Sixteen Questions About Software Reuse

Document Sample
Sixteen Questions About Software Reuse Powered By Docstoc
					    Sixteen Questions About
         Software Reuse

William B. Frakes and Christopher J. Fox
      Communications of the ACM
                  Software reuse

• The use of existing software knowledge and
  artefacts to build new software artefacts
• sometimes confused with porting
   – reuse is using an asset in different systems
   – porting is moving a system across environments and
                     The survey
• Surveyed:
  – software engineers, managers, educators, and other in
    the software development and research community
• About:
  – attitudes, beliefs, and practices in reusing code and
    other lifecycle objects
• Response from
  – 113 people from 29 organizations (28 US and one
  – fairly representative of experienced software engineers
    or managers at high technology companies
                  16 questions

• The survey is used to answer 16 questions
  commonly asked by organizations attempting to
  implement systematic reuse.
• The answers to these questions are often taken for
  granted in the software community but without
  empirical verification
• This article uses empirical data from a survey to
  answer these questions
   Q1: How widely reused are common
• Are reusable assets actually used, and are they found
• Answer: yes and no.
   – Some assets are widely used (e.g.. UNIX tools) and perceived to
     be valuable, others are not used by many and are not perceived to
     be valuable (e.g.. Cosmic collection form NASA)
• Possible reasons:
   – lack of information and education
   – asset quality and ease of access
   – anecdotal evidence suggest that relevant functionality is an
     important factor
 Q2: Does programming language affect
• Does certain programming languages provide better reuse
  support (e.g.. by supporting abstractions, inheritance,
  strong typing)
• The data of the survey shows the correlation of 11
  languages and the level of organizational code reuse.
• Findings:
   – languages usually thought to promote reuse show small correlation
     (e.g.. ADA and C++)
   – Higher level languages are no more strongly correlated with code
     reuse than is assembly language
• Conclusion:
   – Choice of programming language does not affect code reuse levels
   – To increase reuse the focus should be on other factors
    Q3: Do CASE tools promote reuse?

• Many organizations regard CASE tools as a way to
  improve reuse
• The majority (75%) does not agree that CASE tools have
  promoted reuse
• No significant correlation between the respondents degree
  of belief in that CASE tools promote reuse and the
  percentage of actual reuse.
• Conclusion:
   – CASE tools are not currently effective in promoting reuse
   – Reasons:
       • CASE tools may not be used
       • CASE tools may not be used correctly
       • CASE tools may not promote reuse
 Q4: Do developers prefer to build from
          scratch or to reuse?
• Many people believe that software engineers prefer to
  build their own software rather than reuse someone else's
• Statement used in the survey:
   – ”It’s more fun to write my own software than to reuse”
• Most respondent do not agree on it (72%)
• Conclusion:
   – Most developers prefer to reuse rather than to build from scratch
• Contradicts conventional assumptions, but is in agreement
  with other findings of another recent study
Q5: Does perceived economic feasibility
           influence reuse?
• Reuse may not be done if it is not believed to be
  economically feasible.
• Levels of code reuse is compared with respondents
  agreement with the statement: “Reuse is economically
  feasible in my organization”.
• Clear trend towards higher reuse as belief in economic
  feasibility increases
• Conclusion:
   – perceived economic feasibility influences reuse
   – important to convince software engineers about the economic
     feasibility of reuse
    Q6: Does reuse education influence
• Comparing education about software reuse (at school) with
  individual levels of reuse shows that education on reuse
  does affect code and design reuse
• Comparing education about software reuse (at work) with
  organizational reuse levels of lifecycle objects shows that
  education at work does affect reuse
• Conclusion:
   – Education in school and at work improves reuse and is a necessary
     part of a reuse program
      Q7: Does software engineering
       experience influence reuse?
• General assumption that experienced software
  engineers are better practitioners
• The survey shows no significant correlations
  between engineering experience and personal level
  of reuse.
• Conclusion:
   – Software engineering experience has no effect on reuse
     of lifecycle objects
   – Maybe caused by historical lack of training in reuse
    Q8: Do recognition rewards increase
• Reuse incentives are considered to be necessary catalysts for reuse.
    – recognition rewards and cash rewards
• Respondents say that rewards are rare
    – no cash bonuses
    – only a few with other recognition
• Comparing levels of organizational code reuse between the two groups
  ”rec. rewards” and ”no rewards” show no significant differences
• Similar findings for individual reuse
• The findings contradict the common belief that recognition is a
  sufficient reward for reuse.
• It may be that only monetary rewards are sufficient rewards
   Q9: Does a common software process
             promote reuse?
• Statement used in the survey:
    – “A common development process has promoted reuse across projects in
      our organization.”
• Respondents generally do not agree on it.
• This might mean:
    – no defined common process
    – the process has failed to support reuse
• The correlations between the degree of agreement with the statement
  and the organizational reuse shows
    – moderate correlation for designs, test plans, and test cases
    – weak correlations for requirements, code, and user documentation
• Conclusion
    – A defined software process that promotes reuse does affect software reuse
  Q10: Do legal problems inhibit reuse?

