child_process: allow Infinity as maxBuffer value

[cjihrig]

Checklist

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

Affected core subsystem(s)

child_process

Closes#10767

[sam-github]

+1 for feature, but needs docs and tests across all child_process APIs that support maxBuffer. Doesn't have to test that the behaviour is unlimited, but should test that it doesn't abort or throw arg typerrors.

[jasnell]

Perhaps include a fail test for -Infinity

[cjihrig]

Hm, I'm not sure that JavaScript Infinity will convert to C++ properly here.

[cjihrig]

OK, so JavaScript Infinity translates to std::numeric_limits<double>::infinity() in C++. So, I changed max_buffer_ to a double.

[mscdex]

For consistency, we should probably add the same conditionals to the async counterparts.

[sindresorhus]

Error message needs to be updated to reflect the new constraint (It now accepts any positive number, not just unsigned integer).

[jasnell]

That's a good point. The integer check is still required otherwise we'll have nonsense things like 0.1 accepted as the maxBuffer. The check should likely be something like:

if ((Number.isInteger(options.maxBuffer) || options.maxBuffer === Infinity) &&
options.maxBuffer >= 0) {
...
}

And the error message should certainly be updated but that would upgrade this to a semver-major change

[cjihrig]

0.1 is technically fine as a maxBuffer value, even if it is a bit weird. We're just going to be doing comparisons against the value to see if we surpass it.

And the error message should certainly be updated but that would upgrade this to a semver-major change

This builds on an existing semver major change, so this is going to be semver major regardless.

[jasnell]

That's a good point. The integer check is still required otherwise we'll have nonsense things like 0.1 accepted as the maxBuffer. The check should likely be something like:

if ((Number.isInteger(options.maxBuffer) || options.maxBuffer === Infinity) &&
options.maxBuffer >= 0) {
...
}

And the error message should certainly be updated but that would upgrade this to a semver-major change

[mscdex]

Instead of "non-negative", can we use "positive"?

[cjihrig]

But zero is an allowed, non-positive value.

[mscdex]

It depends on which zero you're talking about ;-)

[cjihrig]

Ha. Ok, I'll make it just "positive"

[cjihrig]

CI: https://ci.nodejs.org/job/node-test-pull-request/5889/

[cjihrig]

Another CI run because Windows: https://ci.nodejs.org/job/node-test-pull-request/5891/

[cjihrig]

Another CI run: https://ci.nodejs.org/job/node-test-pull-request/5911/

[sam-github]

@cjihrig why is this semver-major? Is it because you added validation for some args?

[cjihrig]

Yes.