lang
uage
to
supp
ort
prog
ram
spe
cific
atio
n an
d pr
ogra
m d
evel
opm
ent.
SIG
PLAN
Not
ices
, 13
(12)
15-2
4, 1
978.
•
SET
L (J
. Sch
war
tz,
New
Yor
k U
nive
rsity
) •
KID
S (
Kes
trel
inst
itute
), V
DM
, Z,
B,
Even
t-B
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
58
Ref
acto
ring
-Bas
ed E
xtre
me
Prog
ram
min
g (X
P)
Ø M
ore
info
rmal
and
incr
emen
tal p
roce
ss:
Extr
eme
Prog
ram
min
g (X
P)
• Bas
ed o
n re
fact
orin
gs for
str
uctu
ral i
mpr
ovem
ents
, bu
t no
t pa
rtic
ular
ly for
lo
wer
ings
•
Ref
acto
ring
can
be
supp
orte
d by
ref
acto
ring
too
ls
• Ev
ery
requ
irem
ent
is im
plem
ente
d an
d te
sted
in s
epar
atio
n •
Con
tinuo
us t
estin
g an
d co
ntin
uous
inte
grat
ion
(tes
t-dr
iven
dev
elop
men
t)
• Cus
tom
er is
invo
lved
(cu
stom
er-d
rive
n de
velo
pmen
t)
• Pe
rman
ent
revi
ew w
ith p
air
prog
ram
min
g
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
59
Mod
el-D
rive
n Arc
hite
ctur
e as
Tra
nsfo
rmat
iona
l Des
ign
Met
hod
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent D
omai
n m
odel
, R
equi
rem
ents
spe
cific
atio
n
Pla
tform
Inde
pend
ent M
odel
(PIM
)
Pla
tform
Spe
cific
Mod
el (P
SM
)
Pla
tform
Spe
cific
Impl
emen
tatio
n (P
SI)
(Cod
e)
Mod
el m
appi
ngs
Com
puta
tiona
lly In
depe
nden
t M
odel
(CIM
)
60
20.3
.7)
Gen
erat
ive
Des
ign
Ø (
aka
Gen
erat
ive
Prog
ram
min
g)
Ø S
peci
fy t
he s
olut
ion
in
Ø a
"fo
rmal
met
hod"
, a
spec
ifica
tion
lang
uage
Ø a
tem
plat
e w
hich
is e
xpan
ded
(gen
eric
pro
gram
min
g)
Ø I
n U
ML,
whi
ch is
gen
erat
ed in
to c
ode
by a
CASE
tool
Ø G
ener
ate
a so
lutio
n w
ith a
gen
erat
or t
ool t
hat
plan
s th
e so
lutio
n Ø
Plan
ning
the
com
posi
tion
of t
he s
olut
ion
from
com
pone
nts
Ø
Syn
thes
izin
g th
e so
lutio
n
Ø D
ivid
e: d
epen
ds o
n th
e sp
ecifi
catio
n la
ngua
ge
Ø C
onqu
er:
also
Ø F
ully
gen
erat
ive
prog
ram
min
g is
cal
led
Aut
omat
ic P
rogr
amm
ing
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
How
can
I d
eriv
e th
e im
plem
enta
tion
fro
m t
he d
esig
n au
tom
atic
ally
?
61
Gen
erat
ive
Spe
cific
atio
ns
Ø D
evel
opin
g a
spec
ficat
ion
in o
ne o
f th
ese
lang
uage
s is
sim
pler
tha
n w
ritin
g th
e co
de
Ø G
ram
mar
-orien
ted
deve
lopm
ent
(gra
mm
arw
are)
Ø Fi
nite
aut
omat
a fr
om r
egul
ar g
ram
mar
s Ø La
rge
finite
aut
omat
a fr
om m
odal
logi
c (m
odel
che
cker
s)
Ø Pa
rser
s fr
om C
onte
xt-f
ree
gram
mar
s Ø Ty
pe c
heck
ers,
typ
e in
fere
ncer
s fr
om A
ttribu
te g
ram
mar
s Ø Ty
pe c
heck
ers
and
inte
rpre
ters
fro
m N
atur
al s
eman
tics
Ø O
ptim
izer
s fr
om g
raph
rew
rite
sys
tem
s (s
ee c
hapt
er o
n G
RS)
Ø F
eatu
re-o
rien
ted
deve
lopm
ent
(FO
DA):
spe
cify
fea
ture
tre
es a
nd d
eriv
e th
e co
mpo
nent
s fr
om t
hem
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Spe
cific
atio
n
Mod
el –
Gra
mm
ar –
Log
ic -
Feat
ureT
ree
Cod
e
62
Aut
omat
ic P
rogr
amm
ing
Ø I
n au
tom
atic
pro
gram
min
g, a
pla
nner
pla
ns a
way
to
gene
rate
d th
e co
de fro
m t
he r
equi
rem
ent
spec
ifica
tions
•
Usi
ng a
pat
h of
tra
nsfo
rmat
ions
Ø A
.P.
is g
ener
ativ
e, a
nd t
rans
form
ativ
e, a
nd for
mal
met
hod.
