轉貼: OO 方法論

[小瑞]

==> 於 夢幻花園 (Jianzhi@FCU-SkyNet) 文中述及:
: ---------------------------------------------------------------------------
: NCTU-CIS BBS `oop' 版 精華區
: ■■■ OO 方法論 ■■■
: ---------------------------------------------- 整理：william@cis_nctu -----
: ---------------------------------------------------------------------------
==> 於 C++/ASM/Win Master (william@cis_nctu) 文中述及:
: 底下是一些關於 OO 方法論的討論...
: ===============================================================================
: Newsgroups: comp.lang.c++
: From: rma...@rcmcon.com (Robert Martin)
: Subject: Re: OO methodologies (ie. Booch, Rumbaugh et. al., etc. )
: Organization: R. C. M. Consulting Inc. 708-918-1004
: Date: Mon, 1 Nov 1993 15:08:40 GMT
: mi...@barbados.bellcore.com (Mike Kleszynski) writes:
: > I was wondering if anyone out there uses any of the published ( by
: > published I mean there is a text book written on the methods )
: > OOA/OOD methodologies like Booch or Rumbaugh. What I'm wondering is
: > how useful you find it, how closly you follow it, and where is it
: > used (ie. in the system specification, design document, or
: > elsewhere). Do they really help small projects or are they too much
: > overhead; is it possible to manage a large project without using one.
: > ( I know it's possible but will the quality of the product suffer? )
: My recommendation:
: Read Booch, Rumbaugh, Jacobsen and Meyer. Use Booch's
: notation simply because it is the most comprehensive and flexible.
: Use Jacobsen's ideas of "use-cases" and "control/interface" objects.
: Study and apply Rumbaugh's use of finite state machines. Study
: Meyer's Open/Closed principle (which is the one of the few true
: benefits of OO).
: What is a methodology? Do you want a step by step procedure for the
: production of software? You won't get it from these books. If you
: find a book that suggests one, be suspicious! I doubt that any such
: procedure can actually work.
: Booch, in particular, recommends a non procedural approach to software
: analysis and design. This sentiment is echoed by many other OO and
: non OO authors. Software design is the attempt to find patterns in a
: problem domain. The search for these patterns cannot be automated.
: It relies upon human insight and creativity. No methodology can
: provide this insight and creativity for you. Designing a problem from
: the top down, or from the bottom up, or iteratively, or as a data
: model, or as a process model, or as an object model, will *NOT*
: guarantee you good designs. Will not guarantee that you are creative
: and insightful. Will not guarantee that you follow the path where the
: best insights are to be had.
--
│ ─┬─ │ │ │
│ │ └─┴─┘
│ │ │ │ ─┬──
│ │ │ ─┼─ ┌───┐
┘ │ └ │ ││ │ │
┘ ─┴─ ││ │ │
[mΞ Origin: 逢甲蒼穹資訊網 <bbs.fcu.edu.tw> [FROM: 140.134.212.41]