所在位置:专家讲座 - 内容 论坛精华    
Rebecca Wirfs-Brock--"职责驱动设计"讲座

时间:2004年8月25日(周三)上午10:00-12:00

演讲人

Rebecca Wirfs-Brock,对象技术大师,面向对象设计技术的先驱。她在1990年出版的”Designing Object-Oriented Software”是最早的面向对象设计书籍之一(该书中译本《面向对象软件设计经典》已于2003年9月由电子工业出版社出版),书中清晰阐述了“类”、“职责”、“协作”这些我们今天耳熟能详的概念,提出职责驱动设计的方法。她在新书Object Design (Addison Wesley - November, 2002)中很好地融合了其对CRC卡、协作(Collaborations)和灵活性(Flexibility)这些主题的见解。

主持人
*liwei

幻灯下载
*A Brief Tour of Responsibility-Driven Design in 2004

现场图片

声音记录(和文字记录是互补的,不是重复)
*网盘下载(77M)
*ed2k下载(77M)

文字记录

(10:31:25)*doact与所有人说:too hard...
(10:31:47)123与所有人说:en,hard!
(10:32:01)*outmyth与所有人说:14
(10:32:12)*doact与所有人说:i can't remember the idea
(10:32:56)*doact与所有人说:and my english is not good
(10:33:54)123与所有人说:too late!
(10:35:36)*outmyth与所有人说:与实体驱动建模相比,基于职责建模更容易让我们一开始就处于正确的抽象级别上(概念视图),而非实现视图
(10:35:59)*umlchina2与所有人说:资料下载:http://webseminar3.unix160.cn4e.com/webseminar/RDD2004.ppt
(10:36:38)holibut与所有人说:请问:我听到的声音总是断断续续的,是这样的么?
(10:37:02)*outmyth与holibut说:有点,不过还可以
(10:37:23)*outmyth与所有人说:16
(10:37:39)holibut与所有人说:谢谢,概念视图,和logic view有何区别?
(10:38:13)*doact与*umlchina2说:问一下,这是现场还是录音啊?
(10:38:15)*outmyth与所有人说:Candidate,现在CRC的第一个C含义真多
(10:38:29)*umlchina2与所有人说:现场,live
(10:39:08)*doact与所有人说:why she don't speak in chinese?
(10:39:26)[*umlchina_seminar]与*doact说:嘉宾在美国,现在是晚上,在家中 )
(10:39:44)*doact与[*umlchina_seminar]说:ok,i see
(10:40:08)*outmyth与所有人说:Fowler谈论过三种关于抽象的级别(相信可能会有更多):概念视图(conceptual view),规约视图(specification view)和实现视图
(10:40:14)清风扬与所有人说:就是啊,听讲座的都是中国人
(10:40:26)*outmyth与所有人说:(implementation view)。好的CRC卡级描述相当于一个概念视图。
(10:41:23)[*umlchina_seminar]与holibut说:断续--网速? )
(10:41:39)*outmyth与所有人说:page 17
(10:43:17)*doact与[*umlchina_seminar]说:18
(10:43:23)*doact与所有人说:P18
(10:44:59)*outmyth与*doact说:right,
(10:46:13)*doact与所有人说:the statement is not clear,dose she can repeat it more clear?
(10:49:13)*outmyth与所有人说:20
(10:49:25)*doact与所有人说:just at a high level or not?
(10:50:11)*doact与所有人说:not one answer me ?
(10:50:44)[*umlchina_seminar]与*doact说:你可以向嘉宾提问题 )
(10:50:44)*umlchina2与所有人说:关于声音:是我们请专家放慢速度的,为了理解得更好
(10:51:35)*outmyth与所有人说:一个对象设计的概念级别视图描述的是关键的抽象。而有些人正好将关键抽象想象成“候选类(candidate classes)”的高级描述,更愿意从一个稍微不同的角度来考虑概念设计(conceptual
(10:51:59)*outmyth与所有人说:design)――我是在一个稍微不同的级别上考虑设计的。
(10:52:10)*doact与所有人说:high level ?
(10:53:00)*outmyth与所有人说:--from interview of Rebecca Wirfs-Brock(by design by object)
(10:54:25)*outmyth与所有人说:但是我想让这种可能性显现出来,通过在一开始就考虑角色和职责,然后第二个步骤是转向规约级别视图,将这些候选者映射成类和接口。
(10:54:37)hldream与所有人说:大家能看到视频么?>
(10:54:54)sunmoon9898与所有人说:no
(10:55:08)*outmyth与所有人说:一开始概念视图至少要是关键对象角色及其职责(较高级别的陈述)的精华所在。许多你最初建模的候选者将很有可能直接被映射成某种继承体系中的一个单独的类
(10:55:30)wangjingtoo与所有人说:有吗?
(10:55:33)*doact与所有人说:hi,Rebecca..."what it konw" is more important than "what it dose",right?
(10:56:00)*outmyth与所有人说:25
(10:56:25)*outmyth与所有人说:26
(10:56:35)[*umlchina_seminar]与*doact说:如果她没有回答,一会提问时间你再提一次 )
(10:58:07)hldream与[*umlchina_seminar]说:哦,是这样
(10:59:35)sunmoon9898与所有人说:真希望有人把中文大概译一下.
(10:59:47)*outmyth与所有人说:29
(10:59:51)*doact与所有人说:the conception is not clear...
(11:00:41)*outmyth与[*umlchina_seminar]说:一个面向对象应用是一组相互交互的对象。每个对象便是一个或多个角色的一个实现。
(11:00:59)*outmyth与所有人说:一个角色支持一组相关的(具有内聚性的)职责。
(11:01:07)*doact与所有人说:the MSG-Builder is the central?
(11:01:14)*outmyth与所有人说:一个职责便是执行一个任务或知道某种信息的一种责任(obligation)。
