One of my goals when writing talks is to produce content with a long shelf life. Because I'm one of those weird people that prefers to write new talks for new events, I feel like it'd be a waste of effort if my talks didn't at least age well. So how do things measure up if I look back on one of my oldest?
"Teaching Python: The Hard Parts" remains one of my most popular talks, despite presenting it just one time at PyCon US 2016. For most of the past two years, it held steady as a top 10 talk from PyCon 2016 by popularity on YouTube (although it was recently overtaken by a few hundred views 😳), even when counting it against the keynotes (!), and most of the YouTube comments are shockingly nice (!!).
Well, actually
Not everyone was a fan. Obviously I should have known better than to tell instructors they didn't have to use Python 3:
Did I give bad advice? Was mentiontioning the usability advantage of better library support and documentation SEO with Python 2 worth the irreparable damage I might have done to the community?
Matt's not the only one with a chip on his shoulder: the Python 2 → 3 transition has been contentious, and much ink has been spilled on the topic. A popular author infamously wrote a long screed claiming "PYTHON 3 IS SUCH A FAILURE IT WILL KILL PYTHON". Spoiler alert: Python is still alive, and the author updated his book for Python 3.
I've now spent a few years writing 2/3 compatible code, and am on the cusp of dropping Python 2 entirely. I've felt bad for not weighing in on the topic publicly, because people might have looked to this talk for guidance and wouldn't know my advice has changed over the past two years.
A little history
I wrote this talk based on my experiences teaching Python in the winter and fall of 2014, and presented it in early 2016. Back then, it wasn't clear if Python 3 adoption was going to pick up: Hynek wrote an article about Python 3 adoption a few months before PyCon that contained the ominous subheading "GLOOM". Python 3 only reached a majority mindshare of Python developers in May 2017!
Why? That's a topic long enough to fill a series of blog posts, but briefly: the number of breaking changes introduced in the first few releases in Python 3, coupled with the lack of compelling features to incentivize migration led to slow adoption. Personally, until the Python 3.3 release, I don't think Python 3 had that balance right to really take off. Version 3.3 was released in fall of 2012. Python 3.4 was only released in early 2014, just before I mentored at my first set of workshops!
This is a long-winded way to say, "when I gave this talk, it wasn't clear that telling workshop organizers to teach Python 3 would be good advice, because the ecosystem wasn't there yet."
The brave new world
But this is no longer the case! Python 3 adoption is overtaking Python 2 use, even in the enterprise space. The Python 2 clock keeps on ticking. Latest releases of Python 3 have compelling features to lure users, including strong, native concurrency support, formatted strings, better cross-system path support, and type hints.
This is to say, if I had to pick just one change to make to this talk if I gave it today, I would tell folks
USE PYTHON 3! ✨
Other updates
- The documentation for packaging Python is a lot better now. There have been many good talks presented on the subject.
- Distributing Python is still hard. There isn't a widely adopted practice for
cross-platform management of compiled dependencies yet, although wheels are
picking up steam. I'm currently working on the
manylinux2010update to address this problem on Linux systems.
Endorsements
Not to let one YouTube commenter rain on my parade, I am thrilled to say that some people in the community have written some awfully nice things about my talk. Thanks to all for doing so—pulling this together really brightened my day!
Blog Posts
Roxanne Johnson writes, "Elana Hashman’s PyCon talk on Teaching Python: The Hard Parts had me nodding so hard I thought I might actually be headbanging." 😄
Georgia Reh writes, "I am just in love with this talk. Any one who has seen me speak about teaching git knows I try really hard to not overload students with information, and Elana has a very clear idea of what a beginner needs to know when learning python versus what they can learn later." 💖
Tweets
When I presented this talk, I was too shy to attach my twitter handle to my slides, so all these folks tweeted at me by name. Wow!
Elana Hashman's talk on teaching programming has a lot of practical take-aways! @pycon #pycon2016
— Susan Tan (@ArcTanSusan) June 1, 2016
Learned a bunch from this video: Elana Hashman - Teaching Python: The Hard Parts - PyCon 2016 :: https://t.co/IVeaq0XugJ
— Mita Williams (@copystar) June 2, 2016
Muy buenos consejos en la presentación de Elana Hashman - Teaching Python: The Hard Parts - #PyCon2016 https://t.co/5dhNX5KO8y
— José Carlos García (@quobit) June 11, 2016
Elana cool talk! Explaining scope: maybe use analogy of person having same name as a celebrity @pycon #pycon2016 https://t.co/LAoGbY4uGO
— robin chauhan (@robinc) July 17, 2016
If you're about to teach a hands-on #Python workshop/tutorial/class, go watch Elana Hashman's #pycon2016 talk 🎓🐍💖https://t.co/xVJ6YBKFtV
— ✨ Trey Hunner 🐍 (@treyhunner) November 11, 2016
Other
My talk was included in the "Awesome Python in Education" list. How cool 😎
Declaring a small victory
Writing this post has convinced me that "Teaching Python: The Hard Parts" meets some arbitrary criteria for "sufficiently forward-thinking." Much of the content still strikes me as fresh: as an occasional mentor for various technical workshops, I still keep running into trouble with platform diversity, the command line, and packaging; the "general advice" section is evergreen for Python and non-Python workshops alike. So with all that said, here's hoping that looking back on this talk will keep it alive. Give it a watch if you haven't seen it before!
If you like what you see and you're interested in checking out my speaking portfolio or would like to invite me to speak at your conference or event, do check out my talks page.