On Thu, Apr 2, 2020 at 1:25 AM Gregory Nutt <spudan...@gmail.com> wrote:
>
>
> > I was able to do this once before:
> >
> > #if !defined(TODO)
> > # if !defined(NO_PRINT_TODOS)
> > # define DO_PRAGMA(x) _Pragma (#x)
> > # define TODO(x) DO_PRAGMA(message ("TODO - " #x))
> > # else
> > # define TODO(x)
> > # endif
> > #endif
> >
> > IIRC it allowed me to print compiler output that was not treated as warnings
> > then add -Werror= and treat all warnings as errors.
>
> This is a little different topic. You are talking about conditionally
> disabling some of the warnings?
>
> I was asking a different question about checking for new warnings in the
> PR check builds. Let me give the specific example. I had this error in
> the code:
>
> ret - nxsem_wait(&sem);
> if (ret < 0)
> {
> ...
> }
>
> The warning was that the computed value was not used in the line before
> the test annd then that 'ret' was not initialized in the test. Both
> were warnings and the code passed the PR check build successfully. But
> it is clearly wrong; that should have been '=' and not '-' in the line
> before the test;. If we were to detect new warnings in the builds, then
> this could have been caught.
>
> If we had no warnings in the system other than those generated by
> #warning, then your suggestion would a good one and maybe we will need
> to do that down the road.
>
> There are many places with REVISIT in the comments. It would nice if
> those could be made visible as well. REVISIT is the key word for
> flagging issues in the code (although I am sure some people may have
> used TODO as well).
>
Almost warning come from #warning, we can:
1.Add -Wno-cpp to disable #warning for precheck/nightly build:
https://stackoverflow.com/questions/9793293/how-do-i-disable-warning-message-in-gcc
2.Fix the remaining real warnning
3.Enable -Werror for precheck/nightly build
The suggestion from David is also valuable option for item 1, since
too much warning also block the developer find the real issue in daily
work. It is always good to enable people find the bug as early as
possible.
> Greg
>
>
