Closed ChrisRackauckas closed 7 years ago
How about we leave it in this package for now but disentangle it by moving PDE stuff to a PDE.jl file? If that files starts to explode then make a new pkg.
Sounds like a plan.
How about we leave it in this package for now but disentangle it by moving PDE stuff to a PDE.jl file? If that files starts to explode then make a new pkg.
Well, upon first making it it was already more than a 1000 lines (just the basic FEM stuff, not even all of the FDM enhancement stuff) so I went and made this a package. (This will cut down on test dependencies immensely too...)
@mauro3 brought up a good point that the scope of DiffEqBase may not include PDEs. I think this is a good point because:
These facts seem to point that PDEs should be put in a separate DiffEqPDEBase (name pending), which would likely be more lively and user-focused, i.e. developed for giving users "the common interface" on tons of problems, and not for giving developers a common interface on the basic problems / types like DiffEqBase.