There are thousands of JS tests for the JS runtime, as well as integration tests for each of the first-party modules run on real hardware by our continuous integration bot before any change is merged to master in accordance with the not-rocket-science rule. Most bug fixes add regression tests to one of those test suites.
The closest to static analysis we have is
gcc's diagnostics: All C code is compiled with
-Wall -Wextra -Werror. Do you have any suggestions for other static analysis tools we should add to this workflow?
Network would be one major weak spot where there's not enough testing. In part due to finicky hardware/firmware that we have no control over, and in part due to the way that hardware is bound to JS, reliability has only recently reached the point were we're able to gate code merges on a series of tests avoiding failures unrelated to the new code. The existing tests are only at the level of the
require('net') API and run on a Linux host instead of on Tessel hardware.
As Jon hinted on the thread @dan mentioned, we're working on a path to escape both of these traps that we'll be able to give more details on shortly...