Ted Dunning wrote:
Actually, much of the API is fairly perverse. In the statistical code,
particularly, there are typically two three classes where one would do. I
think that somebody read the book on "style" and took it a bit too
literally. Factories and abstracts and interfaces and concrete
implementations are all well and good where there is something complex going
on, but for classes with just a few implementations an abstract super-class
and concrete implementations would do just fine. (and it could grow into the
more complex arrangement if there ever is a need).
Specific suggestions - ideally with patches - for improvements to the
API are welcome. In 2.0 we are making some changes and deprecating some
things, so now is the time to do this.
Phil
On Sat, Nov 29, 2008 at 7:10 AM, Phil Steitz <[EMAIL PROTECTED]> wrote:
An admittedly smelly, but useful feature of RealMatrixImpl is that it
exposes the underlying double[][] array.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
