模型到代码：无神秘空隙的代码生成实践

Models

Code

需积分: 9 29 浏览量更新于2024-07-17 收藏 18.36MB PDF 举报

身份认证购VIP最低享 7 折!

领优惠券(最高得80元）

资源详情

资源推荐

About the Authors

Leon Starr has been developing real-time distributed and embedded

software with object-oriented, executable models since 1984. His models

have been used in fighter jets, factory material transport control systems,

ultrasound diagnostic and cardiac pacing systems, gas chromatography

and semiconductor wafer inspection systems, video postproduction

systems, and networked military battle simulators. He has taught

numerous courses on executable systems and data modeling to systems

engineers and software developers worldwide through his company Model

Integration, LLC (modelint.com) based in San Francisco, California. He

is the author of How to Build Shlaer-Mellor Object Models, How to Build

Class Models, Executable UML: A Case Study, and assorted papers at

uml.org and modeling-languages.com. He regularly assists project teams

who model complex requirements and generate code from those models

for challenging hardware and software platforms.

Andrew Mangogna has developed a number of open source tools,

including pycca, which we use in this book to specify Executable UML

models and translate them into code. He has also worked extensively

in the medical device community. For more than 30 years, Andrew has

been a hands-on builder of embedded software systems. He has worked

in application areas ranging from laboratory instrumentation, remote

data collection, and video special effects, to implantable medical devices.

Andrew has always had a special interest in applying more formal

techniques to the challenge of engineering software to create systems in

a cost-effective manner with demonstrable quality. Trained in the basics

of modeling by Stephen Mellor himself, he has successfully applied

executable modeling techniques and model translation to many projects

and has written several tools to help automate the translation process.

With a keen interest in technology and a practical realization of the

benefits of modeling, he has a mastery of mapping models to appropriate

implementation technology to obtain high-quality software systems.

xvii

Acknowledgments

Any substantial project always depends on the generous help of others. Friends, colleagues, and a host of

experiences have provided valuable service to Models to Code.

We all wish to thank Ed Seidewicz, our technical reviewer, for his many insightful and detailed

comments throughout the text. His eye for detail rooted out quite a few potential and devilishly subtle errors.

He passed many of our “tests” to see whether he was actually reading this stuff. We especially appreciate

his patience with us, as we are well aware that our opinions and experience are often at odds with current

practice in the greater UML community.

I want to express my heartfelt thanks to friends and colleagues who expressed early enthusiasm for

this project. This includes many of my colleagues at SAAB Aerospace here in Sweden. At the Por Que No?

taqueria in Portland, Oregon in March of 2016, Dan George and I had some amazing tacos, and I was

encouraged by his infectious comments and enthusiasm. Back here in Linköping, Sweden, my present home

base, I enjoyed many beers, laughs, and technical discussions with Nils Paulsson of SAAB at The Bishops

Arms pub. He also contributed helpful comments on early drafts (in both the writing and beer sense of the

word). Thanks also to my Scandinavian “agent” and good friend, Christer Andersson, for equal measures of

encouragement and pestering me to get this project over with so I could get back to doing “real” work. My

longtime friend and business partner, Michael M. Lee, has served many a quality margarita on my recurring

visits back to the Bay Area in California. I would like to thank the staff at Babbettes Kafferie in Linköping for

letting me use a corner of their cafe as my second office and keeping me well supplied with coffee, morning

buns, and warm conversation during the dark and cold Swedish winter days. Most important, though, I owe

a massive debt of gratitude to my wife, Kristina, who set me up with a beautiful home office and forbade me

from getting a new Xbox until the book was complete and for putting up with my writing throughout our

summer wedding plans. Along with Andrew’s superb project management skills, you can thank her the most

for this book making it to the shelves in 2017.

—Leon Starr

The many discussions over coffee and beer of modeling and its foundations with Paul Higham were

fundamental to a broader understanding of how all the pieces fit together. There were many people, such as

Melanie Gurunathan, Cary Campbell, and Thomas Brennan-Marquez, who actually used pycca to deliver

real products and whose feedback was essential. Finally, without the patience and support of my wife,

Marjorie Lane-Mangogna, it is hard to envision how this book could ever have happened. Few would have

the patience to deal with the blah, blah, blah chatter that surrounds someone immersed in writing on a

technical subject.

—Andrew Mangogna

xix

Foreword

I published my first book on requirements analysis and modeling back in 1997. It was based on my

experience working with teams to build complex, real-time, model-driven software. Like so many other

books on software modeling, I left the process of transforming the models into code as an exercise for the

student. Ha!

In fact, on actual projects, we were able to make this transformation, but it certainly wasn’t easy.

Worthwhile, yes. The models proved themselves again and again as being worth the effort. But the code

translation problem has never been trivial. We always required a small team dedicated to the translation task

running in parallel with the modeling effort. Note that this was not a waterfall approach, as modeling and

design were being accomplished simultaneously.

Over time, the tools and technology to bridge the path from models to code improved. Unfortunately,

they were locked up in proprietary solutions and expensive licensed products. This put me in an awkward

position when extolling the benefits of modeling to younger engineers. Getting them interested in the

benefits of modeling and a more promising style of model execution and code generation was easy. I would

then be confronted with the inevitable question, “That’s sounds great; how do I get started with this stuff?”

