Christopher Smith

``` What is lacking? (We have sqrtdenest and radsimp.) What other sorts of expressions are you thinking should be supported? ``` Original comment: http://code.google.com/p/sympy/issues/detail?id=2745#c3 Original author: https://code.google.com/u/117933771799683895267/

The side-specific versions could be `ldo` and 'rdo`.

So the `apply*` should be considered low level. We still might want to add higher level, safe methods like `mul` which would split the result into a Piecewise if the...

Could we keep `do` and use `Interval`-like options for left and right: `eq.do(op); eq.Ldo(op); eq.Rdo(op)`?

``` The expression reduces to zero after 2 applications of cancel, but they alone are not sufficient to reduce the expression to 0: as_content_primitive is also needed. I'm not sure...

``` These are the minimal orderings of ops that reduce z to 0: z.as_content_primitive().cancel().cancel() or z.cancel().as_content_primitive().cancel() ``` Original comment: http://code.google.com/p/sympy/issues/detail?id=3730#c3 Original author: https://code.google.com/u/117933771799683895267/

``` I guess the issue has something do do with expansion because z.as_content_primitive()[1].expand().cancel() works; the expand can come before the as_content_primitive call, too. ``` Original comment: http://code.google.com/p/sympy/issues/detail?id=3730#c4 Original author: https://code.google.com/u/117933771799683895267/

cf #23015, #5765

evalf is not as naive as subs; x*0 could be 0 or Nan; since you didn't resolve all symbols you can't get an evaluation. Original comment: http://code.google.com/p/sympy/issues/detail?id=3875#c1 Original author: https://code.google.com/u/117933771799683895267/

I wonder if this is related to what I think is a broken mechanism: if an _eval_evalf routine is defined (as there is for Mul) then the subs result doesn't...