(11:01:26)*outmyth与所有人说:而且对象并不是在孤立地工作,它们与同一群体中的其他对象相互协作来执行整个应用程序的职责。
(11:03:18)*doact与所有人说:31
(11:04:59)*outmyth与所有人说:32
(11:05:07)*doact与所有人说:没意思,全是讲一些概念性的东西,不讲的没什么新意,如果能把“对象”这方面多扩充一下就好了
(11:06:15)*doact与所有人说:in my opinion:"object" is the most important point
(11:07:30)*doact与所有人说:PAGE 34
(11:08:46)*doact与所有人说:how about the protocol?
(11:09:09)*doact与所有人说:???
(11:09:17)sunmoon9898与所有人说:以实例讲解好一些
(11:09:46)*doact与所有人说:35
(11:11:07)*doact与所有人说:who is the administrator?
(11:11:53)*doact与所有人说:有这个讲座的录音下载没?我好多地方没听明白
(11:12:13)[*umlchina_seminar]与*doact说:正在录音,过后会上载的 )
(11:12:42)*doact与[*umlchina_seminar]说:到时候你把录音的地址发到什么地方啊?我们怎么能找到呢?
(11:13:03)*doact与[*umlchina_seminar]说:CSDN上会不会发?
(11:13:07)[*umlchina_seminar]与*doact说:录音以后在http://www.umlchina.com/Chat/Newindex1.htm下载
(11:13:28)*doact与[*umlchina_seminar]说:OK,Thanks
(11:15:53)*outmyth与所有人说:40
(11:20:00)*doact与[*umlchina_seminar]说:下次讲座是什么时候?是什么内容的?我们在哪能查到?
(11:20:21)[*umlchina_seminar]与*doact说:看UMLChina通知
(11:20:32)[*umlchina_seminar]与*doact说:首页
(11:20:44)*doact与[*umlchina_seminar]说:哦,但是我的UMLCHINA的讨论组为什么上不去了?
(11:21:06)*outmyth与所有人说:43
(11:22:07)[*umlchina_seminar]与*doact说:正常啊 )
(11:23:45)*lijinie与所有人说:why not use a picture to illustrate the types of control?
(11:24:21)*doact与[*umlchina_seminar]说:为什么你们的讨论组是挂在smiling.com上的啊?为什么不自己建个专门的呢?
(11:24:24)*outmyth与*lijinie说:page 41有
(11:25:20)*outmyth与*lijinie说:45
(11:26:08)*doact与所有人说:快结束了哦,都没听懂多少
(11:26:25)*lijinie与所有人说:i mean give a example of real world,not abstract relation ship,that might be very helpful!
(11:27:30)*doact与所有人说:agree with *lijinie
(11:29:25)*doact与所有人说:47
(11:31:14)*outmyth与*lijinie说:49
(11:32:38)*outmyth与所有人说:50
(11:35:37)*outmyth与所有人说:52
(11:38:29)*outmyth与所有人说:53
(11:39:32)*doact与所有人说:54
(11:39:57)*doact与所有人说:it's will be over soon
(11:40:34)*doact抓呀抓,把头皮都抓破了,也没有想出个所以然来。
(11:41:35)*doact与所有人说://郁闷的要死
(11:42:03)*doact与所有人说:听了半天,什么都没懂,就OVER了
(11:42:10)*outmyth与所有人说:as for the role sterotypes,are they sufficient source to find candidate class when
(11:42:18)*lijinie与所有人说:could u give us a real world application of the three control types?
(11:42:23)*xionganping与所有人说:
(11:42:26)[*umlchina_seminar]与所有人说:Thank you for the excellent talk
(11:42:39)wangjingtoo与*umlchina2说:谢谢。
(11:42:48)*xionganping与所有人说:thanks
(11:43:10)*lijinie与所有人说:generally speaking,it is good to clarify those abstract and important items!
(11:43:16)*xjmen与*umlchina2说:RDD 是怎么应对 需求的变化呢
(11:43:38)*xionganping与[*umlchina_seminar]说:不知有没有中文的翻译?包括讲话的内容
(11:43:45)*outmyth与所有人说:there is a pretty weak concept of role in uml1.x, and do you think what improvement in
(11:43:54)*outmyth与所有人说:uml2.x for role and role sterotypes?
(11:43:55)jjcao1231与所有人说:I think the talk is constructive. It remaind me sth, and give me some tip. So thanks
(11:43:56)wangjingtoo与*umlchina2说:CRC卡
(11:44:08)[*umlchina_seminar]与所有人说:马上回答问题
(11:44:31)[*umlchina_seminar]与所有人说:问题没有回答的可以重复再贴
(11:44:31)*doact与所有人说:and how about ROSE
(11:44:59)*outmyth与所有人说:are role sterotype sufficient source to find candidate class ?
(11:45:29)*doact与所有人说:see you
(11:45:33)*doact与所有人说:byebye
(11:45:47)*RebeccaWirfsBrock与所有人说:No. I use other techniques for finding objects that I did not talk about.
