Gotchas Slides

class: inverse, center, middle

## Python 2/3 Gotchas!

---

# Integer (truncated) division

.boxed[

.center[
  **Python 2**
]
```python
>>> 5 / 2   # wtf?
2
```
.red[Caution!] When dividing one `int` by another, Python 2 returns an `int` by default. To get a fraction, make sure at least one of the operands is `float`.

```python
>>> 5 / 2.0
2.5
>>> 5 / float(2)
2.5
```
]

.boxed[
.center[
  **Python 3**
]
```python
>>> 5 / 2   # always gives an exact floating point answer
2.5
>>> 5 // 2   # always gives a truncated answer
2
```
]

---
# Printing output

.med-font.boxed[
.center[**Python 2**]

`print` is a statement (and a reserved keyword).
<pre><code class="python">
>>> print "The answer is", 2*2
</code><code>The answer is 4.</code>
</pre>
]

.med-font.boxed[
.center[**Python 3**]

`print` is a function.

<pre><code class="python">
>>> print("The answer is", 2*2)
</code><code>The answer is 4.</code><code class="python">
>>> print("The answer is", 2*2, sep="")
</code><code>The answer is4.</code>
</pre>
]
---
# User input

.boxed[
.center[**Python2**]

Use `raw_input()`
]

.boxed[
.center[**Python 3**]

Use `input()`
]

```python
>>> choice = input("Do you understand how the input function works [Y/N]? ")
```
---
# Encodings, oh my!

.med-font[
- Python 3 makes a very clear distinction between text **strings** (sequences of _unicode characters_) and textual representations of sequences of **bytes** (octets of 1s and 0s).

- Python 2 allowed easy mixing of the two, but this lead to lots of unexpected errors in the way text was read, encoded, and written in applications.
]

.center[

| data type   | Python 3 name| literal notation | has encoding |
|:-------|--------------|------------------|---------------
| unicode string  | **`str`**| `"hello"`| no |
| bytes     | **`bytes`**| `b"hello"`| yes |

]

<pre><code class="python">
>>> s1 = "I am a unicode string"
>>> s2 = b"I am a byte string"
>>> s1 + s2  # Not allowed in Python 3
</code><code style="color: red;">TypeError: Can't convert 'bytes' object to str implicitly
</code>
</pre>

.med-font[
The differences won't matter much for now, but consider reading the following article once you are very comfortable using strings:
]

.center[http://www.diveintopython3.net/strings.html]

---

# Back to the `__future__`

Some backports available in Python 2.7:

```python
from __future__ import division, print_function, unicode_literals

```

- the `division` import activates Python 3 style division semantics
- the `print_function` changes the print statement into a function
- the `unicode_literals` import makes the interpreter treat string literals as unicode text rather than bytes