Re: [tlug] Issuses posed by code reuse

[Raymond Wan <rwan.kyoto@example.com>]

On Fri, May 18, 2012 at 1:51 PM, Nguyễn Vũ Hưng <vuhung16plus@example.com> wrote:
> On Fri, May 18, 2012 at 1:29 PM, Stephen J. Turnbull <stephen@example.com> wrote:
>>
>> An interesting reference to such nuances is in Fred Brooks's Mythical
>> Man-Month, 20th Anniversary Edition, where he discusses how reusable
>> component libraries can be considered a "language" of complexity that
>> starts to compare to natural language.
>
> Interesting. Thanks for point the metaphor out; I've just read a
> summary of the book a few days ago.


While the book is outdated, there are quite a few stories and
anecdotes that a summary (i.e., Wikipedia) probably doesn't do it
justice.  In the field of software engineering, it is one of the
classic books.  Unfortunately, many organizations and software
projects still repeat many of the mistakes cited.  Throwing more
people on to an already late project, for example...


> However, I think that in large project like developing an OS, to
> reduce the numbers of commnication channels
> needed among *modules* and people working on it, the Unix philosophy
> has been a manifesto:
> Building small tools that does exactly what it needs, promoting text
> based format to eliminate
> the effor of learning existing formats and interchange interface
> between modules.


That's true.  However, not all software engineering projects are based
on text-based files.  I know nothing about Linux's development, but I
wouldn't be surprised if communication between processes and modules
is still done using structured data (i.e., C-style struct) and not
text files.  And where there's structured data, there has to be
agreement between people on the interface.

I think what you meant is more for users of an OS (i.e., the sort,
uniq, cut, etc. utilities which are all very useful) and not OS
development.

Ray