C++26 Reportedly Fails Its ISO Ballot, With Contracts the Sticking Point

The C++26 draft failed its ISO Draft International Standard ballot in late October 2026, according to reports that would make it the first C++ standard to do so. The cause cited in every account we found is Contracts, the feature specified in paper P2900, and the committee's next chance to respond is its November meeting in Búzios, Brazil.

We could not find an ISO or WG21 record of the result. The vote figures below come from Stork.ai, published October 28, and an AI-summarised aggregator page on Zeli dated October 26, which says it draws on a post at train-it.eu. Treat them as reported, not confirmed.

What the two reports say about the vote

Stork dates the failed ballot to October 24. Zeli's headline says 12 of 20 voting members approved, with two-thirds required. Both say nine bodies voted no, 41% against a 25% cap on disapproval. Stork's own denominators differ, 20 participating members for the yes count and 22 member bodies for the no count, and it does not name a single country.

Source	Date	Reported tally
Stork.ai	October 28	12 of 20 yes; 9 of 22 no (41%)
Zeli, citing train-it.eu	October 26	12 of 20 yes; 9 countries no
Herb Sutter, March plenary	March 2026	114 yes, 12 no, 3 abstain

A Zeli commenter lists the no votes as Denmark, Finland, Iran, Italy, Japan, Luxembourg, Romania, Spain and Sweden. That is one reader's list on an aggregator page and is not verified.

How C++26 got to the ballot

In his March 2026 trip report, Herb Sutter wrote: "News flash: C++26 is done!" He added that "the ISO C++ committee still wants contracts, and so contracts have stayed in C++26." The plenary vote to finalise the standard was 114 in favour, 12 opposed and 3 abstaining, which he tied mainly to continuing concern about Contracts.

1. In favour114 votes
2. Opposed12 votes
3. Abstained3 votes

Source: Herb Sutter, March 2026 trip report, accessed 2026-11-12

That ballot was of committee delegates. The October one was of national bodies, which is a different electorate. On August 15, Andre Kostur asked on the std-proposals list whether P2900 was "in the DIS that's currently out for national body ballot," and whether reversing it would mean convincing "a number of countries to vote down C++26 at this point." The message shows the vote-no route was being discussed ten weeks before the result.

The case against Contracts

A WG21 paper, P4238R0, titled "Returning C++26 for the Evaluation It Skipped," lists Vinnie Falco, Ville Voutilainen, José Daniel García Sánchez and John Spicer as authors, according to the PDF's metadata. We could not read its body text, so we cannot say what it argues beyond the title.

Stork reports that Bjarne Stroustrup co-authored a related paper, P4334R0, with the same four names, and quotes it calling Contracts an "existential threat to C++." Stork says critics object that the feature added about 56 pages to the draft, that runtime checks can be disabled in production builds, and that its inheritance model is unproven. Supporters, it says, see runtime contracts as a practical way to check assumptions.

Because the ballot covered the whole draft, Stork says static reflection and std::execution are delayed along with Contracts. It lists the options as revising Contracts, removing it, or a narrower compromise. A paper reportedly numbered P4433R0 argues for rejecting removal demands without new technical information, but we did not fetch it.

For other languages, see our reports on Microsoft Forge finding soundness bugs in the Rust standard library and a Rust driver for DisplayLink hardware.

WG21 meets in Búzios in November, and Sutter's March report named that meeting as where C++29 work would begin. Whether it opens with C++26's ballot comments instead is the open question.