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
63
1.3.
8 M
odel
-Drive
n Sof
twar
e D
evel
opm
ent
(MD
SD
)
Ø M
DSD
ble
nds
Tran
sfor
mat
iona
l and
Gen
erat
ive
desi
gn
Ø M
odel
s
• re
pres
ent
part
ial i
nfor
mat
ion
abou
t th
e sy
stem
•
Are
not
direc
tly e
xecu
tabl
e •
But
can
be
used
to
gene
rate
par
ts o
f th
e co
de o
f a
syst
em
Ø M
odel
-drive
n ar
chite
ctur
e (M
DA®
) of
OM
G)
blen
ds
Tran
sfor
mat
iona
l Des
ign
and
Gen
erat
ive
Des
ign
Ø S
ee a
lso
Cha
pter
“M
odel
-Drive
n Arc
hite
ctur
e”
Ø M
DA n
eeds
Asp
ect-
Orien
ted
Mod
elin
g (m
odel
wea
ving
)
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
64
20.3
.9)
Form
al M
etho
ds
Ø A
for
mal
met
hod
is a
des
ign
met
hod
that
Ø H
as a
for
mal
(m
athe
mat
ical
) sp
ecifi
catio
n of
the
req
uire
men
ts
Ø D
evel
ops
a fo
rmal
spe
cific
atio
n of
the
des
ign
Ø T
he d
esig
n ca
n be
ver
ified
aga
inst
the
req
uire
men
ts s
peci
ficat
ion
Ø A
for
mal
met
hod
allo
ws
for
prov
ing
a de
sign
cor
rect
Ø V
ery
impo
rtan
t fo
r sa
fety
-critic
al s
yste
ms
Ø F
orm
al m
etho
ds a
re o
rtho
gona
l to
the
othe
r m
etho
ds:
ever
y m
etho
d ha
s th
e po
tent
ial t
o be
for
mal
Ø I
mpo
rtan
t in
saf
ety-
criti
cal a
pplic
atio
n ar
eas
(pow
er p
lant
s, c
ars,
em
bedd
ed a
nd r
eal-
time
syst
ems)
Ø
Ex.
Petr
i net
s (s
epar
ate
chap
ter)
, B,
Z,
VD
M,
…
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
How
can
I p
rove
tha
t m
y de
sign
is c
orre
ct w
ith
rega
rd t
o th
e re
quir
emen
ts?
65
Che
cker
-Bas
ed S
yste
ms
Ø A
ch
ecke
r-b
ased
sys
tem
is fau
lt-to
lera
nt in
the
sen
se t
hat
for
ever
y co
mpo
nent
, a
chec
ker
exis
ts t
hat
chec
ks t
he c
orre
ctne
ss o
f an
app
licat
ion
Ø A
lso
calle
d a
mon
itor
Ø E
xam
ple:
Ver
ified
com
pile
rs,
faul
t-to
lera
nt 2
4/7
syst
ems
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
C
heck
er
Com
pone
nt
Che
cker
-Wra
pper
?
Err
or
Out
put
Inpu
t
66
Test
-Drive
n Arc
hite
ctur
e
Ø A
tes
t-d
rive
n s
yste
m m
aint
ains
with
eve
ry c
ompo
nent
a t
est
com
pone
nt
Ø T
he t
est
runs
prior
to
the
syst
em
Ø E
xam
ple:
TD
D (
Test
-Drive
n D
evel
opm
ent)
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
C
ompo
nent
Te
st
Com
pone
nt
Test
-Wra
pper
Err
or
Out
put
Inpu
t
67
Vot
ing
Arc
hite
ctur
es
Ø I
n a
votin
g fa
ult-
tole
rant
arc
hite
ctur
e, t
he r
un-t
ime
chec
ker
is a
m
ajor
ity v
oter
(qu
orum
) th
at c
ompa
res
the
resu
lts o
f se
vera
l in
stan
ces
of t
he c
ompo
nent
Ø E
xam
ple:
Spa
ce S
hutt
le
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Vote
r (Q
uoru
m)
Com
pone
nt
? C
ompo
nent
Com
pone
nt
68
20
.3.1
0 A
SP
ECT-
OR
IEN
TED
SO
FTW
AR
E D
ESIG
N
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
69
Deb
uggi
ng
asp
ect
Pers
iste
nce
aspe
ct
Alg
orith
m
Deb
uggi
ng a
spec
t Pe
rsis
tenc
e as
pect
Pers
iste
nce
aspe
ct
Deb
uggi
ng a
spec
t
Wea
ver-
Tool
Deb
uggi
ng a
spec
t
Arc
h. S
tyle
: Asp
ect-
Orien
ted
Sof
twar
e D
esig
n
Ø U
sual
des
ign
met
hods
hav
e on
e as
pect
of de
velo
