C++26 Is Reopened: The 13–9–12 Ballot, the Enforced Label That Answers It, and What Is Still Open
Software / analysis
C++26 Is Reopened: The 13–9–12 Ballot, the Enforced Label That Answers It, and What Is Still Open
Sutter's Búzios report says the committee kept contracts, added one paper, and will send C++26 to a second ballot with no published date.

C++26 was declared finished in March 2026. It is now unfinished again, and the reason is a single feature: contracts, the pre, post and contract_assert keywords that let a function state what it expects and what it promises.
On November 21, Herb Sutter published the trip report from the committee's six-day meeting in Búzios, Brazil. The title says it plainly: "C++26 is done (again)!" The draft passed its first international ballot without unanimity, was "sent back to the ISO C++ committee with additional international comments to address", and was reopened.

How the first C++26 ballot failed on November 13 numbers
The Register reported on November 13 that the Draft International Standard ballot closed at 13 in favour, 9 against and 12 abstentions. ISO's rule, as The Register describes it, is no more than 25% negative votes. Counting only the 22 bodies that took a side, 9 against is 40.9%.
Sutter's report adds the detail the ballot tally lacks. Five national bodies raised contracts concerns for the first time and voted Disapprove. His own history of the feature says "the first DIS was not approved."
| Vote | Body | Yes | No | Abstain |
|---|---|---|---|---|
| DIS ballot, per The Register | ISO national bodies | 13 | 9 | 12 |
| Reship vote at Búzios, per Sutter | WG21 meeting | 111 | 9 | 8 |
The second row is a different electorate. It is the committee's own vote in the room, and Sutter reports it at 92.5% in favour of reshipping C++26.
What the Búzios meeting changed in C++26
The committee did not remove contracts. It added one paper to them. P4451R2, "Early Adoption of an Always-Enforced Label" by Joshua Berne, lets a programmer mark a single contract so that execution never continues past a violated assertion, however the other contracts in the build are configured. The label is std::contracts::controls::enforced.
For anyone writing pre and post conditions in a library, this is the sticking point to watch: the enforced label. Other contracts depend on how a build is configured, and Sutter's description of the label is that it holds "regardless of how other contracts are built." The label lets the author say which checks are never negotiable, which matters most to library authors whose users choose the build flags. Sutter reports that P4451R2 was adopted for both C++26 and the draft C++29.
Whether it answers the five bodies that voted Disapprove is not something either source settles. Neither lists them, and Sutter does not give a count of the comments.
Why contracts split the committee
Contracts let developers attach preconditions, postconditions and assertions to code, which can catch errors that would otherwise lead to undefined behaviour. Ada and Eiffel have had similar features for decades. C++ pulled an earlier version in 2019 over concerns about dangerous defaults, and the revision in C++26 is paper P2900, according to The Register.
The objections are not small. Bjarne Stroustrup and colleagues, The Register reports, call P2900 "an existential threat to C++" and argue it is messy, outdated and under-tested, and would be better as a library. Andreas Fertig, a member of the committee, says those criticisms were already vetted and that a veto this late does damage. Harry Bott, CEO of the C++ Alliance, calls the draft being sent back "the process working as intended."
Thibault Ricord-Marchal of the Margo consultancy laid out three routes: cut contracts so the rest can ship, move them to a separate technical specification, or make a minor revision and vote again. Cutting would be messy, he said, because library hardening already depends on contracts. Sutter's report is consistent with the third: contracts stayed, and one paper was added to them.
What C++29 gained in the same week
The meeting was not only about contracts. Sutter lists five core-language papers and seven library papers adopted for draft C++29. Among them:
- P3568R4, from Jan Schultke, adds
break label;andcontinue label;as named jump targets. - P0085R5 adds
0ooctal prefixes and deprecates the bare leading0. - P3411R7 adds
std::ranges::any_view, and P3216R6 addsviews::slice. - P4189R0, from Nevin Liber, adds
.get_ptr()tostd::optional.
What happens next for C++26
The committee is preparing an updated C++26 document for a second DIS ballot, followed by editorial work and ISO publication "in the near future." Sutter gives no dates. The Register said that a slip into next year would make the official name ISO/IEC 14882:2027, with an informal C++27 label likely to follow.
Teams that pinned a compiler upgrade to C++26 on a published date, much as Vim Classic states which Python versions it drops, have nothing firm to plan around until that second ballot closes. The next WG21 meeting is in Istanbul in March, where C++29 work continues, and Sutter's report does not say whether the ballot will finish before it.
Compare that open-ended wait with the way a database project documents its limits at release: our Postgres 19 REPACK piece lists what shipped and what it skips. C++26 has the first half of that list, the adopted papers, and not yet the second.
Sources
More in Software
- 01Levien's Next-100 Rust GUIs Post Counts About 20 New Toolkits and Bets Against egui's Immediate ModeThe Linebender author credits AI rewrites for the surge; egui's own README concedes the trade-off he is pressing on.
- 02OpenAI Says Codex Cannot Run in Docker for iOS Builds, and Docker Launches It With Codex's Sandbox OffBolin's November 18 post lists Seatbelt, Bubblewrap and MXC as Codex's three sandboxes; Docker's page defaults to --dangerously-bypass-approvals-and-sandbox.
- 03CLI-Anything Has 62,119 Stars and a README With Three Different Test TotalsThe HKUDS project turns desktop software into agent-callable CLIs; its own per-harness lines sum to 158 more tests than it reports.
- 04Brazil's Receita Federal Posts Tax Calculator Under MIT With 16 Open IssuesThe agency published three repositories on Nov. 16 for the new consumption taxes, and outside developers had filed bug reports against the first within days.