[gouarin]

Checklist

• The title and commit message(s) are descriptive.
• Small commits made to fix your PR have been squashed to avoid history pollution.
• Tests have been added for new features or bug fixes.
• API of new functions and classes are documented (no public API changes).

Description

The svector move constructor, move assignment and swap are noexcept, but they allocate when the vector holds more than N elements. Both copy into a new heap buffer, so a std::bad_alloc calls std::terminate. Moving an xarray with more than four dimensions goes through this path.

Moves now take over the heap buffer of the source, together with its allocator. When the source uses its inline buffer, its elements are moved into the current storage, which always holds at least N elements. If *this owned a heap buffer that gets replaced, it is freed with its own allocator first. Self-move assignment is a no-op. Swapping two svectors of the same type goes through moves and does not allocate.

Behavior change: a moved-from svector is now always empty. Move assignment used to leave the source untouched.

svector.move checks that the heap buffer is transferred and the source is left empty; on master it fails on the buffer address. I also ran heap and inline moves, self-move and swaps with a stateful tracking allocator under AddressSanitizer and LeakSanitizer: each buffer was freed by the allocator that allocated it, and nothing leaked.

[codspeed]

Merging this PR will degrade performance by 18.92%

1 warning

⚠️ Unable to measure the execution of 1 benchmark

This benchmark measured no execution time, so its result is not comparable and counts as unchanged. This usually happens when the compiler removes the code under test.

Prevent compiler optimizations

See affected benchmarks

❌ 1 regressed benchmark
✅ 254 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	xshape_access[xt\:\:svector<std\:\:size_t, 4>]	125 ns	154.2 ns	-18.92%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing gouarin:fix-svector-move-noexcept (c92fe75) with master (57f3d07)}