Re: quality measured by tool
"Perhaps this tool isn't that great. I've seen this sort of thing be crap at actually judging code quality."
I can think of a number of reaons why a pull request should not be accepted...
a) use of global variables for no good reason. I had someone submit a patch that did this to a simple utility I wrote, years before github actually. I basically shelved it. NOT polluting the desgn.
b) use of hard tabs. *NO* - just *NO* (it doesn't display the same on every editor, duh)
the only exception are when required by the file format, like make utilities
c) violating basic code standard. If everything is ALLMAN STYLE, don't you DARE put a K&R style 'if()' block in there.
d) single character variable names - NO
e) anything whitesmith's style - double indenting is an irritation and an eyesore
f) you didn't include comments to explain anything? Even if it's trivial PUT FRICKING COMMENTS. The next confused person might be YOU 5 years from now, going "what was I thinking?"
g) code is just "unreadable" by people like me, who don't like looking at alphabet soup code and having to 'figure things out' instead of getting work done. Do the work up front, use comments and format it in a readable and consistent manner.
h) fears using 'goto' when it's not only practical, it's more efficient to use it. You find a lot of 'goto' in kernel code, worthy of mention.
Now... how does that tool do on THOSE things ???