pmen
t in
foc
us
(“ty
rann
y of
dec
ompo
sitio
n”)
Ø A
spec
t-or
ient
ed s
yste
ms
spec
ify d
iffer
ent
aspe
cts
of a
sys
tem
in
sepa
ratio
n (s
epar
atio
n of
con
cern
s)
Ø
The
slic
es a
re r
eint
egra
ted
by g
ener
ativ
e Asp
ect
Wea
vers
(Asp
ect/
J)
Ø
Mor
e in
cha
pter
s “A
spec
t-or
ient
atio
n”,
“Fea
ture
-bas
ed p
rodu
ct li
nes”
and
cou
rse
CBSE
TU D
resd
en,
Pro
f. U
. A
ßm
ann
7
0
20
.3.1
1 A
DA
PTI
VE
AR
CH
ITEC
TUR
ES
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
71
To b
e fil
led…
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
72
20
.4 A
RC
HIT
ECTU
RA
L S
TYLE
S S
PEC
IFIC
TO
LA
YER
S
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
73
Laye
red
Arc
hite
ctur
e
Ø T
he m
ost
gene
ral a
rchi
tect
ural
sty
le,
whi
ch c
an b
e co
mbi
ned
with
al
l oth
ers
are
laye
rs
Ø L
ayer
s ca
n be
com
bine
d w
ith m
any
othe
r st
yles
Ø I
ngre
dien
ts:
Ø
Con
nect
ors:
pro
cedu
re c
alls
or
stre
ams
Ø
Port
s: c
ompo
nent
inte
rfac
es
Ø
Con
trol
flo
w m
ostly
syn
chro
nous
Ø
Dat
a flo
w a
long
the
laye
rs a
nd t
he
call
grap
hs,
mos
tly s
ingu
lary
Ø
Dat
a- a
nd c
ontr
ol flo
w a
re is
omor
phic
Ø D
omin
atin
g st
yle
for
larg
e sy
stem
s
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
74
Exam
ple:
4-T
ier
Arc
hite
ctur
es in
GU
I-ba
sed
App
licat
ions
(BCED
)
Ø A
lrea
dy p
rese
nted
in S
T-1
Ø A
cycl
ic U
SES
Rel
atio
n,
div
ided
into
3 (
resp
. 4)
laye
rs:
Ø G
UI
(gra
phic
use
r in
terf
ace)
Ø M
iddl
e la
yer
(App
licat
ion
logi
c an
d m
iddl
ewar
e, t
rans
port
laye
r)
Ø D
ata
repo
sito
ry (
data
base
)
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Dat
a R
epos
itory
Lay
er (d
atab
ase,
mem
ory)
Mid
dlew
are
(mem
ory
acce
ss, d
istr
ibut
ion)
Gra
phic
al u
ser i
nter
face
App
licat
ion
logi
c (b
usin
ess
logi
c)
75
Exam
ple:
Ope
ratin
g Sys
tem
s
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Ker
nel
Use
r Spa
ce
UN
IX:
Ker
nel
Use
r Spa
ce
App
le-U
NIX
:
Mic
roke
rnel
(Mac
h)
Ker
nel
Use
r Spa
ce
Win
dow
s N
T/X
P:
Har
dwar
e A
bstra
ctio
n La
yer (
HA
L)
76
Arc
hite
ctur
al S
tyle
s Can
Be
Laye
r Spe
cific
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
77
Dat
a R
epos
itory
Lay
er (d
atab
ase,
mem
ory)
Mid
dlew
are
(mem
ory
acce
ss, d
istr
ibut
ion)
Gra
phic
al u
ser i
nter
face
App
licat
ion
logi
c (b
usin
ess
logi
c)
Even
t-bas
ed
Dat
a-ba
sed
Act
ion-
base
d
Func
tion
-ba
sed
Test
-dr
iven
Tr
ans-
form
ativ
e MDA
Asp
ect-
orie
nted
Form
al m
etho
ds
Dom
ain-
Spe
cific
Arc
hite
ctur
al S
tyle
s
Ø O
ften
an
appl
icat
ion
dom
ain
need
s its
ow
n st
yle,
its
refe
renc
e ar
chite
ctur
e Ø I
t's h
ard
to s
ay s
omet
hing
in g
ener
al a
bout
tho
se
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
78
Impo
rtan
t
Ø A
n ar
chite
ctur
al s
tyle
res
ults
fro
m a
spe
cific
dev
elop
men
t m
etho
d Ø F
unct
iona
l, m
odul
ar d
esig
n: c
all-
base
d st
yle
Ø A
ctio
n de
sign
: da
ta-f
low
sty
le,
wor
kflo
w s
tyle
, re
gula
r pr
oces
sing
, pr
oces
s tr
ees
Ø O
bjec
t-or
ient
ed d
