24小时热门版块排行榜    

查看: 274  |  回复: 5
当前主题已经存档。
当前只显示满足指定条件的回帖,点击这里查看本话题的所有回帖

p1801321

金虫 (著名写手)

[交流] 【求助】重金求数段的翻译(有效期2天)

分割线以上20币,分隔线以下10币,谢绝翻译软件的翻译,不给币
Having developed a conceptual architecture, we then used LDM to reveal extra dependencies. An unexpected benefit of building a DSM with a cleaner architecture was that by going over the extra dependencies and expanding the hierarchy in the matrix, it was possible to very quickly to find the cause of the extra dependencies. In fact, in the DSM in Figure 18, four regions (numbered 1-4) are supposed to be free of dependencies; regions 1-3 because they are in the top-right diagonal and would represent cycles, and region 4 because (as shown in the architectural diagram) it represents the independence of the UI from the Server Analysis of the dependencies in these regions suggests the remediation necessary in order to clean up the architecture:
Region 1 represents dependencies from what is known as the project’s data-model. These modules form the building blocks of the entire system and therefore are not supposed to depend on any other modules. Examining the extra dependencies shows that they are caused mainly due to static methods. These dependencies show the practice in Haystack of not using any particular criteria in placing non-instance based helper methods. While these methods should ideally be
in another module, they are in the data model for ease in finding and using them.
• Region 2 represents dependencies from the project’s inference engines. These engines have resulted from the evolution and extension of two smaller inference engines one primarily used in the server and the other in the UI. Over
time, they had evolved to have different strengths and are being used inconsistently in the code. A redesign of this component was already being contemplated. The extra dependencies in the grid highlight this need.
Region 3 and 4 dependencies represent minor project inconsistencies. Examining these extra dependencies shows that resolving each one needs a small and local design decision, and have likely been caused because of the absence of an architectural diagram in the past.
-----------------------------------------------------------------------------------------------
Beyond helping identify these regions of extra dependencies, the rearchitected DSM yielded significant benefits. Just from the DSM it is apparent that subsystems for server extensions and user interface (ozone) extensions have no major dependencies on the rest of the system. This means that they can be changed or newer extensions added without significant risk to the system. In fact, a
large number of these components are now hidden subsystems of the server and the user interface. Similarly, we were able to create a new abstraction for inference
engines (region 2) with the DSM without actually writing code. While we still needed to refactor the code so that the new abstractions would be properly layered and easily available, the DSM allowed us to define the details of the new abstraction by
allowing us to examine all dependencies that would arise from creating the abstraction.

[ Last edited by p1801321 on 2009-12-8 at 17:24 ]

» 猜你喜欢

已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

syh9288

捐助贵宾 (著名写手)

海军上将

★ ★ ★ ★ ★ ★
p1801321(金币+6,VIP+0):thanks 12-8 23:37
我们已经开发出了概念性架构,然后用LDM展示其外延链接。构建一种结构更加简洁的DSM
(设计性结构矩阵)有着一个出乎意料的的优点:通过检查外延链接及扩大层级矩阵,这使得我们能快速地定位外延链接的原因。事实上,如在图18所示的DSM中,4个区域(标记有1-4)被认为没有相关性:1-3区域,因为他们在对角线的上端和表现出周期性,以及区域4,如架构图中所示,它表示:为了使架构更加简洁在这些区域中源自于链接代理服务器分析的UI(用户界面)的独立性需要必须的修复。
人生如梦,一樽还酹江月
3楼2009-12-08 21:46:42
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖
查看全部 6 个回答

zcy86520

铁杆木虫 (著名写手)

★ ★ ★
p1801321(金币+3,VIP+0):谢谢参与,不如楼下的兄弟翻译的到位 12-8 23:37
分隔线以下的:
除了帮助确定这些额外的依赖区域,重新设计的用电需求管理产生了重大利益。刚刚从帝斯曼很明显,对服务器扩展和用户界面(臭氧)扩展子系统对系统的其余部分没有大的依赖性。这意味着它们可以改变或者没有重大风险的系统较新的扩展补充。事实上,这些组件大批人现在隐藏的服务器子系统和用户界面。同样,我们能够创造一个没有实际编写代码的推理引擎(第2区与DSM)的新的抽象层。虽然我们仍然需要重构代码,以便将新的抽象层次,适当地用,帝斯曼使我们定义使我们能够检查所有的依赖关系,将产生创造了新的抽象抽象的细节
2楼2009-12-08 17:51:34
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

syh9288

捐助贵宾 (著名写手)

海军上将

★ ★ ★ ★ ★ ★ ★ ★ ★ ★
p1801321(金币+10,VIP+0):thanks 12-8 23:38
区1表示被称为计划数据模式的链接。这些模块是组成整个系统的构建模块,因而不被认为依赖于任何其它模块。检查外延链接显示他们主要是由静态方法引起的。在解释基于助方法的非实例中,这些链接显示了在不用任何特殊标准的堆积方式中的实际运用。虽然这些方法理念上应该是在另一个模块,他们在数据模型中是为了使得查找和使用更加便捷。
人生如梦,一樽还酹江月
4楼2009-12-08 22:19:51
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖

syh9288

捐助贵宾 (著名写手)

海军上将

★ ★ ★ ★ ★ ★ ★ ★ ★ ★
p1801321(金币+10,VIP+0):thanks 12-8 23:38
区2表示项目推理引擎的链接。这些引擎因两个较小推理引擎的演化和扩展而产生,一个主要用于服务器,另一个主要用于用户界面(UI)。
  随着时间的变化,他们已经进化到拥有不同的优势并且在代码的使用也不一致。这种组件的重新设计先已被纳入考虑的范围。在格图中的外延链接更突出了这种需要。
  区3和区4中的链接有着细微的项目不连续性。通过这些外延链进行接检验表明;解决每一个链接问题需要一个局域的设计理念,并且可能因为没有架构图而被引起。
人生如梦,一樽还酹江月
5楼2009-12-08 22:40:49
已阅   回复此楼   关注TA 给TA发消息 送TA红花 TA的回帖
普通表情 高级回复 (可上传附件)
信息提示
请填处理意见