Right from the DjangoCon Circus

May 16, 2013
We are two programmers from CodeSyntax here the DjangoCon in Warsaw, happy to be in this circus (yes, we are in a tent!). These are some of our notes from Day 1.

So, some notes from Day 1 in DjangoCon in Warsaw.

Advanced POSTGRESQL – DJANGO – Christopher Petoff

Being Postgresql one of the leading open source DBs of the market, good to know how to mixes with Django. Most interesting of Petoff's presentation: how you can store JSON and also the ways to add attributes. He has stressed also the importance of performance. In short, praising the customizable properties of Postgresql, he has made a good point in favor of it.

Getting recommendations – Ania Warzecha

How you can build and manage a recommendation system like the big ones like Amazon or Facebook use, by means of Django. Particularly interesting: track users before the login, and once they do, bring that previous information to their profile. She has mentioned that in order to do all the numerous calculations in a painless way, she uses Redis DB.

Circus process and socket manager, Tarek Ziade

Supervisors are important to check how all components of an application work. We Python-programmers mostly use Supervisord but we all now its flaws: flapping, no real time stats, etc. So Ziade decided to create his own supervisor, and naming it Circus, it was most convenient that he presented it in the DjangoCon circus tent (!).  Main feature of Circus: chyanneling information thru sockets.

Processing payment for paranoids: Andy McKay

Good to know that for upcoming FirefoxOS mobile OS, the marketplace of the system will have a payment platform made in Django. Andy McKay is building it, and he told us all about security needs and parannoias. They're quite sure of the robustness of what they have built, as they've opened the code, offering money to anyone that cracks it. We are glad that Django and maximum security match well.

Thread profiling - Amjith Ramanujam

Profiling is the way by which we know how computing time is spent in our developments. In Python we mostly use a profiler called cProfile, but it's not well suited for really big projects. Processes consuming a lot of time are easily detected, but not very rapid things that are repeated a lot (and they can be troublesome). So Ramanujam has developed what he calls statistical profiling and assures that he gets much better results with it. We may try, we'll see.

JS & Django & originality

And a couple of final notes, one for Reinout van Rees's Django & Javascript talk, in which, we have discovered an interesting tool (Coffee script), and the other one for Zed Shaw for giving themost amusing and different talk of the day, not exactly about programming matters, but also about things that are important when working properly and sharing a sense of community.

(these notes are not only mine, but also partly from my coleague Aitzol Naberan; we've written more extensively in Basque; you see, English is both's 3rd language, so perhaps you'll also excuse the crudeness of this text ;-)


To comment this article you have to log in using your Facebook, Twitter or Google account.

Jatsu Argarate

Django programmer