Glibc scanf fixes, Julia LS rework, Swift and Python pitches

Security backports for a glibc scanf overflow lead the day, alongside a Julia language-server overhaul and active language design threads in Swift, Python, and Rust. Git maintainers also pressed contributors on human oversight after buggy AI-assisted patches.

Glibc backports fix scanf %mc one-byte overflow

Glibc maintainers posted backports to the 2.40 and 2.39 branches for BZ #34008, a one-byte buffer overflow in the scanf %mc and %mC allocation path (also tracked as CVE-2026-5450 on the 2.39 series). The patches correct how the dynamically allocated buffer size is computed so the write cannot overrun. Downstream distributors and anyone shipping older stable glibc should treat the fix as a routine security update.

Julia VS Code extension 1.224 pre-release reworks language server

A pre-release of the Julia VS Code extension (1.224) landed with a substantial language-server rewrite aimed at fewer false positives and better performance. The change set is the main topic of a multi-participant thread on the Julia discourse. Editor and tooling users who rely on the extension have an early build to evaluate before the stable channel picks it up.

Swift pitch: same-file memberwise initializers in extensions

A Swift Evolution pitch proposes allowing memberwise initializers to be declared in same-file extensions, primarily so public initializers can be attached to macro-generated structs. The thread examines the visibility and source-compatibility constraints such a rule would impose. Library authors who generate types via macros are the clearest beneficiaries if the pitch advances.

Git maintainers push back on AI-generated patch noise

A patch converting show-branch object.flags usage to a commit-slab drew heated replies after maintainers found the contribution buggy and the follow-ups apparently AI-generated. Discussion turned to expectations of human review and accountability on the Git list. Contributors are being reminded that automated assistance does not relieve them of technical ownership.

PEP 841 frozen-syntax debate continues

An active discuss.python.org thread is working through PEP 841, which would add f{} literal syntax for frozen sets and dictionaries. Proponents argue the form would make immutable collections easier to write and open optimization doors; others question syntax cost and interaction with existing frozen types. The volume of replies shows the idea still lacks consensus.

Glibc dlopen lock release around constructors

A patch for BZ 15686 proposes dropping dl_load_lock while ELF constructors run inside dlopen, addressing reported deadlocks. Reviewers weighed that benefit against possible changes to initialization ordering, and a simpler alternative patch was also floated. The exchange matters for anyone hitting constructor-related lock cycles in heavily dlopen-dependent processes.

Pre-RFC: official LTS for the Rust toolchain

A pre-RFC on Rust internals asks whether the project should publish official long-term support toolchain releases and what the maintenance cost would be. Participants are mapping feasibility, policy, and the burden on release engineering. Teams that need multi-year compiler stability have a concrete thread in which to register requirements.