On Wed, Jul 10, 2013 at 2:37 PM, Manuel López-Ibáñez <lopeziba...@gmail.com> wrote: >> This points to other ideas: >> 1) how about adding a helper switch to show what is included in Wall? >> such as -Wall-print > > Doesn't > > gcc -Q -Wall --help=warnings > > give you this? >
Yes it does work as expected. > Otherwise, I think it is a bug. > >> 2) how about making -Wall configurable -- a default config file is >> looked at by the compiler, but user can change the config or use a >> different config they like. > > Isn't this exactly like using -Wno-error=* or -Wno-* ? It is a small > step to encode those in some file and pass the content of the file to > gcc as extra command-line options. > yes. > To be honest, I agree with Gabriel here. And I would go a step > forward, I would say that we are too timid with the warnings we enable > by default or by -Wall. We should warn more agressively, and let users > disable the warnings that they don't care about or are too buggy for > their taste. Because, at the end, it is a matter of taste. If > -Wmaybe-uninitialized saved your neck you will wonder why it is not on > by default. If it only shows false positives for your code, you will > wonder why it exists at all. > > -Wmaybe-uninitialized is able to catch real bugs, and the false > positives have easy work-arounds. It has some bugs, yes, but one can > always work-around them. Despite this very long thread, we don't > actually get that many complaints about it (we do get a lot of > complaints about false negatives, in particular PR18501). I agree with you here. thanks, David > > Cheers, > > Manuel.