wro.cpp wro.cpp
short

The May 2026 WG21 mailing in five papers (pre-Brno, 116 total)

· english · audience: working-cpp

WG21’s pre-Brno mailing landed in early May — 116 papers, the largest pre-meeting volume since the C++20 design crunch. Browse the full list at open-std.org/jtc1/sc22/wg21/docs/papers/2026/ or via the searchable wg21.org mailings index. Brno meets 8-13 June 2026.

Five papers are doing the heavy lifting for the threads wro.cpp covers.

1. P5000R1 — “Direction for ISO C++29”

Authors: Daveed Vandevoorde, John Garland, Paul E. McKenney, Roger Orr, Bjarne Stroustrup, Michael Wong (R0 2026-02-23, R1 2026-05-11)

The formal direction-setting paper for C++29 (not to be confused with P2000R5 “Direction for ISO C++,” the general direction paper). Safety is named as the headline axis — the line the committee adopted explicitly is that “C++29 must deliver on the safety story C++26 began but did not finish.” Concurrency continues to evolve (std::execution is the framework C++26 shipped; C++29 builds on it), and the paper sketches incremental work on reflection, modules, and language ergonomics.

For wro.cpp readers: this is the document that retroactively justifies the memory-safety-cpp26-and-beyond toolset page’s “deferred to C++29” framing. The direction paper makes that deferral non-negotiable — you cannot ship C++29 without a profile attribute story.

2. D3704R0 — “A type-safety profile”

Author: Bjarne Stroustrup (Jan 2026, revised for May mailing)

The next iteration past the C++26 deferral. Stroustrup’s R0 explicitly re-aligns the type-safety profile on top of Gabriel Dos Reis’s general profiles framework (P3589), addressing the Bloomberg counter-paper (P3543) critiques head-on — granularity at the per-rule level, structured error reporting, explicit migration paths from non-conforming code.

SG23 focus at Brno. The page above gets a re-review the week after Brno wraps.

3. P3970R0 — “Profiles and Safety: a call to action”

Authors: Vandevoorde, Garland, McKenney, Orr, Stroustrup, Wong — same author group as P2000R0 (Jan 2026)

A short paper (8 pages) co-signed by all the C++29-direction authors. The argument: the committee owes the community concrete profile timelines, not just direction prose. Names three milestones (SG23 framework R1 at Brno, type-safety profile R1 by autumn, vote at the post-Brno meeting in November). This is the committee leadership defending the timeline — the May mailing version is signal that the C++29 deliverable isn’t slipping further.

4. P4003R0 — “Coroutines for I/O”

Authors: Phillip Liard, Lewis Baker, Pavel Voržáček (Croydon onward)

Targets first LEWG review at Brno in June. The proposal is the standard-library complement to std::execution: a coroutine-aware I/O model that composes with the C++26 senders / receivers framework. Doesn’t promise C++29 (LEWG first review is the earliest possible milestone for major library proposals); doesn’t promise to look exactly like Lewis Baker’s libunifex or Eric Niebler’s pre-WG21 sketches; does promise to stop reinventing socket-loop integration for every reactor library in the ecosystem.

Worth tracking in particular because: every wro.cpp reader who has hand-rolled an async-I/O coroutine wrapper for boost::asio or liburing is the target audience.

5. P3294 — “Token sequence injection” (latest revision in May mailing)

Authors: Barry Revzin, Andrei Alexandrescu, Daveed Vandevoorde

The paper that wro.cpp’s “Where this is heading” triptych sections all gesture at. Revzin re-revised for the May mailing with refined ergonomics and updated example output. Implementation status remains: no shipping compiler. Clang-p2996 is the closest to having the substrate; a prototype likely lives in a research branch but isn’t surfaceable through Compiler Explorer yet.

Cross-link: Revzin’s C++Now keynote earlier this month is the design-space tour these papers live inside. The mailing version of P3294 is one specific point in that design space; the keynote is the careful evaluation of which axes any future code-gen paper — this one included — has to navigate.

Honourable mentions

Beyond the five above, the mailing has:

  • P2300 / std::execution ratification follow-ups — a half-dozen LEWG-issued fixups
  • Pattern matching (P1371) revisions — still alive, slowly grinding toward a complete-enough proposal
  • Linear algebra (P1673 / <linalg>) maintenance papers — the C++26 feature is shipping; the May mailing has bug fixes and a couple of small API revisions
  • Reflection follow-ups — most notably P3687R0 “Final Adjustments to C++26 Reflection,” which proposes deferring splice template arguments to C++29 (wording and implementation not fully baked for C++26). Additional papers tighten annotation lifetime, friend reflection, and error-quality edge cases
  • Modules — continued progress on import std; ergonomics across implementations

How to read along

Brno (8-13 June) is the next milestone. Sutter’s trip reports typically post within 48 hours of the meeting closing. wro.cpp will run a Brno trip-report news short in the week of 15 June.

Also from this week: if you ship C++, the ISO C++ Foundation’s 2026 Developer Survey “Lite” is open. 10 minutes of feedback that the standards committee + the major tool vendors (GCC, clang, MSVC, Bloomberg’s clang-p2996, NVIDIA, Intel) all read directly.