[IlyasShabi]

What does this PR do?

Removes kSamplingForceGC from allocation profiling on Node.js 26+, so collecting a profile no longer requests a full GC. It also updates the documentation in the code and adds a simple regression test.

Motivation

We identified that allocation profiling forces a full GC every time we collect a profile. This adds GC work and can cause long pauses on the main thread. Removing the flag avoids those extra collections and their pause overhead.

Additional Notes

Some objects that are no longer reachable may still be reported as live because V8 has not collected them yet. This can overestimate live heap usage at collection time. For a statistical profiler, we prefer this tradeoff to forcing a full GC on every profile.

Jira SCP-1457

[github-actions]

Overall package size

Self size: 2.62 MB
Deduped: 3.73 MB
No deduping: 3.73 MB

Dependency sizes

| name | version | self size | total size | |------|---------|-----------|------------| | pprof-format | 2.4.0 | 912.36 kB | 912.36 kB | | source-map | 0.8.0 | 185.66 kB | 185.66 kB | | node-gyp-build | 4.8.4 | 13.86 kB | 13.86 kB |

_{🤖 This report was automatically generated by heaviest-objects-in-the-universe}

[szegedi]

If I understand correctly, this does not even affect the precision of allocation profiles, since they report live + collected and the sum remains the same with or without GC, only the summands change. If anything, this improved the precision of heap live size profiles, since without it it can report samples for unreachable, but not yet collected objects as live, which is how it worked before allocation profiling anyway. Do I see it right?

[IlyasShabi]

@szegedi Yes that’s right! Removing the flag affects the freshness of live heap, not the accuracy of allocation profiles.