October 11, 2013
Wellfire was at DjangoCon 2013 in Chicago this year and it’s high time we reflected a bit about the conference and our experience there. This year we had the chance to share with the community in the form of a talk I gave on implementing search in a Django project.
This talk grew out of a talk given at Django District in the spring of 2013 and reprised last month in its latest form. The talk is a talk without heavily textual slides, so if you missed it you can catch a recorded version of the talk.
It’s an intermediate-level talk, expecting that you have a reasonable level of experience with Django regardless of your hands on experience with search. It starts from the beginning, defining the search problem, explaining how search engines help, integrating search into a Django project using Haystack, and some strategies to consider. But you can flip through the slides yourself.
One question I wish I had answered differently was about why to use
Haystack; in light of some of the extensive-looking work arounds I
implemented and discussed, why not just use the
library directly? Especially when you want to avoid database hits when
My revised answer would be to mention the
load_all keyword argument
which will reduce the database calls if you need to fetch model
attributes from a search result set and, more importantly, content
indexing. Even if you use a more direct connection to the search engine,
pyelasticsearch or ElasticSearch’s new official Python driver,
Haystack is still going to be an immense aid in identifying content to
add, update, and remove from the search engine.
The conference’s two tracks meant that a lot of speakers got to present and of course that it wasn’t possible to get to every talk. I met a lot of conference regulars and first-time attendees, not to mention first-time speakers. It spoke very highly of the community at large that so many people felt comfortable and compelled to speak for the first time, and at their first conference, and that the quality of these first-time presentations was so high.
The DC Django community got some great representation. Josannah Keller gave a community-oriented talk, My Bootcamp Brings all the Nerds to the Yard: Lessons from GeekChic, which I heard was awesome but I missed (again, the two tracks!). Longtime DC Python and Django denizen Eric Palakovich Carr shared in Winning an Election with Django and jQuery Mobile how he used these tools to rapidly build and deploy to the field a tool for his wife’s city council campaign. While it was great to see how he used these tools, it was just as interesting to see someone create a project for a real end (not just another toy project). Matt Makai gave an overview of some of the third-party services that further enable a site, Django or otherwise, and presented a thorough set of strategies for choosing and integrating them in Making Django Play Nice with Third Party Services.
Three - three! - talks explicitly focused on some aspect of database migrations. Apparently this is a big deal. The capstone of these was Andrew Godwin’s Everybody Loves Migrations. Andrew is the original author of South and is wrapping up the first part of a successful Kickstarter funded project to add database migrations into Django core as of Django 1.7.
One of the talks I enjoyed the most was Eli Ribble’s Django Toolsets: what are they buying you, what are they costing you?. In one of the few advanced talks Eli discussed how his company successfully transitioned from their old system to Django. Along the way they found that some of the great libraries and plug-and-play apps out there don’t always scale to your use case that well. From migrating schema changes with South to building APIs, this talk covers what worked, what didn’t, how they improved their system, and how large projects should approach out-of-the-box solutions.
You can watch all the talks on the Open Bastion’s YouTube channel.
DjangoCon US 2014 is heading to Portland and we’re already scoping out the city. See you all next September on the West Coast!
Another fine post by Ben Lopatin.
© 1997-2021 Ben Lopatin: follow me on Twitter; fork me on GitHub; connect, sync, and circle back with me on LinkedIn.