esig
n: o
bjec
t-or
ient
ed c
all-
base
d sy
stem
s, c
lient
-ser
ver,
ac
tors
(pr
oces
s sy
stem
s)
Ø U
ses-
orie
nted
des
ign:
laye
red
syst
ems
Ø S
peci
fic la
yers
nee
d sp
ecifi
c st
yles
Ø R
elia
ble
syst
ems
need
spe
cific
sty
les
Ø T
he d
edic
ated
eng
inee
r kn
ows
whe
n to
app
ly w
hat
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
79
Sum
mar
y: M
ost
Impo
rtan
t Arc
hite
ctur
e Sty
les
Ø D
ata
flow
sty
les
Ø S
eque
ntia
l pip
e-an
d-fil
ter
Ø D
ata
flow
gra
ph/n
etw
ork
Ø W
orkf
low
sys
tem
s (m
ixed
with
co
ntro
l flo
w)
Ø C
all-
styl
e
Ø M
odul
ar s
yste
ms
Ø A
bstr
act
data
typ
es
Ø O
bjec
t-or
ient
ed s
yste
ms
Ø C
lient
/ser
vice
pro
vide
r
Ø H
iera
rchi
cal s
tyle
s Ø L
ayer
ed a
rchi
tect
ure
Ø I
nter
pret
er
Ø C
heck
er-b
ased
Arc
hite
ctur
es
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
Ø I
nter
actin
g pr
oces
ses
(act
ors)
Ø T
hrea
ds in
a s
hare
d m
emor
y Ø D
istr
ibut
ed o
bjec
ts
Ø E
vent
-bas
ed s
yste
ms
Ø A
gend
a pa
ralle
lism
Ø D
ata-
orie
nted
(Rep
osito
ry
arch
itect
ures
) Ø T
rans
actio
n sy
stem
s (d
ata
base
s)
Ø Q
uery
-bas
ed s
yste
ms
Ø B
lack
boar
d (e
xper
t sy
stem
s)
Ø T
rans
form
atio
n sy
stem
s (c
ompi
lers
) Ø G
ener
ativ
e sy
stem
s (g
ener
ator
s)
Ø D
ata
base
d st
yles
Ø C
ompo
und
docu
men
ts
Ø H
yper
text
-bas
ed
80
Law
of M
etho
d an
d Sty
le
Ø F
unct
iona
l and
act
ion
desi
gn è
cal
l-ba
sed
arch
itect
ural
sty
le o
r co
mpo
nent
-bas
ed s
tyle
Ø O
bjec
t-or
ient
ed d
esig
n è
obj
ect-
orie
nted
cal
l sty
le o
r ac
tor
styl
e Ø A
ctio
n-ba
sed
desi
gn (
with
dat
a-flo
w) è
dat
a-flo
w a
rchi
tect
ures
(p
ipe-
and-
filte
r ar
chite
ctur
es)
or E
CA s
yste
ms
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
A sp
ecifi
c de
sign
met
hod
lead
s to
a s
peci
fic a
rchi
tect
ural
sty
le
81
Ø A
spe
cific
app
licat
ion
dom
ain
need
s a
spec
ific
arch
itect
ural
sty
le,
and
due
to t
hat,
a s
peci
fic d
esig
n m
etho
d, e
.g.,
•
Embe
dded
sof
twar
e ne
eds
form
al m
etho
ds
• En
terp
rise
sof
twar
e ne
eds
wor
kflo
w-b
ased
sty
le
• In
form
atio
n sy
stem
s ne
ed r
epos
itory
sty
le
Whi
ch D
esig
n M
etho
d fo
r th
e Sat
ellit
e Rad
io?
Ø R
eal w
orld
obj
ects
mus
t be
sim
ulat
ed
Ø O
bjec
t-or
ient
ed d
esig
n?
Ø E
vent
s in
the
rea
l wor
ld
Ø E
vent
-con
ditio
n-ac
tion
base
d de
sign
?
Ø F
low
of da
ta fro
m t
he s
atel
lite
to t
he r
adio
to
the
user
Ø D
ata-
orie
nted
des
ign?
dat
a-flo
w a
rchi
tect
ure!
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
82
Wha
t H
ave
We
Lear
ned?
Ø T
here
is n
o si
ngle
“th
e w
ay t
o th
e sy
stem
” Ø E
very
pro
ject
has
to
find
its p
ath
empl
oyin
g an
app
ropr
iate
des
ign
met
hod
Ø T
he b
asic
des
ign
ques
tions
are
pos
ed o
ver
and
over
aga
in,
until
a
desi
gn is
fou
nd
Ø S
elec
t a
desi
gn m
etho
d Ø P
ose
the
desi
gn m
etho
d's
basi
c qu
estio
n Ø P
erfo
rm t
he d
esig
n m
etho
d's
proc
ess
Ø P
erfo
rm t
he d
esig
n m
etho
d's
step
s: e
labo
rate
, re
fine,
str
uctu
re,
chan
ge
repr
esen
tatio
n, .
