write tests for...

[sbillinge]

Problem

• eq
• add
• radd
• sub
• rsub
• mul
• rmul
• truediv
• rtruediv

before doing this talk to @yucongalicechen or @sbillinge about new data structure for the arrays in the DOs.

Proposed solution

[bobleesj]

For the group's testing standards, please read #236 and #225 and #255

In the future, group's testing standards will be more thoroughly documented.

[bobleesj]

@sbillinge I am trying to write tests for these. Two important questions below:

Q1. What's the difference between, for example mul and rmul shown below?

 def __mul__(self, other):
 multiplied = deepcopy(self)
 if isinstance(other, int) or isinstance(other, float) or isinstance(other, np.ndarray):
 multiplied.on_tth[1] = other * self.on_tth[1]
 multiplied.on_q[1] = other * self.on_q[1]
 elif not isinstance(other, DiffractionObject):
 raise TypeError("I only know how to multiply two Scattering_object objects")
 elif self.on_tth[0].all() != other.on_tth[0].all():
 raise RuntimeError(x_grid_emsg)
 else:
 multiplied.on_tth[1] = self.on_tth[1] * other.on_tth[1]
 multiplied.on_q[1] = self.on_q[1] * other.on_q[1]
 return multiplied

 def __rmul__(self, other):
 multiplied = deepcopy(self)
 if isinstance(other, int) or isinstance(other, float) or isinstance(other, np.ndarray):
 multiplied.on_tth[1] = other * self.on_tth[1]
 multiplied.on_q[1] = other * self.on_q[1]
 elif self.on_tth[0].all() != other.on_tth[0].all():
 raise RuntimeError(x_grid_emsg)
 else:
 multiplied.on_tth[1] = self.on_tth[1] * other.on_tth[1]
 multiplied.on_q[1] = self.on_q[1] * other.on_q[1]
 return multiplied

Q2. We've made all_arrays immutable - This makes the existing __mul__, __add__ methods invalid. We must create a new object. For example, the on_tth function above within __mul__ isn't valid since it can't modify all_arrays.

I am thinking we might have to change from

 def __add__(self, other):
 summed = deepcopy(self)
 if isinstance(other, int) or isinstance(other, float) or isinstance(other, np.ndarray):
 summed.on_tth[1] = self.on_tth[1] + other
 summed.on_q[1] = self.on_q[1] + other
 elif not isinstance(other, DiffractionObject):
 raise TypeError("I only know how to sum two DiffractionObject objects")
 elif self.on_tth[0].all() != other.on_tth[0].all():
 raise RuntimeError(x_grid_emsg)
 else:
 summed.on_tth[1] = self.on_tth[1] + other.on_tth[1]
 summed.on_q[1] = self.on_q[1] + other.on_q[1]
 return summed

to

 def __add__(self, other):
 if not isinstance(other, DiffractionObject):
 raise TypeError("I only know how to sum two DiffractionObject objects")
 tth_sum = self.on_tth[1] + other.on_tth[1]
 return DiffractionObject(xarray=tth_sum, yarray=self.yarray, xtype="tth", wavelength=self.wavelength)

The above assumes 1) self's yarray is used in the new object 2) tth xtype is used 3) and the self's wavelength is used.

[sbillinge]

Q1. What's the difference between, for example mul and rmul shown below?

they are the same because the mul operation commutes.

[sbillinge]

Q2. We've made all_arrays immutable - This makes the existing __mul__, __add__ methods invalid. We must create a new object. For example, the on_tth function above within __mul__ isn't valid since it can't modify all_arrays.

Those arrays are not immutable, we just made them not settable, right? I think we can update them how we want inside a method.

[bobleesj]

Those arrays are not immutable, we just made them not settable,

Right, all_arrays isn't settable. I guess we can modify _all_arrays internally which is mutable.

[bobleesj]

Since 3.6.0 deadline is approaching, I will resolve this issue

[bobleesj]

I will keep progress here:

• eq
• add Refactor __add__ operation in DiffractionObject and add tests #285
• radd
• sub
• rsub
• mul
• rmul
• truediv
• rtruediv

Note to myself:

• Refactor __add__ operation in DiffractionObject and add tests #285 (comment)