The second day of the PyCon 2017 conference was kicked off by breakfast with people from NASA and Pixar, among others, followed by several lightning talks. I didn’t see them all, but they were kind of fun. Then they moved on to the day’s first keynote by Lisa Guo & Hui Ding from Instagram. I hadn’t realized that they used Django and Python as their core technology.
They spoke on how they transitioned from Django 1.3 to 1.8 and Python 2 to 3. It was a really interesting talk and had a pretty deep dive into how they use Python at Instagram. It’s really neat to see Python being able to scale to several hundred million users. If I remember correctly, they also mentioned that Python 3 saved them 30% in memory utilization as compared with Python 2 along with a 12% boost in CPU utilization. They also mentioned that when they did their conversion, they did in the main branch by making it compatible with both Python 2 and 3 while continually releasing product to their users. You can see the video on Youtube:
The next keynote was done by Katy Huff, a nuclear engineer. While I personally didn’t find it as interesting as the Instagram one, it was fun to see how Python is being used in so many scientific communities and in so many disparate ways. If you’re interested, you watch the keynote here:
After that, I went to my first talk of the day which was Debugging in Python 3.6: Better, Faster, Stronger by Elizaveta Shashkova who works for the PyCharm team. Her talk focused on the new frame evaluation API that was introduced to CPython in PEP 523 and how it can make debugging easier and faster, albeit with a longer lead time to set up. Here’s the video:
Next up was Static Types for Python by Jukka Lehtosalo and David Fisher from the Dropbox team. They discussed how to use MyPy to introduce static typing using a live code demo as well as how they used it at Dropbox to add typing to 700,000 lines of code. I thought it was fascinating, even though I really enjoy Python’s dynamic nature. I can see this as a good way to enforce docstrings as well as make them more readable. Here’s the video:
After lunch, I went to an Open Space room about Python 201, which ended up being about what problems people face when they are trying to learn Python. It was really interesting and gave me many new insights into what people without a background in computer science are facing.
I attempted my own open space on wxPython, but somehow the room was commandeered by a group of people talking about drones and as far as I could tell, no one showed up to talk about wxPython. Disappointing, but whatever. I got to work on a fun wxPython project while I waited.
The last talk I attended was one given by Jean-Baptiste Aviat entitled Writing a C Python extension in 2017. He mentioned several different ways to interact with C/C++ with Python such as ctypes, cffi, Cython, and SWIG. His choice was ctypes. He was a bit hard to understand, so I highly recommend watching the video yourself to see what you think:
My other highlights were just random encounters in the hallways or at lunch where I got to meet other interesting people using Python.
This is my first PyCon in 6 years. My last being in Atlanta. This year, PyCon is in Portland, OR. My first conference day I managed to miss the morning talks due to rooms getting overly full. So I ended up on the hallway track instead and got to speak with various members of the Python community. For example, I was able to speak with Al Sweigart, author of “Automate the Boring Stuff with Python” and Eric Matthes author of “Python Crash Course”. It was fun getting to know fellow authors.
Backing up a bit, the keynote was given by Jake Vanderplas and he spoke on how Python is used in science, specifically astronomy. It was extremely interesting and really neat that so many astronomy programs are using Python for data analysis.
For the afternoon, I spent some time working as a volunteer in the “Green Room”. These are the behind-the-scenes people who make sure the talks go off without a hitch. It was fun and pretty busy.. I met a lot of people and I thought that it went quite well.
My first talk was Nicole Zuckerman’s (@zuckerpunch) The Glory of pdb’s set_trace. She works for Clover Health and discovered that people were still using Python’s print() statement for debugging rather than using pdb. She is a huge advocated of using pdb’s set_trace() method. This method is useful for setting breakpoints in your code, traverse frames in a call stack, inspect variables, etc. You can read some more about pdb in Python’s documentation. Personally I like the debugger that comes with Wingware’s IDE, however pdb can be useful when you’re on a machine or remote server that you can’t use a more powerful debugger on. Anyway, this was an interesting talk that might be useful for people who need to learn more about the pdb module and how to use it constructively.