..
Ø I
f pr
oces
s ge
ts s
tuck
, ch
ange
des
ign
met
hod
and
try
anot
her
one!
Ø A
rchi
tect
ural
sty
les
are
the
resu
lt of
a d
esig
n pr
oces
s Ø T
hey
give
us
a w
ay t
o ta
lk a
bout
a s
yste
m o
n a
rath
er a
bstr
act
leve
l Ø A
rchi
tect
ural
sty
les
can
be d
istin
guis
hed
by t
he r
elat
ion
of d
ata-
flow
and
co
ntro
l-flo
w (
para
llel v
s an
tipar
alle
l)
Ø .
. an
d th
e ty
pe o
f sy
stem
str
uctu
ring
rel
atio
n th
ey u
se
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
83
Wha
t is
run
ning
in P
art
III
– D
esig
n?
Ø P
rese
ntat
ion
of D
esig
n M
etho
ds w
ith N
otat
ions
, Pr
oces
ses,
H
eurist
ics
Ø P
rese
ntat
ion
of t
he D
evel
opm
ent
Focu
s Ø P
rese
ntat
ion
of r
esul
ting
Arc
hite
ctur
al S
tyle
s Ø P
rese
ntat
ion
of V
aria
bilit
y an
d Ex
tens
ibili
ty m
echa
nim
s, t
o pr
epar
e pr
oduc
t lin
e en
gine
erin
g
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
84
Why
Is
This
Im
port
ant?
(T
he E
ngin
eer's
Park
inso
n D
isea
se)
Ø D
on't
be d
isco
urag
ed a
bout
the
div
ersi
ty o
f th
is le
ctur
e. T
here
is
som
ethi
ng t
o w
in..
. Ø A
goo
d ob
ject
-orien
ted
desi
gner
is n
ot a
utom
atic
ally
a g
ood
soft
war
e en
gine
er
Ø A
sof
twar
e en
gine
er k
now
s a
larg
e to
olbo
x of
diff
eren
t m
etho
ds t
o be
abl
e to
cho
ose
the
righ
t m
etho
d!
Ø U
sual
ly,
peop
le s
tick
to t
he m
etho
ds in
whi
ch t
hey
have
bee
n ed
ucat
ed
Ø C
OBO
L pr
ogra
mm
ers
Ø I
mpe
rativ
e vs
fun
ctio
nal p
rogr
amm
ers
Ø O
bjec
t-or
ient
ed p
rogr
amm
ers
vs p
roce
dura
l pro
gram
mer
s
Ø D
o yo
u w
ant
to g
et s
tuck
? Ø Y
ou w
ill h
ave
a la
rge
adva
ntag
e if
you
are
open
-min
ded
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
85
20
.5 D
ESIG
N H
EUR
ISTI
CS
A
ND
BES
T P
RA
CTI
CES
Gen
eral
Str
ateg
ies
in D
esig
n Pr
oces
ses
Lite
ratu
re
Ø O
blig
ator
y Rea
ding
Ø [
Bro
oks]
Fre
derick
P.
Bro
oks
jr.
No
Silv
er B
ulle
t. E
ssen
ce a
nd A
ccid
ents
of
Sof
twar
e En
gine
erin
g. I
n [T
haye
r].
Won
derf
ul a
rtic
le o
n w
hat
soft
war
e en
gine
erin
g is
all
abou
t.
Ø O
ther
Lite
ratu
re
Ø [
Bud
gen]
Dav
id B
udge
n. S
oftw
are
Des
ign.
Add
ison
-Wes
ley.
Exp
ands
on
the
Bud
gen
pape
r. P
rett
y sy
stem
atic
. Ø [
Endr
es/R
omba
ch]
A.
Endr
es,
D.
Rom
bach
. A H
andb
ook
of s
oftw
are
and
syst
ems
engi
neer
ing.
Em
pirica
l obs
erva
tions
, la
ws
and
theo
ries
. Add
ison
-Wes
ley.
Ver
y go
od c
olle
ctio
n of
sof
twar
e la
ws.
Nic
e!
TU D
resd
en,
12
.05
.20
09
Seb
asti
an R
ich
ly
Folie
87
von
18
Des
ign
Heu
rist
ics
Bro
ok's
Par
adox
on
Sof
twar
e Bea
uty
Exci
ting
Ø U
nix
Ø O
S/2
Ø A
PL
Ø P
asca
l Ø M
odul
a Ø A
lgol
68
Ø S
mal
ltalk
Use
ful,
but
unex
citin
g Ø M
VS/3
70
Ø M
S-D
OS
Ø C
obol
Ø P
L/1
Ø F
ortr
an
Ø A
lgol
60
Ø p
hp
Nic
e sy
stem
s ar
e of
ten
too
late
in th
e m
arke
t
... b
e th
e fir
st o
r the
sec
ond
bird
!
