On Fri, 22 May 2026 at 12:49, Matthias Kretz <[email protected]> wrote:
>
> Jason Merrill via Gcc [Thursday, 21 May 2026, 20:12:01 CEST]:
> > > 3) add diagnostic group -Wmiddle-end (name pending);
> >
> > This seems too broad a category;
>
> I think it's a very interesting category. But, with that name, it's not very
> discoverable for users.
>
> My experience with these kinds of warnings (and I implemented library
> precondition checking in a similar manner) is that they can be resolved by
> precondition checking that is visible to the optimizer. I like to think of
> this as "forcing precondition checks to bubble up".
Sometimes the precondition checks are there, but don't survive IPA, or
something makes them get forgotten before the location of the warning.
When those warnings happen inside std::string or std::vector, it can
be very difficult to figure out how to provide enough information to
the compiler to make it stop being annoying.
>
> Consider:
>
> char str[] = "Hello";
>
> // precondition: cond == false
> char
> f(bool cond)
> { return cond ? str[100] : str[0]; }
>
> This warns about str[100] being out of bounds. But the real problem was that f
> was called out-of-contract. If we add "pre (cond == false)" and compile with
> -fno-contracts-conservative-ipa (and a terminating contracts evaluation
> semantic) the warning goes away.
>
> So you can think about these warnings as "your function is missing a
> precondition check". Conceptually. Because C++ contracts currently default to
> hiding from the optimizer.
>
> Thus:
> 1. middle-end warnings should hint at missing precondition checks
> 2. middle-end warnings maybe should only be on by default if contracts
> checking is visibly (to the optimizer) terminating. (I've been using
> __builtin_trap for this.)
>
> On the name: "-Wmiddle-end" is an explanation to compiler devs and nobody
> else. How about something in the direction of "-Wstatic-precondition-
> checking"?
>
> - Matthias
>
> > let's focus on the most common problems,
> > namely the ones supported by -fdiagnostics-show-context=N. Maybe
> > -Wrange-flow or something like that?
> >
> > I also notice that pass_warn_access (-Wstringop-overflow) runs multiple
> > times in passes.def; maybe the early run could stay in -Wall? Maybe even
> > adjust all of the offending warnings to run before jump threading in -Wall
> > and also optionally later?
>
>
> --
> ──────────────────────────────────────────────────────────────────────────
> Dr. Matthias Kretz https://mattkretz.github.io
> GSI Helmholtz Center for Heavy Ion Research https://gsi.de
> std::simd
> ──────────────────────────────────────────────────────────────────────────
