Performance regression Date.toISOString()

[RafaelGSS]

These are the last results of nodejs-bench-operations

• v18 - https://github.com/RafaelGSS/nodejs-bench-operations/blob/main/RESULTS-v18.md#date-toisostring
• v19 - https://github.com/RafaelGSS/nodejs-bench-operations/blob/main/RESULTS-v19.md#date-toisostring

and as you can see, toISOString might have been affected by a regression.

[RafaelGSS]

And it seems to be there since v19.3.0 (at least) https://github.com/RafaelGSS/nodejs-bench-operations/tree/main/v19.

[anonrig]

We updated to V8 10.8 in v19.3.0. Could this be related?

[c8421204b0] - deps: patch V8 to 10.8.168.21 (Michaël Zasso) nodejs/node#45749

[RafaelGSS]

No. This was caused by a semver-major change. I've adjusted the repository to run the benchmark against 18.0.0 and 19.0.0 and the regression is still there.

• v19.0.0 https://github.com/RafaelGSS/nodejs-bench-operations/blob/main/v19/RESULTS-v19_0_0.md#date-toisostring
• v18.0.0 https://github.com/RafaelGSS/nodejs-bench-operations/blob/main/v18/RESULTS-v18_0_0.md#date-toisostring

[targos]

It's probably the update to V8 10.7

[targos]

Implementation of toISOString was changed in v8/v8@126d477

[anonrig]

Is there an open issue in the V8 bug tracker for this?

[targos]

probably not

[anonrig]

Is the regression still there with V8 11.3? @RafaelGSS

[ronag]

I don't think this is actionable?

[RafaelGSS]

Yeah, it's more of a tracking issue in case someone wants to fix it in the upstream.