TU D
resd
en,
12
.05
.20
09
Seb
asti
an R
ich
ly
Folie
88
von
18
Des
ign
Heu
rist
ics
Heu
rist
ic:
Lazy
or
Eage
r D
esig
n
Ø I
n ca
se o
f a
diffic
ult
desi
gn d
ecis
ion
Ø (
whe
n el
abor
atin
g, r
efin
ing,
ref
acto
ring
or
chan
ging
rep
rese
ntat
ion)
Ø …
def
er it
(la
zy d
esig
n)
Ø It
erat
ive
Sof
twar
e de
velo
pmen
t m
etho
ds s
uch
as E
xtre
me
Prog
ram
min
g Ø …
dec
ide
it (e
ager
des
ign)
Ø …
anti
cip
ate
furt
her
deve
lopm
ents
in t
he d
esig
n (a
ntic
ipat
ory
desi
gn)
Ø T
ime-
boxe
d de
sign
: (S
CRU
M X
P pr
oces
s)
Ø D
o ite
ratio
ns in
fix
ed t
ime-
slot
s (1
mon
th)
Ø F
ix r
equi
rem
ents
onl
y fo
r on
e tim
e-sl
ot
Ø H
ave
it ru
nnin
g un
der
all c
ircu
mst
ance
s Ø U
pdat
e re
quirem
ents
with
cus
tom
er a
fter
the
tim
e-sl
ot
TU D
resd
en,
12
.05
.20
09
Seb
asti
an R
ich
ly
Folie
89
von
18
Des
ign
Heu
rist
ics
Prep
are
for
Evol
utio
n: G
row
Liv
ing
Sof
twar
e
Ø B
uild
dev
elop
men
t: “
build
, no
t w
rite
” [B
rook
s]
Ø S
oftw
are
is a
livi
ng t
hing
Ø L
ehm
an's
first
law
of so
ftw
are
evol
utio
n: “
A s
yste
m t
hat
is u
sed
will
be
chan
ged”
Ø I
ncre
men
tal d
evel
opm
ent
Ø “
grow
, no
t bu
ild s
oftw
are”
[Bro
oks]
Ø R
efac
toring
s an
d re
finem
ent
shou
ld a
lway
s be
pos
sibl
e
TU D
resd
en,
12
.05
.20
09
Seb
asti
an R
ich
ly
Folie
90
von
18
Des
ign
Heu
rist
ics
Heu
rist
ic:
Div
ide
and
Con
quer
Str
ateg
y
Ø D
ivid
e et
impe
ra (
from
Ale
xand
er t
he G
reat
) Ø d
ivid
e: p
robl
ems
into
sub
prob
lem
s (s
impl
ifica
tion)
Ø T
o fin
d so
lutio
ns in
ter
ms
of t
he a
bstr
act
m
achi
ne w
e ca
n em
ploy
. W
hen
this
map
ping
is
com
plet
e, w
e ca
n co
nque
r Ø c
onq
uer
: so
lve
subp
robl
ems
(hop
eful
ly e
asie
r)
Ø c
omp
ose
(mer
ge)
: co
mpo
se t
he c
ompl
ete
so
lutio
n fr
om t
he s
ubso
lutio
ns
Ø R
euse
of pa
rtia
l sol
utio
ns is
pos
sibl
e
(the
n th
e tr
ee is
a d
ag)
Ø W
here
do
we
begi
n?
Ø S
tepw
ise
refin
emen
t (t
op-d
own)
Ø A
ssem
blag
e (b
otto
m-u
p)
Ø D
esig
n fr
om t
he m
iddl
e (m
iddl
e-ou
t, y
o-yo
)
......
.
......
.
......
.
??
TU D
resd
en,
12
.05
.20
09
Seb
asti
an R
ich
ly
Folie
91
von
18
Des
ign
Heu
rist
ics
Ste
pwis
e Ref
inem
ent
(Top
-Dow
n, C
lass
ic D
ivid
e-an
d-Con
quer
)
Ø P
oin
twis
e re
fin
emen
t Ø F
rag
men
t re
fin
emen
t Ø C
ontr
ol r
efin
emen
t (o
pera
tion
refin
emen
t)
Ø W
e gu
ess
the
solu
tion
of t
he p
robl
em in
te
rms
of a
hig
her-
leve
l abs
trac
t m
achi
ne
Ø W
e re
fine
thei
r op
erat
ions
unt
il th
e gi
ven
ab
stra
ct m
achi
ne is
rea
ched
Ø D
ata
refi
nem
ent
Ø W
e m
ay a
lso
refin
e th
e da
ta s
truc
ture
s
of t
he a
bstr
act
mac
hine
Ø S
ynta
ctic
ref
inem
ent
does
not
res
pect
se
man
tics
of t
he o
rigi
nal m
odel
Ø S
eman
tic
refi
nem
ent
prov
es c
onfo
rman
ce o
f th
e re
fined
mod
el t
o th
e or
igin
al m
odel
, i.e
. w
heth
er it
is s
eman
tical
ly e
quiv
alen
t or
ric
her
than
the
origi
nal m
odel
Ø D
isad
vant
age:
Ø W
e m
ight
nev
er r
each
a r
ealiz
atio
n Ø O
ften
"w
areh
ouse
sol
utio
ns"
are
de
velo
ped,
tha
t ar
e in
appr
opriat
e ...
