test: fix flakiness of stringbytes-external

[ofrobots]

Pull Request check-list

• Does make -j8 test (UNIX) or vcbuild test nosign (Windows) pass with
  this change (including linting)?
• Is the commit message formatted according to [CONTRIBUTING.md][0]?
• If this change fixes a bug (or a performance problem), is a regression
  test (or a benchmark) included?

Affected core subsystem(s)

test

Description of change

The tests used to rely on precise timing of when a JavaScript object
would be garbage collected to ensure that there is enough memory
available on the system. Switch the test to use a malloc/free pair
instead.

Ref: #5945
R=@bnoordhuis, @Trott
CI: https://ci.nodejs.org/job/node-test-pull-request/2146/
CI: https://ci.nodejs.org/job/node-test-pull-request/2147/
CI: https://ci.nodejs.org/job/node-test-pull-request/2148/

[Trott]

Interesting. I believe all the other tests in addons are testing native addons. But in this case, you're building an addon so you can get some info to determine whether or not to skip a memory-intensive test that otherwise does not involve a native addon. If we want to go down this path, I wonder if we want to make that clear somehow. Not sure how, though. (And maybe it doesn't matter?) @nodejs/testing

[ofrobots]

Indeed, although my interpretation of addons test was 'anything that needs an addon to test'.

An uglier alternative would be to expose the EnsureAllocation function in node-core to be used by these types of tests only. I really would like to avoid adding this test-only useless APIs to core however.

[jasnell]

This is certainly an interesting case. Overall it LGTM it's just definitely quite a bit different from our other tests.

[gibfahn]

@ofrobots@Trott

my interpretation of addons test was 'anything that needs an addon to test'.

If that's the consensus then presumably the tests README should be updated to clarify that. It currently says:

Tests for addon functionality.

[evanlucas]

LGTM

[jbergstroem]

Would perhaps the cctest suite be a better fit?

[ofrobots]

Adding a new suite for this is probably overkill. I don't think this pattern is going to be very common to require a suite.

I would like to land this PR today as it is blocking #5945 which needs to land very soon in order to get ready for the Node v6 RC schedule.

[Trott]

LGTM. After it lands, we can update the README in the test directory to call out this particular test.

[ofrobots]

Thanks. Landed as f4ebd59. I will update the README in a follow-on.

[thefourtheye]

Closing, as the changes have landed already.

[ofrobots]

Follow-on: #6073.

[MylesBorins]

@ofrobots this is not merging cleanly onto v5.x, would you be able to backport?

[MylesBorins]

since this is not landing on v5.x I'm marking as dont-land-on-v5.x

We will still want this backported to v4.x-staging though