Colin B. Macdonald
Colin B. Macdonald
Ideally, just have one place to deal with all that, but not urgent or anything
Hey, welcome back! There are a bunch of functions in `@double/`. They should probably each respect `f(nan)` gives `nan`. I haven't thought about this in a while: its possible this...
So in `@symvar/diff` we call `diff(f.sym, ...)` (effectively casting to `@sym`). I think instead we should call the superclass method `@sym/diff` on `f` itself. If this was classdef, that should...
might punt this until we are based on classdef (#545) and see if it can be easily fixed after that.
You might look at `@pyobject` which is classdef and works properly: https://bitbucket.org/mtmiller/pytave/src/d0a7f66393fcaeaad789d83c41462ed6a4c353ed/@pyobject/?at=default
Min Octave version is 4 which has classdef so this might be possible now.
whatever we do, make sure to update the "differences" document.
I still get `None` from SymPy: ``` In [11]: y = S('x') + oo In [12]: y Out[12]: x + oo In [13]: y.is_finite is None Out[13]: True In [14]:...
I think I agree with all that.
This is because of a dumb loop done in vpa.m. We can fix it by passing all the doubles in at once and avoiding all those round-trips. But it might...