Phew, another PyOhio has come and gone. This year was a great event. I can't say enough good things about the group that puts it together. It's really nice to have something somewhat local to head to every year.
- Data-Transfer Objects Are a Disease. Meet the cure.
The first I went to on a whim, I wasn't really sure what kind of Data Transfer Object (DTO) was going to be discussed. It turns out, the talk was a praising of the NamedTuple as a great way to pass data in Python. It's got some nice sanity checks, is very lightweight, and helps prevent developers from going crazy adding all kinds of business logic to simple data containers. I'm not sure how they're passing them around their applications, but I can see the appeal. I know I've been bit a few times when unserializing some JSON and a typo in a dict key bites me. I wonder if there's a way to easily get back a list of NamedTuples vs dicts when loading up some JSON transferred around.
- Aspen: A Next-generation Web Framework
I checked out the Aspen web framework talk just to check out what new ideas people are playing with when it comes to web tools. I wasn't disappointed, Aspen has some very interesting ideas. The author has done some work to bring back the url meaning something on the project layout. The idea of your layers of the app in a single file is kind of interesting, and I can see how that'd be helpful in some development cases. I end up sitting my controller/template code side by side when I work anyway.
The weakness I see, is that it's got the same issue PHP has when it comes to helping new developers start good practices. One of the great things about web frameworks, is that they help tell you where and how to organize your code. They give you test directories out of the box, they help bootstrap a good way to get your database connection up in a way that avoids shared sessions across requests, etc. Aspen does a lot less of that, and I could see a younger dev doing way more copy/paste of code than I'd want. It's a little bit of a bare framework, which is great to help integrate you tools of choice, but also provides a barrier to entry for new developers at times.
- Django and Google App Engine: Why I'm using Flask and EC2
This talk was from a friend on the west side of the state. As I've been part of the IRC discussions where he's been trying to go through various tools to build some small and simple web apps, it was cool to see the story in one swoop. He's a fan of the microframework. For his use cases, the full stack just threw hurdles in front of him. It's great to hear he's found a tool he loves in flask.
- Evolving an internal web service
Taavi gave a great talk on how they worked to rebuild an aging PHP app in Python over a long period of time. It's a great example of what I've been trying to get going for a while, APIs all the time. Everything seems to talk to the new application base via an array of remote methods and this great decoupling has been a boon for them to help provide data to several different systems in a clean way.
I wish I had more time to try to chat with him. They're doing some cool things with SqlAlchemy, migrating data from old systems to new, and just some very good work and testing done on performance.
- Creating web apis that are a joy to use
This was something I really was interested in. Since I've been pushing apis like mad at work and I've been working on my first public one for Bookie, I knew I needed some help/guidance on this. Issac did a great job hitting home the big rules I kind of knew, but wasn't following that well.
First, document, document, document. I loved his graph of user happiness vs amount of documentation. Users of your api don't get happy until the docs are near complete. Until then, it's just as bad as no documentation. I've got some typing ahead of me.
The second point was something I was battling with a bit. I tend to think of the api in terms of usage. "You want to do task XXX". At that point, I'm deciding what the user wants to do. In reality it's more about the term "resouce". A resource can be data, a function (send email), or something else along those lines. However you want to expose them via the api in simple distint manner. Just taking your current html view you push to browser users and building an api that is the same doesn't work. After all, the great thing about the api is that people build and do things you didn't think to do on your current application implementation.
- My Talk, Sqlalchemy Tutorial
Finally, I was of course at my talk. This year I decided to really didn't like myself and I should do more than a talk, but a two hour tutorial with some hands on coding exercises. The room was full of people of all levels and was a bit more challenging than I originally thought it would be. On top of that, the AC broke and the rooms were over 85 degrees which made holding an audience's attention all that much more challenging.
In the end, I think things came out ok. I've heard from an array of people that they enjoyed the talk. Once the first hour/talk part was over, most of the room left. We had about seven people that did the hands on code for the second hour. In the end, it sucks, but I can't blame them. If it wasn't my talk I'd have searched out cooler air as well. I hope that people still take the time to try out the hands on code and let me know if they run into any issues. If you do, feel free to email me email@example.com.
Thanks to PyOhio for letting me take a shot at something more classroom like. It's a new challenge to go from a talk to a tutorial and I encourage people to try it out.
I think it's true what they say, as you go to more and more conference you tend to do and learn less in the talks themselves, but make up for it with the great networking. This year it was very noticeable. What was great was that I got a chance to meet several people I've been following on Twitter for a while. These are people that are interesting, respected, and meeting them was kind of mini-starlet moments.
I got to have a great discussion around apis and self bootstrapping application installs with Issac Kelly. I met Michael Trier whos been a great Python presence on Twitter for a while now. I also caught up with the Ohio crew and guys Dave, Mike, and Catherine. If you run into these guys start a conversation, it'll be worth it.
I also had a ton of great conversations on things from new people that I really wish I did a better job of remembering and tracking down names for. Sorry that I don't call you all out.
All the hallway stuff really helped make this PyOhio a great one for me.
Another great thing PyOhio does are the sprints. Unfortunately I could only do them on Saturday, but man what fun it was. I think we managed to get 6 or so people with their own up and running Bookie instance running. We had nearly a dozen people hacking on things at one point. We had some fun hammering pypi from the wifi network and some really good ideas came up to help make the installs a bit easier. At the end of the night we had a pull request and some definite interest going forward. I hope that the people that sprinted on Bookie found it interesting to take part in and maybe learned something. I know I've got a lot of work to do still
I'll have a separate Bookie status report out later with some details on changes and things.