#6: Python 3.6 is going to be awesome, Kite: your friendly co-developing AI

This is Python Bytes, Python headlines and news deliver directly to your earbuds: episode 6, recorded on Monday, December 12th. In this episode we discuss why Python 3.6 is going to be awesome, kite: your friendly co-developing AI, and more!

This episode was brought to you by Rollbar: they help you take the pain out of errors.

This is the last episode of 2016. Thank you everyone for a great launch. We’ll be back early January. 😉 Be sure to check out Talk Python and Test and Code if you want more Pythonic listening over the break.

News items

#1 Make your Python code more readable with custom exception classes

• This is a 5 min video + text. Good introduction into why you should define your own exceptions instead of using the built in ones, and how to do it.
  • It makes errors from your code more readable.
  • Better communication between your code and the person using your code.
  • It allows you to give more context of the error to the caller of the function.
  • Remember to derive from Exception or from another builtin exception.
• Do people create enough fine-grained exception types? I would say probably not.
• This advice is good because it encourages EAFP (easier to ask for forgiveness than permission) style of programming which is generally Pythonic.
• Allows for multiple except statements for different errors in one try block
• Dan also featured our show in The ultimate list of Python Podcasts (thanks Dan!)
• If you have a package that defines it’s own exceptions, please read another article.
  • The definitive guide to Python exceptions
    • Julien Danjou
    • https://julien.danjou.info/blog/2016/python-exceptions-guide
    • Covers having a common base exception for your package, organization within a package, and some examples of packages that organize their exceptions well, including requests

#2 Kite

• Kite augments your coding environment with all the internet’s programming knowledge.
• Is an AI pair programmer, or mentor really.
• Contextual info for
  • language
  • packages
  • e.g. “import r” → shows list of popular packages
  • then detailed docs, examples, etc.
  • autocompletions… by global popularity
  • BYOE
  • even works on your code
  • be sure to watch the video
• kite.com is implemented mostly in Go according to the founder Adam Smith.
• Thanks Gilberto Diaz for sending this one to us.

#3 Tidy Data in Python (by Jean-Nicholas Hould)

• This article caught my attention because of the notion that the data as you receive it might not be in a form that is ideal to use it. This I am used to. But the article give some attributes of what problems to look for in data sets, and how to transform the data into a more usable structure using pandas.
• Great example of someone taking a good idea from someone else, summarizing it, and showing how to use it in Python.
• Based on a paper named Tidy Data by Hadley Wickham
  • In this post, I will summarize some tidying examples Wickham uses in his paper
  • Will demonstrate how to do so using the Python pandas library
• Tidy data has the following attributes:
  • Each variable forms a column and contains values
  • Each observation forms a row
  • Each type of observational unit forms a table
• A few definitions:
  • Variable: A measurement or an attribute. Height, weight, sex, etc.
  • Value: The actual measurement or attribute. 152 cm, 80 kg, female, etc.
  • Observation: All values measure on the same unit. Each person.

#4 What's new in Python 3.6

• By Brett Cannon
  • Works at Microsoft Azure Data Science team
  • Python core developer
• 16 PEPs in Py3.6 - more than any other release than Py 3.0
• PEP 498 Formatted string literals
  • You learn about internals
  • That this is actually faster than str.format() because optimizations that can be done on the string itself (f””)
  • PEP 524: On Py 3.5 would fall back to unsecure. On Py 3.6 os.urandom() now blocks os.urandom() for cryptographically strong random numbers or os.getrandom() raises error if not enough randomness. Usually not a problem, but with things like containers and IoT, it has become one! Fix: use new secrets module.
• There are also other interesting things that aren’t PEPs
• Python 3.6 is generally significantly faster (than Py3.5 and legacy Python)
• Python 3.6.0 release candidate is now available, final expected end of the week
• Something that hasn’t been as highly publicized is the deprecation of pyvenv.
  • https://docs.python.org/3.6/library/venv.html
  • “The pyvenv script has been deprecated as of Python 3.6 in favor of using python3 -m venv to help prevent any potential confusion as to which Python interpreter a virtual environment will be based on.”
  • This is important for me and you and anyone who teaches people to use Python. We often recommend virtual environments, and it’s good to recommend -m venv to make sure people know which Python interpreter they are tying to their virtual environment.

#5: Who Tests What

• I had Ned Batchelder on Test&Code to discuss coverage. Episode 12.
• Ned is planning a new feature for coverage.py, which would tell you not just which code has run, but which test (or something else) caused that code to run.
• Discusses the stages of coverage, measurement/storing data/combining/reporting. Discusses the issue of how to decide who is calling the code in question. His current model is based on coverage having a plugin hook for someone to tell coverage a string that defines the “what” that is causing the code to be run. He also discusses some decisions about storage concerns and what coverage does now.
• He has questions remaining that he wants help with:
  • Today coverage.py keeps everything in memory until the end of the process, then writes it all to disk. Q: Will we need something more sophisticated? Can we punt on that problem until later?
  • Q: Is it important to try to conserve memory?
  • Today, the .coverage data files are basically JSON. This much data might need a different format. Q: Is it time for a SQLite data file?
  • Q: How would you use the data?
  • And a couple more questions regarding reporting.
• I like the direction this is going and I encourage everyone who has some nonstandard usage of coverage to take a look at this and give Ned some feedback.

#6 Threaded Asynchronous Magic and How to Wield It (by Cristian Medina)

• This is your async programming in Python 3.5+ via async / await article
• Covers:
  • Tasks
  • Scheduling tasks
  • Scatter / gather example
  • Moving the asyncio loop to a background thread
• Examples:
  • Real World Example #1 — Sending Notifications (email)
  • Real World Example #2 — Parallel Web Requests
    • via aiohttp: HTTP client/server for asyncio (PEP 3156)

Update: From episode 3: pynini https://en.m.wikipedia.org/wiki/Pāṇini