• Legal issues e.g.. regarding contracting, ownership are still
  unresolved and are thought to be a serious impediment to
• The statement:
   – I am inhibited by possible legal problems.
• Most respondents (68%) report that legal problems do not
  appear to be an impediment
• Conclusion:
   – No significant correlation between levels of reuse and reported
     inhibition by legal problems.
   Q11: Does having a reuse repository
          improve code reuse
• A reuse repository is a searchable collection of
  reusable assets.
• Slightly higher median for reuse levels for
  organizations with a repository
• Not statistically significant
• Conclusion
   – Having a reuse repository does not improve reuse
   – A repository should not be the focus for organizations
     trying to improve systematic reuse
  Q12: Is reuse more common in certain
• Respondents were asked to classify their company or
• Significant differences in reuse between different
   – higher level for telecommunication
   – lower level for aerospace industries
• Reasons not currently known
• Industries with low reuse might benefit from studying and
  adopting reuse practice in industries that lead in software
     Q13: Are company, project sizes
    predictive of organizational reuse?
• Small organizations:
   – Is systematic reuse a realistic goal given the the scope of their
     domain and limited resources?
• Large organizations:
   – Introducing systematic reuse may be unrealistic because of the
     large investment and time required
• No significant correlation were found between
  organizational size and reuse levels
• Conclusion:
   – Company and project size are not predictive of reuse levels
   – Organizations of any size may succeed to introduce systematic
   Q14: Are quality concerns inhibiting
• It is commonly thought software engineers distrust assets
  developed outside and thus are less likely to reuse them.
• No correlation between:
   – ”software developed elsewhere meets our standards”
   – ”what percentage of the parts you reuse are form external sources”
• Experience in quality of reusable software have generally
  been favourable (69% agreement)
• No correlation between personal reuse level and
  experience in quality of external software
• Conclusion
   – Satisfaction with quality does not influence reuse levels
   – Does NOT mean that quality is unimportant
Q15: Are organizations measuring reuse,
      quality, and productivity?
• Measuring reuse is important to determine success of the
  reuse program
• Few organizations are currently measuring reuse (at most
• Quality is more widely measured (42%)
• Productivity measurement is relatively rare (32%)
• Conclusion:
   – Many organizations do not measure reuse levels, quality, and
   – These organizations cannot properly manage their software
     processes and products, including reuse
Q16: Does reuse measurement influence
• Measuring an activity will tend to increase it.
• Comparing median level of software reuse for
  organizations that are or are not measuring reuse.
• The study shows no correlation
• Conclusion:
   – Organizations that measure reuse do not use these
     measurements to improve reuse levels.
             How to improve reuse

• To improve systematic reuse concentrate on:
   – education about reuse
   – developers understanding on the economic feasibility of
   – introduce a common development process that
     promotes reuse
   – make high quality assets available to developers