My talk at PyCon Italy

I’ve been giving a talk on Vulture for almost half a decade now. It takes me new places, and teaches me new things, and I quite like it. This time, it was even more special – it was in Italy. And rather than focussing on Vulture specifically, I wanted to spend more time on automated code analysis in general.

I also wanted to have a nice ice-breaker to loosen up the awkwardness which is generally present in technical conferences. And since we were in Italy, why not Football?

I started off with some old shots of Baggio head faking:

The phrase “head fake” was first used by Randy Pausch to denote hidden indirect learnings from other activities, e.g. when children play football, they don’t just learn how to kick a ball, they also learn skills like leadership, communication, comradery, etc.

Limited by how little I actually know about football, I promptly segued back to code analysis.

As they say, if you want someone to do something, convince them why. There’s a better option though: Tell them the cost of not doing it. That’s how we start.

37 seconds into descent in June 1996, the Ariane5 rocket flipped by 90 degrees and tore itself apart. Reason, it turned out, was a floating point error in the IMU: trying to stuff a 64-bit value into a 16-bit variable. The thing which made it even more unfortunate was that this value wasn’t even being utilized in the guidance system. That’s right: writing invalid data to an unused variable ended up obliterating a $500mn rocket.

Detecting these kinds of mistakes is what code analysis tools excel at. To tell you whether or not your code follows best practices, or if the function you’re using is prone to attack vectors or whether you comply with the style guide or not, or if you’re doing strange things like leaving dead code or unused imports behind. These tools, together, give voice to the collective wisdom of the community.

“The voice of the people has been likened to the voice of God.”

Niccolo Machiavelli

During the later half, I gave a brief demo of Vulture – how to run it, tweak all its knobs and joints to optimize it for your codebase, etc.

The last part, however, was the most exciting: I drew out how Vulture worked under the hood with Abstract Syntax Trees.

Conclusion

This talk was a head fake of sort: “Let’s learn about Vulture” to “Let’s take charge of our code!”

Here’s the deal: how Vulture works is irrelevant for most, if not all, of its users. But, knowing that there’s a way you can just pluck the AST and mess with it. It’s incredibly powerful.

P.S. The video isn’t out yet. I’ll make sure to add a link here as soon as it’s out.

Here you go:

TAGGED AS:

PYTHON

CONFERENCE

TALK

Share on Twitter