I then had to concede that, while you might toy around with the ideas on your own, you really needed to get

attached to a project that could fund the required tooling. Consequently, I didn’t want to write another book

without giving the reader a real path, any path, that took the models into running code without requiring the

purchase of any expensive tools.

Andrew and I have worked together on and off on various projects over the years. One of his early

translation systems extracted models from a tool called BridgePoint at the time (now xtUML) and

transformed those models directly into assembly language. Because of the specialized CPU, no sufficiently

optimal C compiler was available. It was a proprietary solution for a particular embedded device company,

but it was a compelling proof of concept. Abstract models could be compacted and reorganized directly to

the machine instruction level and run on specialized and limited resource hardware to yield the precise

behavior specified by the models.

Andrew has since moved on to build a number of open source translation systems of which one, pycca,

is used in this book. I eventually realized that these open source tools were the key to explaining how to

bridge the gap from models to code while providing an accessible solution to any interested party.

Thus began our latest book project. After coaxing Andrew to take part, and enticing the primary author

of Executable UML and Agile signatory, Stephen Mellor, to join in, we had the perfect mix of expertise.

As you read along, you may find that this is not an easy subject, and may require some hard work to

make it through. The reward is the ability to take executable models that solve real problems through to an

efficient implementation without destroying the models in the process. I want you to know that you are not

alone in this journey, and we’re more than happy to help if you have questions or run into obstacles. Feel

free to use our contact page at the end of the book. Happy reading!

—Leon Starr

Linköping, Sweden

March 15, 2017

xxi

xxii

■ Foreword

In conversations with colleagues who share a similar outlook about software development, the topic

invariably turns to why the practice of software development using models and translation is not more

widespread. Many reasons are offered, but usually the conversation trails off into consternation over an

industry that claims to want well-engineered software juxtaposed to the repeated practices that do not

deliver.

For my part, I have concluded that modeling and translation are hard because abstraction is hard

and requires practice to master. One need only look at the facial contortions of a young teenager when

initially exposed to the idea of variables in beginning algebra to understand that we are creatures first of the

immediate and concrete. Gone are the comforts of explicit numbers, and suddenly a new set of rules is in

place to cope with the idea that an arbitrary symbol, which doesn’t look anything like a numeral, can now

range over many values. The look is similar to that of a beginning programming student struggling to fix

firmly the difference between a variable and the value contained in the variable when the same symbol is

used for both and only context provides the distinction. Even though I have successfully produced software

systems using modeling and translation for many years, there are times when looking at pages of boxes and

arrows that it is hard to see how the gap between symbols on a page and running software will ever be filled.

It looks like a jump across the Grand Canyon to be attempted only by an extreme stunt rider equipped with a

magic, rocket-propelled vehicle.

The young algebra student, by working through many specific problems, does come to grasp the

underlying abstractions. Reasoning, and sometimes wrestling, with the concrete yields up the abstract.

In this book, we undertake to illustrate general principles of translating models into code by examining

specific examples. We practice a specific way to model requirements and translate to a specific type of target

platform using a specific programming language. We hold tightly to our tagline of “no mysterious gaps” in

our attempt to show, at each step along the way, exactly how the translated software is produced and runs.

We show lots of models and lots of code, confident in the awesome power of an individual to grasp the

abstractions embedded in the specifics.

There is a danger that some readers will look at the details of the platform or translation technique

and conclude that, because the specifics don’t match their interests or preferences, the idea of translating

models into programs is itself not generally applicable. In that case, we will have failed as authors to provide

the vision of model translation as a tangible engineering approach. But I am confident that there are readers

who will realize that symbols on a diagram can be made tangible as a running program (as tangible as

software ever is) and that obtaining code by translating models looks more like driving over a bridge than

jumping the Colorado river.

—Andrew Mangogna

Nipomo, California

March 15, 2017

剩余317页未读，继续阅读

处女座程序员的朋友

粉丝: 347
资源: 104

模型到代码：无神秘空隙的代码生成实践

Models to Code: With No Mysterious Gaps

Models to Code With No Mysterious Gaps 无水印原版pdf

codes.pdf

Evaluating Large Language Models Trained on Code.pdf

AttributeError: Model of type <class 'gensim.models.keyedvectors.KeyedVectors'> can't be loaded by <class 'gensim.models.word2vec.Word2Vec'>

ModuleNotFoundError: No module named 'torchvision.models.video.io'

org.apache.ibatis.binding.BindingException: Invalid bound statement (not found): com.xh.models.account.mapper.UserMapper.selectUserEmail

timm.models.layers.drop

加载一个word2vec模型时出现AttributeError: Model of type <class 'gensim.models.keyedvectors.KeyedVectors'> can't be loaded by <class 'gensim.models.word2vec.Word2Vec'>

models.Sequential.fit

models.chanpinxinxi.objects.filter啥意思

runtimeerror: failed to import transformers.models.bert.modeling_bert becaus

linux no module named 'models.ops.cython_bbox

gensim.models.ldamodel.LdaModel.save

ModuleNotFoundError: Nomodule named "timm.models.layers.helpers'

timm.models.layers.droppath

timm.models.layers.helpers

最新资源