ok, then I don't see any problems other than swig-related. Yes, 1.6 before 2.0 is good.
Regarding interface changes, I think we should allow for that now (if they are good), but the challenge is to get them in place before 2.0 and then again be conservative with interface changes during 2.x. -- Anders Tue Feb 17 2015 at 9:14:02 PM skrev Garth N. Wells <[email protected]>: > On Tue, Feb 17, 2015 at 8:10 PM, Anders Logg <[email protected]> > wrote: > > Where will this lead to interface changes? > > > > Not at the user level. I had more in mind for some new additions, and > maybe in cleaning up some low level interfaces. > > > Anyway, I think we should move to enums if it does not lead to big > > complications with swig versions. Perhaps if there are significant > interface > > changes, we could make the next release FEniCS 2.0 in June > > Let's make a release before June, maybe 1.6? > > 2.0 and radical interface changes - I like the sound of it :). > > Garth > > > (and perhaps it > > should be 2.0 even without enums). > > > > -- > > Anders > > > > > > Tue Feb 17 2015 at 6:15:13 PM skrev Garth N. Wells <[email protected]>: > >> > >> For a while now, we've tended to use strings rather than enums in > >> DOLFIN to make interfacing to Python easier. It may be a SWIG 3 > >> things, but SWIG seems to handle enums really well now. From C++, > >> enums are cleaner than strings, and from IPython we get tab-completion > >> for enums. From Python, the enum syntax is close to the NumPy syntax > >> for various things, e.g. np.inf. > >> > >> Is there any reason we should keep using strings over enums when enums > >> are appropriate? Should we worry about SWIG 2 compatibility? > >> > >> Garth > >> _______________________________________________ > >> fenics mailing list > >> [email protected] > >> http://fenicsproject.org/mailman/listinfo/fenics > _______________________________________________ > fenics mailing list > [email protected] > http://fenicsproject.org/mailman/listinfo/fenics >
_______________________________________________ fenics mailing list [email protected] http://fenicsproject.org/mailman/listinfo/fenics
