On 9/19/2010 1:43 PM, David Eddy wrote:
> Must confess that my going in assumption would be that the source
> code & data would be the only useful, accurate reflection for what
> the system is doing... & that whatever documentation might be found
> would be woefully out-of-date. (01)
There was a huge amount of documentation from various sources of
varying degrees of reliability. Both the software and the docs
were up to 40 years old, and both had been patched innumerable
times with varying degrees of metadata (ranging from zero to
exact descriptions of what had been done). (02)
The customer wanted to check the degree to which the documentation
agreed with the specifications. In some cases, new patches to
the software were based on assumptions about what other programs
were doing -- and those assumptions were often based on the
documentation instead of the original code. In other cases,
management had made business decisions based on the results,
but the documentation did not agree with what the code did. (03)
The goal was to compare each statement that mentioned any piece
of program or data to the actual code and check whether that
statement was accurate. (04)
It was also necessary to correlate different terminology that had
been in use over the years and produce a glossary that showed the
terms and definitions at different dates, the cross references
to the code that used those definitions, and the level of
agreement of the code and the definition (as far as possible). (05)
There was 100 megabytes of documentation of a wide variety of
different kinds -- the text amounted to about 100 typical books.
The consulting firm estimated that it would take 80 person years
to read all that documentation, check each statement against
the code, report all discrepancies, and generate the glossary
for humans and the metadata (data dictionary, data structure
diagrams, data flow diagrams) from *both* the documentation
and the code. (06)
And that's what the Arun and Andre did in 15 person-weeks
with the aid of suitable software. They produced it on
one CD-ROM, which the customer said was exactly what
they wanted. (07)
See slides 91 to 98 of (08)
http://www.jfsowa.com/talks/iss.pdf (09)
John (010)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (011)
|