| 查看: 269 | 回复: 1 | |||
| 当前主题已经存档。 | |||
p1801321金虫 (著名写手)
|
[交流]
求一段翻译
|
||
|
请不要使用翻译软件,不给金币 The LDM tool uses a standard notion of dependency, in which a module A depends on a module B if there are explicit references in A to syntactic elements of B. Currently, a module is a Java class, but a more fine-grained analysis is possible. This simple but effective notion of dependency works well for understanding design dependencies, in which modifications to one module might affect another. It is less well suited to determining runtime properties (such as how failures can propagate between modules), which require a deeper static analysis. As in other dependency tools (such the Reflexion Model Tool [3]), the extraction of dependences can be decoupled from their analysis, so the techniques we describe here would apply equally well on top of more sophisticated static analyses. In the LDM tool, the DSM can be configured to display an ‘X’ for a dependency or to display a dependency strength representing the number of references between two modules that is responsible for their dependence. LDM offers DSM algorithms for partitioning; it does not currently offer automatic clustering. The default decomposition used by LDM is based on the code organization: the Java package structure. The matrix presented is hierarchical (as in Figure 4), and the algorithms and manual intervention can be applied at different levels. Users can edit the systems structure. They can reorder modules and partition by hand, and create, delete, and move subsystems to reflect their understanding of the architecture. Dependencies are automatically recalculated and re-aggregated as the structure is changed. [ Last edited by p1801321 on 2009-12-8 at 08:30 ] |
» 猜你喜欢
参与限项
已经有3人回复
假如你的研究生提出不合理要求
已经有7人回复
实验室接单子
已经有4人回复
全日制(定向)博士
已经有4人回复
对氯苯硼酸纯化
已经有3人回复
求助:我三月中下旬出站,青基依托单位怎么办?
已经有12人回复
不自信的我
已经有12人回复
所感
已经有4人回复
要不要辞职读博?
已经有7人回复
北核录用
已经有3人回复
syh9288
捐助贵宾 (著名写手)
海军上将
- 翻译EPI: 6
- 应助: 2 (幼儿园)
- 金币: 1598.6
- 散金: 543
- 红花: 1
- 帖子: 1046
- 在线: 475.4小时
- 虫号: 735351
- 注册: 2009-03-30
- 性别: GG
- 专业: 应用高分子化学与物理
★ ★ ★ ★ ★ ★ ★ ★ ★ ★
p1801321(金币+10,VIP+0):专业!十分感谢! 12-8 15:28
p1801321(金币+10,VIP+0):专业!十分感谢! 12-8 15:28
|
LDM工具中应用的是标准链接的概念,这种概念说的是:如果在A中存在对B语法元素的引用,那么A模块依赖于B模块。现在有一种模块用Java语言编写的,但是对这种模块进行更加精细纹理分析是可能的。这种简而有效的链接概念能很好被应用于解释设计中的从属关系,在设计依存性中对一个模块进行修改不会影响到另一某块。但是这并不能很好地适用于决定运行时间的性能(例如模块间信息传递失败),做到这一点需要更深的静态分析。但在其他的链接工具中(例如the Reflexion Model Tool [3]),链接的筛选能从他们的分析中分离得到,因此我们在这篇文章中所描述的技术将会比更精细的静态分析更好地被应用于解决这方面的问题。在LDM工具中,DSM的配置能显示出某个具体的链接或者显示出某个链接的强度,这种强度用两个有着联系的模块之间相互被引用的数目。 LDM使用DSM的算法式而达到分区的目的:现在它不提供自动归类的功能。LDM使用的默认值重置是基于代码的组合:JAVA组件结构。我们所介绍的矩阵是分层的(如4图所示),这种算法式和人工干预能在不同水平得到应用。用户能编辑系统的结构:为了反映他们自己对架构的认识,他们能通过手动重新调整模块的顺序和分区,能创建,删除以及改变子系统。用于结构的改变链接的数目应重新计算和求和。 |

2楼2009-12-08 13:46:00












回复此楼