Le mardi 08 septembre 2009 à 15:59 +0100, G. Perendia a écrit :
- Shall I then make the changes to its sylv library GeneralMatrix as per my
earlier email (enclosed) so that it can be used for kalman C++ and then integrated into the new estimation's k-order based solver too?
The idea is to keep only one source tree of dynare++. So the sylv/ subdirectory should disappear. The idea is to merge your changes in the main dynare++ tree, but without breaking the compilation of the standalone dynare++ binary, so go ahead with it. If necessary, use #ifdefs to enclose your specific changes.
- Eventually, the make-files for kalman may be changed too to refer to
Dynare++ sylv and kalman sylv library subdirectory probably removed. This is not necessary but may be advantageous as a reduction of duplication.
Yes, indeed, the kalman/sylv dir should be removed once you merge your changes into the main dynare++ tree.
I am going to create an appropriate build system which would build both dynare++ standalone and the MEX files which use parts of dynare++. Until I do that, you can use your own local Makefiles, using VPATH if necessary.
Best,