R
Ron Bullman
asj,
Case studies are for learning, not for making comparisons.
Inorder to properly compare you need to have two (or more) implementations,
where the entities to be compared are known and measurable. This works well
for tangible artifacts, for example you measure the max speed of car A and
car B. If you make your measurement properly and such a way that somebody
else can repeat them, you could state that car B is faster/ slower than car
A.
However software systems and any system including intangible atrifacts do
have different nature. For example a reasonable complicated software product
is developed by groups A and B. A is developed with Java and EJB and B is
developed with .NET and C#. So you like to compare whether A or B is better.
You run them in similar machines and the output is similar and you find that
product A runs 10% faster than product B. Can you draw the conclusion that
EJB and Java is better than .NET and #C? Or could it be that the group B had
better developers than group A?
Inorder to find out whether it depends on tools or developers you need to
make several experiments. AFAIK nobody has ever made any such experiments.
Ofcourse you are entitled to your opinions, but that's what they are, just
your opinions (or somebody elses)!
Ron
Case studies are for learning, not for making comparisons.
Inorder to properly compare you need to have two (or more) implementations,
where the entities to be compared are known and measurable. This works well
for tangible artifacts, for example you measure the max speed of car A and
car B. If you make your measurement properly and such a way that somebody
else can repeat them, you could state that car B is faster/ slower than car
A.
However software systems and any system including intangible atrifacts do
have different nature. For example a reasonable complicated software product
is developed by groups A and B. A is developed with Java and EJB and B is
developed with .NET and C#. So you like to compare whether A or B is better.
You run them in similar machines and the output is similar and you find that
product A runs 10% faster than product B. Can you draw the conclusion that
EJB and Java is better than .NET and #C? Or could it be that the group B had
better developers than group A?
Inorder to find out whether it depends on tools or developers you need to
make several experiments. AFAIK nobody has ever made any such experiments.
Ofcourse you are entitled to your opinions, but that's what they are, just
your opinions (or somebody elses)!
Ron