....
......
.
......
.
??
Folie
92
von
18
Des
ign
Heu
rist
ics
Ste
pwis
e Con
stru
ctio
n (B
otto
m-u
p)
Ø I
n th
is c
ase
we
star
t w
ith a
giv
en a
bstr
act
mac
hine
and
Ø a
ssem
ble
mor
e co
mpl
ex o
pera
tions
of a
hi
gher
-lev
el a
bstr
act
mac
hine
Ø o
r as
sem
ble
the
mor
e co
mpl
ex d
ata
st
ruct
ures
Ø G
ood:
Ø A
lway
s re
alis
tic
Ø A
run
ning
par
tial s
olut
ion
Ø B
ad:
Ø D
esig
n m
ight
bec
ome
clum
sy s
ince
gl
obal
pic
ture
was
not
tak
en in
to a
ccou
nt
......
.
......
.
......
.
??
Folie
93
von
18
Des
ign
Heu
rist
ics
Mid
dle
out
Ø F
ix s
ome
subp
robl
ems
in
the
mid
dle
and
solv
e th
em
by r
efin
emen
t Ø T
hen
wor
k yo
ur w
ay u
p Ø O
ften
avo
ids
the
disa
dvan
tage
s
of t
op-d
own
and
bott
om-u
p Ø F
indi
ng le
mm
as in
a
mat
hem
atic
al p
roof
is s
imila
r
......
.
......
.
......
.
??
Folie
94
von
18
Des
ign
Heu
rist
ics
Heu
rist
ic:
Use
Hie
rarc
hies
and
Red
ucib
le G
raph
s
Ø T
rees
, tr
ees,
tre
es
Ø D
ags
(direc
ted
acyc
lic g
raph
s)
Ø C
an b
e la
yere
d Ø R
educ
ible
gra
phs
Ø C
an b
e la
yere
d to
o, o
n ea
ch
laye
r th
ere
are
cycl
es
Ø E
very
nod
e ca
n be
ref
ined
in
depe
nden
tly a
nd a
bstr
acts
th
e lo
wer
leve
ls
TU D
resd
en,
12
.05
.20
09
Seb
asti
an R
ich
ly
Folie
95
von
18
Des
ign
Heu
rist
ics
Heu
rist
ics
on S
ize
Ø L
imit
your
self
to a
sm
all n
umbe
r of
item
s Ø N
ever
use
mor
e th
an 5
item
s
Ø o
n a
page
Ø o
n a
slid
e Ø o
n an
abs
trac
tion
leve
l of a
spec
ifica
tion
or m
odel
Ø K
ISS (
keep
it s
impl
e st
upid
) Ø R
emov
e al
l sup
erflu
ous
thin
gs,
mak
e it
fit o
n 1
page
Ø S
impl
ifica
tion
take
s a
long
tim
e “I
did
n't
have
the
tim
e to
mak
e it
shor
ter”
Ø E
inst
ein:
"M
ake
it as
sim
ple
as p
ossi
ble,
but
not
sim
pler
!”
Ø S
teph
en K
ing:
“W
hen
I th
ink,
I a
m r
eady
, I
usua
lly h
ave
to r
educ
e ab
out
30%
fat
fro
m m
y bo
ok."
