On Wed, May 4, 2016 at 8:32 AM, 'Bill Hart' via julia-users <julia-users@googlegroups.com> wrote: > At the very least the warning should print correctly. Currently there is no > end-of-line when the message is printed, so we end up with a handful of > these errors stuck one after the other. > > But I too wonder what is the purpose of the warning.
Throwing inside a C callback is usually a bad idea (most C libraries won't handle it correctly and cleanly). However, the issue with the warning is that while it does capture the most obvious case where you made a type and the function always throws an error, there are cases as you describe where this might be useful and the error also doesn't capture the case where the function can conditionally throw (which IMHO is worse than always throw since at least you'll notice while testing if that's not actually what you want). > > On 4 May 2016 at 14:27, Yichao Yu <yyc1...@gmail.com> wrote: >> >> On Wed, May 4, 2016 at 8:20 AM, 'Bill Hart' via julia-users >> <julia-users@googlegroups.com> wrote: >> > Julia is now issuing warnings because functions do not return, e.g: >> > >> > function flint_abort() >> > error("Problem in the Flint-Subsystem") >> > end >> > >> > >> > What is the standard way of making Julia accept this as a valid function >> > (it >> > is not meant to return)? >> > >> >> This warning was added in >> https://github.com/JuliaLang/julia/pull/15972 . I personally think it >> should be removed. @Jameson. >> >> > > >