Thanks your detailed explanation to nxstyle. I'll try my best to update nxstyle step by step.
Gregory Nutt <spudan...@gmail.com> 于2020年1月1日周三 上午4:42写道： > > > For (1) false negatives, I just create a few simple test files and > > verify that nxstyle detects the error. Pretty easy. > > > > For (2) false positives, it is a little more difficult. I use > > nuttx/sched to do these because it probably follows the coding > > standard the best. ... > > > There is another test that is needed. When we add a new test to > nxstyle.c, we may often break other logic that detects different coding > standard violations. The above do not address that. > > I think we would need to collect a directory full of code with known > coding standard violations and run and modified nxstyle against those to > to assure that we have not broken other detection logic. > > The code in nxstyle is not pretty. It uses many booleans and line > number varialbes to detect problems through some really rather crude > heuristics (it knows nothing about C syntax). Since it is convoluted, > unexpected side effects and unexpected breakage is really common. > > Greg > > > >
