http://d.puremagic.com/issues/show_bug.cgi?id=8757
--- Comment #9 from bearophile_h...@eml.cc 2013-06-26 14:01:57 PDT --- (In reply to comment #8) > Visual Studio is one of the worst compilers I've ever seen with regards to > warnings. It has tons of annoying, useless warnings which don't help one whit, > forcing you to shut them off. So, I would consider Visual Studio to be a > horrible example of what you should or shouldn't warn against. What evidence do you have to back your claim? This article shows a significant amount of people enjoying Visual Studio static analysis to detect bugs in a large amount of C++ code: http://randomascii.wordpress.com/2013/06/24/two-years-and-thousands-of-bugs-of-/ They enable only a certain subset of the warnings and they use a Python script that shows new warnings only the first time they appear in the code base. This is a simple but very useful memory, to solve one of the most important downsides of warnings. > And I would put > any warning about operator precedence on the list of warnings that should be > removed. It subverts the language when you're forced to add parens rather than > use operator precedence. Experience has shown again and again that humans become unreliable at keeping in mind operator precedence when it goes beyond a small number of rules and levels. From the evidence on the frequency of bugs related to operator precedence in C and C++ code, they go past such limit. This enhancement request is asking for an warning, then for it to become a deprecation, and later an error. -- Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email ------- You are receiving this mail because: -------
