DeepCode Updates for May 2020




DeepCode Static Code Analysis comparison to sonarqube and lgtm semmle

Blog

DeepCode up to 54 times faster than comparable tools

Performance is crucial for DeepCode in two respects: First, it is needed to process large amounts of code for training; secondly, when applying the model in your IDE, it needs to be close to real-time. We compared DeepCode with some other leading solutions in JavaScript and found DeepCode to be up to 54x faster. 

See full report here







DeepCode Infer Types for Python

Infer types - Teaser

Ability to Infer Types or What Python Got to do With It

Dynamically typed languages like JavaScript or Python provide a lot of flexibility. The downside is that mistyping can lead to exceptions, crashes, and are hard to trace. But there is only so much a typical IDE can do. We added new functionality to the DeepCode runtime that enables us to infer type information by analyzing the source code.

Learn more




Bug Fix

Analysis is now running only once

We found a bug (how meta!) that sometimes the analysis was performed again after the initial run. We released a fix a week or so ago and since then you should see a running analysis only once. This should make the analysis appear fast after the initial run so you can dive into the DeepCode suggestions.

Check out the fixed version