util: fix formatting of objects with SIMD enabled

[addaleax]

Checklist

• make -j4 test (UNIX), or vcbuild test nosign (Windows) passes
• tests and/or benchmarks are included
• commit message follows commit guidelines

Affected core subsystem(s)

util

Description of change

When SIMD is enabled, util.format couldn’t display objects (with at least 1 key) because the formatter function got overridden.

[bnoordhuis]

For robustness and performance reasons (no double lookup), it might be better to set the formatter in the else block around line 510, but LGTM either way.

[addaleax]

@bnoordhuis Updated, if I understood your suggestion correctly.

CI: https://ci.nodejs.org/job/node-test-commit/4267/

[bnoordhuis]

Not quite what I had in mind but your approach is a bit better still. LGTM.

[bnoordhuis]

New CI, last one failed with a Jenkins error on Windows: https://ci.nodejs.org/job/node-test-pull-request/3421/

[jasnell]

oh, thank you for this, I'd noted this down a while back but never got around to it! This LGTM!

[targos]

Shouldn't we run the test with --harmony-simd flag?

[addaleax]

@targos I’ve added some tests from test-util-format.js to test-util-inspect-simd.js, too. Does that work for you? You’re making a good point but I think I’d rather have both code paths tested (SIMD on/off).

[cjihrig]

LGTM. Updated CI: https://ci.nodejs.org/job/node-test-pull-request/3483/

[jasnell]

Build bot failure on Windows in the last CI run, running again just to be safe: https://ci.nodejs.org/job/node-test-pull-request/3487/

[targos]

LGTM

[addaleax]

That last CI failed completely, new attempt: https://ci.nodejs.org/job/node-test-commit/4370/

[addaleax]

Landed in 1b24b37, thanks for the reviews!