Le 14/11/2010 17:19, Luc Maisonobe a écrit : > Hello, > > One large change that was done for version 2.1 was to add a way to > compute the partial derivatives of an ODE final state with respect to > both its initial state and problem parameters. This was done by creating > a new ode.jacobians package. > > As time as shown, this attempt is really clumsy. Some issues have been > raised (MATH-380, MATH-381, MATH-388). The package is difficult to use > and it duplicates many features from the raw ode package. From a user > point of view, the current implementation is bad. > > As I suggested in the comment for MATH-380, a completely new > implementation should be done. We are working with Pascal on a new > implementation which would be much more smoothly integrating with the > top level ode package and which would allow the same user code to be > used in both cases : with or without partial derivatives. The new > implementation implies adding a few methods to public interfaces, so it > breaks compatibility and cannot be introduced in 2.X branch. > > I would like to deprecate the complete ode.jacobians package right now > to prevent people from building application on it, to postpone MATH-380 > to 3.0 (MATH-381 is already scheduled for 3.0). > > What do you think about this change ?
There was no comment on this, so I will go ahead and deprecate the complete package in 2.X and remove it in 3.0. Luc > > best regards, > Luc > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