If you’re like me, you missed PyCon North America 2014 this year. It happened last weekend. While the main conference days are over, the code sprints are still running. Anyway, for those of you who missed PyCon, they have released a bunch of videos on pyvideo! Every year, they seem to get the videos out faster than the last. I think that’s pretty awesome myself. I’m looking forward to watching a few of these so I can see what I missed.
Not sure how I missed this, but PyCon 2013 is already open for proposals, which means if you like to talk about Python, now’s your chance to show your chops! You can propose a talk, a tutorial or a poster. Head on over to their prospectus for more information. PyCon is a lot of fun and a good place to go to expose yourself to new things in Python. You can learn a lot just in the hallway circuit, let alone the actual talks! Put your thinking caps on because they’re only accepting ideas until September 28th!
The PyCon USA talk videos are finally starting to come online. You can check them out here: http://pyvideo.org/category/17/pycon-us-2012. I’m wondering why they chose this over the miro site that they’ve been using for the last few years. Maybe someone in the know can comment on that.
I noticed the streams I linked to seemed to be pretty hit or miss, so hopefully this will work better for those of you who missed out on PyCon like I did. Enjoy!
In case you didn’t realize it, PyCon has officially started today in Santa Clara, CA. Sadly my organization was too slow footed to secure a ticket for me, so I won’t be there to report on what’s going on this year. Instead, I’ll try to post links to other people’s blogs so you can get an idea of what it’s like there. You can also follow along on Twitter: https://twitter.com/#!/pycon. Finally, the PyCon blog seems to be staying fairly up-to-date with fun announcements. Hopefully next year, I’ll be able to return to PyCon and give it some live coverage once again.
On Sunday, March 13th, the final official PyCon conference day occurred (there were sprint days that followed, of course). Anyway, they had three interesting plenaries to help lead us into that day’s talks. Read on to find out what they were about.
The Threadless Plenary
Chris McAvoy, VP of Technology at Threadless in Chicago gave a plenary he called Going Full Python. Threadless is a t-shirt company that’s been around at least 10 years and has been a darling of the business world for a while now. I guess it’s also a popular campaign location for local Democrats.
Anyway, the reason they were at PyCon was because they use django for their website now after switching from php. He spoke on the history of the company and how they now support various causes, including the Japanese tsunami tragedy via the selling of special shirts. He had lots of lame jokes and I think his talk had the most cursing of any of them. You can read the comments about it on the convore thread that happened mostly in real-time. You can also see the talk here. I thought it was an interesting talk overall.
The Disqus Plenary
Jason Yan and David Cramer gave a plenary called disqus – world’s largest django system! They had lots of interesting statistics such as these:
disqus serves 500 million users
founded 4 yrs ago
only 16 employees of which 8 are engineers
traffic increasing 15-20% a month
doubled amount of traffic in 6 months
They created a program called Gargoyle that they open sourced that is some kind of feature switch decorator. I didn’t really understand it’s use case though. They also mentioned that they use the following projects: Hudson, Open Sentry, Monitor Graphite, pylint and pyflakes (I think). You can read the conference goer’s take on the talk here or watch the plenary here.
The OpenStack Plenary
Andy Smith gave a plenary on OpenStack, a project that has origins in NASA and Rackspace. Here are a list of the related projects that he mentioned:
swift – object storage system
nova – compute size, provisioning VMs
glance – image and registry storage
burrow – in erlang distributed message
dashboard (dash) – django admin interface
Mr. Smith mentioned that NASA uses OpenStack to detect asteroids, so that’s neat. NASA also uses it to take and manage pictures with WYSE (some kind of satellite, I think). The following all use OpenStack too: Citrix, SCALR, cloudkick, OPSCODE, NTT, and piston.
You can read the related convore thread here or the video here.