Re: [lmi] More enabled warnings, less boost

[Greg Chicares]

On 2017-02-06 17:58, Vadim Zeitlin wrote:
> On Mon, 6 Feb 2017 15:35:42 +0000 Greg Chicares <address@hidden> wrote:
> 
[...pragma GCC push, pop, etc. ...]
>
> GC> No, thanks. I already tried those pragmata. I wrote them around
> GC> calls to boost facilities. I wrote them around inclusions of
> GC> boost headers. I still got the same number of lines of compiler
> GC> diagnostics (8932 lines or so). Searching the web, all signs
> GC> point to an open defect on gcc's bugzilla.
> 
>  I couldn't find any bugs related to this, could you please give a link to
> this one?

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=53431

> GC> So I took a different approach. Here's one thing that helped:
> GC> 
> GC>   -DBOOST_STATIC_ASSERT_HPP \
> GC>   -D'BOOST_STATIC_ASSERT(A)=static_assert((A),"")' \
> GC> 
> GC> If that tickles your fancy, I'm willing to share the whole set
> GC> of changes with you,
> 
>  The simplest would be to commit them to some experimental/Wconversion
> branch in the git repository, then I could not only look at it, but it
> would also make working on it and completing these changes easier in the
> future.

They're far too embarrassing to publish. "Pauca sed matura", as Gauss said.

> GC>     int cents() const
> GC>         {
> GC> //      return cents_ % cents_per_dollar;
> GC>         // C++11 [5.6/4]: "if the quotient a/b is representable in the
> GC>         // type of the result, (a/b)*b + a%b is equal to a"
> GC>         return static_cast<int>(cents_ - (cents_ / cents_per_dollar) * 
> cents_per_dollar);
> GC>         }
> 
>  I don't understand how is this preferable to
> 
>       return static_cast<int>(cents_ % cents_per_dollar);
> 
> ?

IIRC, with '-Wconversion', gcc flagged "cents_ % cents_per_dollar"
as suspicious in itself because the operands are of different types.

> GC> Maybe I should have used modf() or fmod(), but that would
> GC> have required too much thought for this quick exercise.
> 
>  I don't see how would it have helped neither, to be honest?

Superstition perhaps. I'm aware that some very smart people added
functions like those to C for a reason, and the reason, AIUI, is
somehow related to what gcc seemed to be objecting to here. I know
the motivation is to prevent splitting $1.99 into dollars and cents
and getting an answer like two dollars and negative one cents, but
I always seem to get this wrong if I try to go too fast.