Ø A
bstr
actio
n is
neg
lect
ion
of u
nnec
essa
ry d
etai
l Ø F
ocus
at
one
prob
lem
at
a tim
e an
d to
for
get
abou
t ot
hers
Ø D
ispl
ay o
nly
esse
ntia
l inf
orm
atio
n Ø C
hang
e re
pres
enta
tion
if de
velo
pmen
t st
rate
gy c
hang
es
Ø T
his
lead
s to
des
ign
met
hods
or
deco
mpo
sitio
n m
etho
ds
TU D
resd
en,
12
.05
.20
09
Seb
asti
an R
ich
ly
Folie
96
von
18
Des
ign
Heu
rist
ics
Heu
rist
ics
on A
bstr
actio
n
Ø S
epar
atio
n of
Con
cern
s (S
oC)
Ø D
iffer
ent
conc
epts
sho
uld
be s
epar
ated
so
that
the
y ca
n be
spe
cifie
d in
depe
nden
ly
Ø D
imen
sion
al s
peci
ficat
ions
: sp
ecify
fro
m d
iffer
ent
view
poin
ts
Ø I
f se
para
ted,
the
n co
ncer
ns c
an b
e va
ried
inde
pend
ently
Ø E
xam
ple
of S
oC:
Sep
arat
e Po
licy
and
Mec
hani
sm
Ø M
echa
nism
: Th
e w
ay h
ow t
o te
chni
cally
rea
lize
a so
lutio
n Ø P
olic
y: T
he w
ay h
ow t
o pa
ram
eter
ize
the
real
izat
ion
of a
sol
utio
n Ø I
f se
para
ted,
the
n po
licy
and
mec
hani
sm c
an b
e va
ried
inde
pend
ently
TU D
resd
en,
12
.05
.20
09
Seb
asti
an R
ich
ly
Folie
97
von
18
Des
ign
Heu
rist
ics
But
Con
side
r Bro
oks
Law
..
The
cent
ral q
uest
ion
in d
esig
n is
how
to
impr
ove
on t
he s
oftw
are
art
cent
ers
- as
it a
lway
s ha
s be
- o
n
peop
le.
[Bro
oks]
TU D
resd
en,
12
.05
.20
09
Seb
asti
an R
ich
ly
Des
ign
Heu
rist
ics
Folie
98
von
18
Ref
lect
ions
on
Bro
oks'
Law
Ø E
duca
tion
of p
eopl
e is
ver
y im
port
ant!
Ø H
owev
er,
the
differ
ence
s ar
e no
t m
inor
- t
hey
are
rath
er li
ke t
he d
iffer
ence
s be
twee
n Sal
ieri a
nd M
ozar
t.
Ø S
tudy
aft
er s
tudy
sho
ws
that
the
ver
y be
st d
esig
ners
pro
duce
str
uctu
res
that
are
fas
ter,
sm
alle
r, s
impl
er,
clea
ner,
and
pro
duce
d w
ith le
ss e
ffor
t.
Ø G
reat
des
igne
rs a
nd g
reat
man
ager
s ar
e bo
th v
ery
rare
Ø H
owev
er,
Fark
as' L
aw:
Figh
ting
help
s!
Ø F
arka
s, a
pro
min
ent
trom
bone
tea
cher
, no
ticed
tha
t th
e m
ost
tale
nted
pu
pils
did
n't
mak
e it
Ø I
nste
ad,
the
mid
dle-
clas
s su
rviv
ed t
hat
lear
ned
how
to
wor
k ha
rd
TU D
resd
en,
12
.05
.20
09
Seb
asti
an R
ich
ly
Folie
99
von
18
Des
ign
Heu
rist
ics
Oth
er L
itera
ture
Ø S
imon
Sin
gh.
Ferm
ats
letz
er S
atz.
Die
abe
nteu
erlic
he G
esch
icht
e ei
nes
mat
hem
atis
chen
Rät
sels
. dt
v.
Ø G
ute-
Nac
ht-G
esch
icht
e üb
er F
erm
at‘s
jah
rhun
dert
ealte
s Rät
sel.
Erkl
ärt
den
kom
pliz
iert
en B
ewei
s And
rew
Wile
s‘ für
Nic
ht-E
xper
ten.
Zum
Ver
sche
nken
! (G
aloi
s in
klus
ive.
.)
Ø U
hren
arith
met
ik.
Ellip
tisch
e G
leic
hung
en.
Mod
ulfo
rmen
.
Ø M
erke
: G
enie
ent
steh
t au
s vi
el,
viel
Fle
iss
(man
bea
chte
das
Er
lebn
is W
iles‘
bei
der
Kor
rekt
ur d
es B
ewei
ses!
) Ø W
enn
selb
st s
olch
gro
sse
Mat
hem
atik
er F
ehle
r in
ihre
n Bew
eise
n pr
oduz
iere
n...
.. k
eine
Ang
st v
or g
ross
en A
ufga
ben.
..
Ø E
xcel
lenc
e is
the
res
ult
of e
norm
ous
corr
ectio
n..
TU D
resd
en,
12
.05
.20
09
Seb
asti
an R
ich
ly
Folie
10
0 v
on 1
8
Des
ign
Heu
rist
ics
The
End
Ø I
n th
e fo
llow
ing,
we
will
see
sev
eral
exa
mpl
es fo
r se
lect
ed d
esig
n m
etho
ds
Ø W
ith t
he c
once
pts
of s
impl
e gr
aph-
base
d m
odel
s, w
e ca
n se
e co
mm
on c
once
pts
in a
ll of
the
m
TU D
resd
en,
Pro
f. U
. A
ßm
ann
D
evel
opm
ent
10
1