Review of Mou-Sadowsky Book 
1) I believe that this book potentially fills a need and could
be well done. As is obvious the proposal is not very detailed
and one cannot be very precise in such evaluations. Its
interesting feature is the practical (or engineering) approach,
which is comparatively unusual. Usually a more abstract or
formal style is adopted. I would personally find the proposed
approach attractive. 
2) The organization seems sound but it is very difficult to
judge content as many important details are unsaid. The
systematic use of Java is a reasonable choice but it is not
clear how the authors will address the comparitive paucity of
quality algorithms implemented in Java. I note the area of
numeric algorithms (such as special functions and matrices) is
subject of much work as there are difficulties in achieving
effcient implementations. It is not obvious how much attention
this area will get the book. I can recommend the work of the
NIST and Visual Numerics groups to the authors. This field also
emphasizes another difficulty with the choice of Java.
Currently the computer science community is debating
appropriate standard classes (for things like differential
operators, arrays, matrices and complex numbers). The authors run
the danger of making choices that are obsoleted. Perhaps more
important is the issue of parallelism which appears to be a
feature of book. Java has a natural thread parallelism but my
guess is that they would also address distributed memory
architectures. Such details appear important but are not
discussed in this brief proposal. I should also note that I
view the cited system Divacon as interesting but rather
idiosynchratic (specialized) for inclusion in such a book.
3) I believe that the authors are well qualified to write this
book. I am familiar with Mou's work which is of high quality.
4) I think such a book could be used effectively in many core
computer science courses. Its enginering flavor is its strength
and weakness in this regard. There are not so many books of
this style but also many computer science faculty would prefer
to teach the more conventional formal approach.
5) I would support investigating publication seriously but I
suggest that the authors think through the details first.