add memory leak test

[nickdesaulniers]

continuing from issue #151, at commit cfa6ca7 (version 3.2.3) we know we had a memory leak based on running @deepakprabhakara 's leak test https://github.com/redsift/node-nanomsg-leak-test for ~10-20 minutes.

We should find a way to speed up that test, and check it in to our test suite. By checking out the know bad commit (git checkout cfa6ca76b734d2dfeeabd347f2f83bf0c2761d3f), we can verify the test fails fast, and passes on newer versions, once PR #169 is merged.

[deepakprabhakara]

Should be possible if we reduce the timer interval and also create more rep sockets on a range of ports. And nodejs can be run with the '--expose-gc' flag to call global.gc() manually.

[nickdesaulniers]

I wonder if there's anything we can do to force pressure on the GC, like create a lot of objects?

[deepakprabhakara]

@nickdesaulniers Here's one that leaks within 5 seconds - redsift/node-nanomsg-leak-test@c2835b6

We can do better if needed.

[deepakprabhakara]

Here's a better one - redsift/node-nanomsg-leak-test@7fc0a91

[nickdesaulniers]

Great thanks. I'll try it when I get home. can you add a compatible license or send a PR? On Dec 20, 2016 7:29 AM, "Deepak Prabhakara" <[email protected]> wrote: Here's a better one - redsift/node-nanomsg-leak-test@7fc0a91 <redsift/node-nanomsg-leak-test@7fc0a91> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#170 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABvUXy86UIoALXNMpZTaig1CWw61FXl7ks5rJ8o5gaJpZM4LQab1> .

[deepakprabhakara]

Sure, I'll try to drop it in as a test and send a PR.

[deepakprabhakara]

I've also added an MIT license in the meantime.