GME_教程lesson1

2021年1月30日发(作者：alcazar)

Long Tutorial lessons

1

Lesson 1 Creating the First Paradigm and Model

Introduction to Lesson 1

In

this

first

lesson,

you

will

configure

GME

to

allow

the

drawing

of

basic

diagrams

for

a

specific

application

domain.

Later

lessons

will

show

you

how

to

extend the project with more advanced features.

The

application

domain

throughout

this

tutorial

is

that

of

networking

systems:

routers, networked computers, and other devices, as well as the connections between

these devices. Your application domain is probably quite different from this one, but

looking

at

a

specific

domain

as

an

example

is

probably

the

best

way

to

help

you

identify

the

main

steps

of

modeling

and

learn

the

techniques

of

customizing

GME.

There

are

some

sections

that

describe

the

details

of

networking,

and

serve

as

Those sections appear in red italics

. If you are not familiar with networking, you may

find them irrelevant or hard to understand. Feel free to skim or even skip them, and

focus on the rest of the tutorial, which deals with the more practical aspects of GME.

This

tutorial

provides

enough

information

to

give

the

average

user

a

well- developed

understanding

of

GME,

without

having

to

rely

on

extra

documentation. Of course, to gain a deeper understanding of GME and to learn all of

its features, you will need to consult the GME Users Manual as well.

1.1

GME and the modeling process

GME

is

a

generic,

configurable

modeling

environment.

In

other

words,

the

configuration of GME is not an option; it is the first step that must be taken before

anything

meaningful

can

be

done

with

it.

The

configuration

process

itself

is

also

a

form of modeling - the modeling of a modeling process. This is called metamodeling.

The output of the metamodeling process is a compiled set of rules, the paradigm, that

configures GME for a specific application domain. (GME

是一个通用的 、可配置的

建模环境。换句话说，

GME

的配置不是一个选项，而是用

GME

做任何有意义

< p>
的事情之前必须做的第一个步骤。配置过程本身也是一种类型的建模

-

对于建模

处理过程的建模。

这叫做元建模< /p>

metamodeling

。

元建模过程 的输出是一系列编译

过的规则，范例，它们为特定的应用域配置

GME

。

)

Does everyone need to

be a metamodeling expert to use GME? Of course not.

Normally,

only

a

few

people

participate

in

the

metamodeling

process.

By

contrast,

there

are

numerous

users

of

the

configured

environment.

The

users

do

not

need

to

know

much

about

metamodeling;

what

they

see

is

a

graphical

tool

with

editing

capabilities

that

is

already

customized

for

their

domain.

The

main

advantage

of

model-integrated computing is that the work of the users is guided by the modeling

environment.

This

tutorial

is

written

for

those

who

are

interested

in

metamodeling

and

customization.

The

majority

of

the

tutorial

deals

specifically

with

metamodeling

issues;

examples

of

customized

environments

are

usually

explained

in

much

less

detail.

Remember,

the

metamodeling

environment

itself

is

also

based

on

GME.

So

while acting as a metamodeler, you are also a user of the customized environment of

GME, based on its meta- paradigm. By the time you begin to build models, you will

have already had plenty of practice simply by using the metamodeling environment!

1.1.1

问题说明

Problem specification

The

first

thing

that

a

metamodeling

expert

needs

is

a

specification

(or

at

least

some

vague

idea)

of

the

modeling

application

to

be

implemented.

This

typically

comes in a natural-language description.

（一个元建模老手需 要的第一件事是一个

要实现的建模应用的说明书或者至少是粗略的想法。

）

As an example, the proposed

network modeling application is described below:

want

to

create

networking

diagrams.

The

diagrams

will

contain

routers.

Each router has several router-ports.

We also

have hosts

(e.g. servers). We want

to

connect router-ports and hosts to form networks.

following:

or

[i.e.

we

have

Cisco

and

Linux

routers].

Ports

are

identified

by

portname,

IP

address,

speed

(in

kbps),

and

type

(which is either

are

identified

by

name

and

address

(for

now,

they

all

have

a

single

Ethernet port). Networks are identified by a network address and a netmask.

1.1.2

建模方法

The modeling approach

We

now

have

a

description

of

the

concepts

to

be

modeled

using

the

GME

metamodeling environment. The modeling technique we are using is the well-known

and widely accepted Unified Modeling Technology, especially a specific subset called

UML Class Diagrams.

（我们使用的建模 技术是众所周知和广泛接受的统一建模

技术，特别是一个叫做

U ML

类图的特殊子集。

）

The

most

basic

step

in

the

metamodeling

process

consists

of

determining

two

things:

the

entities

used by

the

model,

and

the

relations

between

them.

Information

used to identify and qualify certain entities and relations will be assigned to them as

attributes. Metamodeling, in a nutshell, is the mapping of specification concepts onto

entities, relations and attributes.

（元建模最基本的步骤由确定

2

件事情决定：模

型使用的实体，

以及实体间的相互关系。

用于确定和限定特定实体和相互关系的

信息将作为属性指定给它们。

简单的说，

元 建模就是将特定概念映射到实体、

相

互关系和属性。

< p>
）

We have been given the specification

（

1.1.1

< p>
）

. Now let's describe the modeling

paradigm:

is also an entity, one that is always associated with (i.e. contained by) a router.

The

model

will

contain

two

different

types

of

relations:

one

that

represents

connections between a network and a router-port or host (

association

), and one that

links

a

router

to

its

router- ports

(

containment

).

The

main

difference

between

containment and association is that contained entities are

and the contained entity cannot exist without its container.

Certain entities have specific attributes. All of the attributes are mentioned in the

specification, such as

Luckily for us, the networking specification in this example turned out to be brief

and fairly precise. It often takes much more to extract the essence of the customer's

ideas.

1.1.3 Generic modeling concepts in GME

GME supports a set of generic modeling concepts that are variations of the UMT

entities, relationships, and attributes. The differences reside partly in the nomenclature

and

partly

in

additional

features,

semantics,

and

constraints

which

are

expected

to

guide and simplify the metamodeling process.

Atoms

are

a

basic,

limited

type

of

entity.

The

name

indicates

that

it

has

no

internal

structure (i.e. contained objects). Every feature of an atom that can be represented in a model is

contained in the atom's name, attributes, and the relations it participates in.

Models, the second generic type of entity, are very similar to atoms. The main difference lies in

the

ability

of

models

to

contain

atoms,

other

models,

and

other

types

of

objects.

Thus,

models

have

internal

structure.

When

viewed

together,

they

form

tree-like

containment

hierarchies

of

entities. Models can be opened, showing a diagram of their internal structure (Fig 1.2).

* Please note that - mostly for historical reasons - the term model has a double meaning in GME:

it denotes either the full description of something (the artifact of a modeling project) or a model

entity located somewhere in a containment hierarchy of entities.