ci(release): build the linux prebuilds on ubuntu 22.04

[GuillaumeLagrange]

Build the Linux prebuilds of @codspeed/core on Ubuntu 22.04, for both x64 and arm64.

In 6.0.0 the arm64 prebuild moved to an ubuntu-26.04-arm runner, whose GCC 14 libstdc++ adds a CXXABI_1.3.15 requirement, so the addon fails to load on older distros such as Ubuntu 22.04. The arm64 prebuild now runs on ubuntu-22.04-arm, and the build job, which produces the x64 prebuild, is pinned to ubuntu-22.04 instead of ubuntu-latest, so a future runner bump cannot hit x64 the same way.

Published as 6.0.1-alpha.1 from this branch. Symbol versions required by the Linux prebuilds:

prebuild	6.0.0	6.0.1-alpha.1
linux-arm64	CXXABI_1.3.15, GLIBCXX_3.4.21, GLIBC_2.17	CXXABI_1.3.9, GLIBCXX_3.4.21, GLIBC_2.17
linux-x64	CXXABI_1.3.9, GLIBCXX_3.4.21, GLIBC_2.17	CXXABI_1.3.9, GLIBCXX_3.4.21, GLIBC_2.17

The alpha also passed our integration benchmarks with @codspeed/vitest-plugin.

Closes #96
Closes COD-3881

[greptile-apps]

Confidence Score: 5/5

[High impact] The PR appears safe to merge; no actionable issues were found.

Summary

The release workflow now builds both Linux prebuilds on Ubuntu 22.04.

• Linux native prebuilds now come from Ubuntu 22.04 runners.

Diagram

%%{init: {'theme': 'neutral'}}%%
flowchart LR
  X["Ubuntu 22.04: x64"] --> A["Upload prebuilds"]
  R["Ubuntu 22.04: arm64"] --> A
  M["macOS: arm64"] --> A
  B["Build libraries"] --> D["Delete local prebuilds"]
  D --> F["Download and merge prebuilds"]
  A --> F
  F --> P["Publish packages"]

Loading

_{Reviews (1) · Last reviewed commit: e9f99ab · Reviewed by Greptile}

[codspeed]

Merging this PR will regress 5 benchmarks

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚠️ 14 benchmarks spent significant time in system calls

System calls cannot be consistently instrumented, so they are not included in the measure, which understates the real cost. Please switch to the Walltime instrument to accurately measure system calls.

Measurement and system calls

• switch 1
• wait 500ms
• wait 1sec
• switch 1
• wait 1ms
• wait 1ms
• wait 1sec
• test_recursive_fibo_10
• switch 1
• test_recursive_fibo_10
• …and 4 more

ℹ️ 25 benchmarks have no flame graph

The flamegraph could not be generated because too much time was spent in runtime internals, making the visualization not useful for performance analysis.

• switch 1
• switch 1
• wait 1ms
• switch 2
• test_recursive_cached_fibo_10
• test sync baz 100
• test_iterative_fibo_10
• test_recursive_fibo_10
• switch 2
• test sync baz 10
• …and 15 more

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 10 improved benchmarks
❌ 5 regressed benchmarks
✅ 249 untouched benchmarks
⏩ 56 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	WallTime	test sync baz 10	120 ns	144 ns	-16.67%
❌	WallTime	test_iterative_fibo_10	120 ns	144 ns	-16.67%
❌	WallTime	test sync baz 100	216 ns	252 ns	-14.29%
❌	WallTime	switch 1	72 ns	84 ns	-14.29%
❌	WallTime	switch 2	84 ns	96 ns	-12.5%
⚡ ⚠️	Simulation	switch 1	1,075.9 µs	245 µs	×4.4
⚡	WallTime	fibo darwin	9.3 ms	5.2 ms	+78.35%
⚡	Simulation	short body	242.5 µs	153.6 µs	+57.85%
⚡	Simulation	short body	242.7 µs	154 µs	+57.62%
⚡ ⚠️	Simulation	wait 500ms	49.3 ms	36.2 ms	+36.23%
⚡ ⚠️	Simulation	wait 1sec	89.3 ms	72.9 ms	+22.45%
⚡	Simulation	long body	639.6 µs	548.5 µs	+16.6%
⚡	Simulation	long body	639.6 µs	548.8 µs	+16.56%
⚡	WallTime	switch 2	96 ns	84 ns	+14.29%
⚡ ⚠️	Simulation	switch 1	269.7 µs	245.1 µs	+10.04%

Tip

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

---

_{Comparing cod-3881-600-fails-on-linux-arm64-with-older-libstdc-cxxabi_1315-not (8110099) with main (f08a6b7)}

Footnotes

1. 56 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