(ref.doc)iesw189

[posted to dshp01 news by Claude]

"The [Discovery] mission [which landed on 89oct.3] actually flew with
some known software errors. In fact, the average backlog of known
errors in the 14 millions lines of code is about 1800. [...]"

"Following the 1986 Challenger disaster, the Rogers commision made
many recommendations to NASA, leading to changes in shuttle and its
software. [...] In all, there were 3800 requirements changes. These
resulted in 900 software releases [...]. We built SW the old-fashioned
way: lots of sweat, a lot of control, a lot of discipline, basically
manual work. [...] We have an exhaustive test program. [...]The issue
is good management practices [...]. We have to recognize that SW
development is hard, complex, tedious work and there is no management
or technical substitute for understanding the problem."

Just two comments: their bugs per line ratio is slightly bigger than
0,01%, one bug every 10 000 line, am I right?

Then, they do *exhaustive* tests on their SW!!