[docs] [issue21227] Decimal class error messages for integer division aren't good

[leewz]

leewz added the comment:

Total list of issues now:
- Error message for `DivisionImpossible` is
      [<class 'decimal.DivisionImpossible'>]
  instead of an actual error message.
- `decimal.DivisionImpossible.__doc__` is empty.
- Calling `help(decimal.DivisionImpossible)` turns up nothing useful.

I checked all of these just now on my 3.2.3 (Windows). These issues are a CHANGE from 3.2 to 3.3. For example:
- Old error:
  decimal.InvalidOperation: quotient too large in //, % or divmod
- New error:
  InvalidOperation: [<class 'decimal.DivisionImpossible'>]

I assume that the issues also apply to the other InvalidOperation types. So I guess what I'm really asking for is to pull forward the 3.2 docs and messages.

> That doesn't mean the implementation should change.

Er, I was explaining why it wasn't really surprising once I thought about it.

> There may be a documentation issue, but then again, people who are "surprised" usually haven't studied the docs in detail.

To clarify:
- I looked at the Decimal docs
    - skimmed for references to modulo and division
    - looked for `DivisionImpossible`
- Then looked on Google for '"DivisionImpossible" Python'.
- Experimented with changing precision and fooling around with bigger and smaller numbers, and realized why the operation was logically impossible.
- Came here, posted, got a response that it was a flag raised by libmpdec.

I'm not asking for a change in behavior. I did point out that it was inconsistent with regular floats, in case consistency with Pyfloats was a thing that mattered. But I don't care about that. I only worry about anyone else who would use Decimal coming across the same unhelpful error.

> AFAICT, this particular "confusion" has never arisen before (i.e. bug reports, questions in python classes, posts on stackoverlow, blog posts, etc).  Likewise, the issue does not seem to have ever arisen in the many other non-Python implementations of the decimal spec.

I agree it'd be (very) rare, but part of the reason why it might not appear online is that the issues at the top of this email are relatively new (3.3).

----------

_______________________________________
Python tracker <report at bugs.python.org>
<http://bugs.python.org/issue21227>
_______________________________________