(11:45:58)*umlchina2与*RebeccaWirfsBrock说:Welcome back,Professor.
(11:46:08)*outmyth与所有人说:how to predict the flexibility of system as possible without fall into the trap of overengineering
(11:46:13)*RebeccaWirfsBrock与所有人说:Perhaps I can give you a talk about finding objects sometime
(11:47:14)*xjmen与*RebeccaWirfsBrock说:you know that requirment often is changed ,how do RDD deal with it ?
(11:49:31)*RebeccaWirfsBrock与所有人说:I am not sure that you can hear me very well. I will type
(11:49:31)*xjmen与*RebeccaWirfsBrock说:thanks
(11:49:45)*outmyth与所有人说:there is a pretty weak concept of role in uml1.x, and do you think what improvement in
(11:49:52)*outmyth与所有人说:uml2.x for role and role sterotypes?
(11:50:34)*RebeccaWirfsBrock与所有人说:UML 2 has the idea of role,and that is good, but it is only used to label an association
(11:50:55)*RebeccaWirfsBrock与所有人说:But UML 2 lets you define multiple stereotypes and that is good
(11:51:37)*outmyth与*RebeccaWirfsBrock说:thank you
(11:51:56)*outmyth与*RebeccaWirfsBrock说:how to predict the flexibility of system as possible without fall into the trap of overengineering
(11:52:06)*RebeccaWirfsBrock与所有人说:You are most welcome
(11:52:32)*xjmen与*RebeccaWirfsBrock说:can we read th context of this lecture in your book?
(11:52:39)*RebeccaWirfsBrock与所有人说:I think that flexibility must be planned for. It does not happen just because you use objects
(11:52:53)*xjmen与*RebeccaWirfsBrock说:"Designing Object-Oriented Software"?
(11:53:15)*RebeccaWirfsBrock与所有人说:Yes. Much of this material is in my new book, Object Design
(11:53:29)*xjmen与*RebeccaWirfsBrock说:thanks
(11:54:36)*umlchina2与*RebeccaWirfsBrock说:Dear Professor, how do you think the book of Eric Evens
(11:54:55)*umlchina2与*RebeccaWirfsBrock说:Sorry, Eric Evans
(11:55:07)*RebeccaWirfsBrock与所有人说:I like Eric's book and talk with him regularly. I think that domain modeling fits with rdd
(11:55:51)*outmyth与*RebeccaWirfsBrock说:I want to know do the trust regions mean package design in implementation view?
(11:55:57)*RebeccaWirfsBrock与所有人说:Finding objects from the domain is important (and then there are other ways to find objects, too)
(11:56:34)*RebeccaWirfsBrock与所有人说:I like to use packages as trust regions and that is very good to think that way...
(11:56:52)*outmyth与*RebeccaWirfsBrock说:thanks so much
(11:57:10)*RebeccaWirfsBrock与所有人说:Which means that you have a way of making good packages if you think about trust
(11:58:28)*outmyth与*RebeccaWirfsBrock说:I know everything in one package should be release and reuse together
(11:58:52)*outmyth与*RebeccaWirfsBrock说:so they are trusted each other
(11:59:19)*RebeccaWirfsBrock与所有人说:Yes. But you can think about trust when you use components and services and libraries, too
(11:59:44)*outmyth与*RebeccaWirfsBrock说:I get
(12:00:19)*RebeccaWirfsBrock与所有人说:Any more questions? If not, thank you so much.
(12:00:39)*outmyth与*RebeccaWirfsBrock说:thanks,professor
(12:00:47)*umlchina2与*RebeccaWirfsBrock说:21:00 in your time now :)
(12:00:52)*RebeccaWirfsBrock与所有人说:Please send me email at rebecca@wirfs-brock.com if you have any
(12:01:01)[*umlchina_seminar]与*RebeccaWirfsBrock说:Thank you! )
(12:01:09)*lijinie与所有人说:thank you,professor
(12:01:12)*xjmen与*RebeccaWirfsBrock说:ok