This is the mail archive of the libstdc++@gcc.gnu.org mailing list for the libstdc++ project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [v3] Simplify a tad _Bind<_Functor(_Bound_args...)>


Hi,
> It passes all my tests so far: if Jason can fix the damn serious issue
> with class types, I think we should apply it.
>
> If everything goes well, I also have in my tree: some trivial changes to
> the overloads for volatile / const volatile which allow to enable the
> code (just use consistently volatile and const volatile also when
> passing around tuple<_Args...>); trivial changes int -> size_t in
> various places indexing std::tuple (this changed since TR1), then a
> check __i >= can also be removed, of course.
>
> .... and we should add a *ton* of testcases.
>   
So... for now I'm applying this incremental improvement, without
attempting anything special for the volatile / const volatile issue (at
some point I add an hack which passes the testsuite, but I didn't trust
it, treated differently those overloads in an ad-hoc way). Besides what
we discussed already, I also noticed inconsistencies in the way
temporary tuple<_Args...> were passed around, and decided to uniform
everything to use rvalue refs, which seems the right thing to do. If you
are aware of testcases that breaks because of that, more than welcome of
course...

Tested x86_64-linux, committed to mainline.

Paolo.

/////////////////////

Attachment: CL_bind
Description: Text document

Attachment: patch_bind
Description: Text document


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]