typecast to uv_async_cb to fix node-webkit compile

[springmeyer]

• Fixes node-webkit compile error TryGhost/node-sqlite3#387
• Unblocks io.js release of node-sqlite3 - node-sqlite3 failing on IOJS TryGhost/node-sqlite3#385

[rvagg]

oh, not so great that we let this slip in, we obviously don't have test cases for this. I'm 👍 on this change and pushing a patch-release for 1.5. While we wait for additional input @springmeyer can you have a quick look at the tests to see why this isn't flagged anywhere?

[bnoordhuis]

I suspect a bug in the NAUV_UVVERSION macro because the NAUV_WORK_CB macro is supposed to take care of the differences in uv_async_cb prototypes. @brett19? I believe you originally wrote that?

[springmeyer]

Just confirmed that all of the crazy node-sqlite3 build matrix works with this patch: https://travis-ci.org/mapbox/node-sqlite3/builds/47314165

@springmeyer can you have a quick look at the tests to see why this isn't flagged anywhere?

Okay, taking a look. As a byproduct see #240 which should help those new to running the tests like me.

[springmeyer]

I've come up short on a reduced testcase to replicate/prevent regression. But I've found that it is easy to replicate within the nan directory just by doing:

npm install nw-gyp
./node_modules/.bin/nw-gyp rebuild --directory test --target=0.11.5

Is it reasonable for the nan .travis.yml to test against node-webkit to be able to catch more subtle breakages like this or do you need something more reduced?

[rvagg]

I don't mind using nw-gyp as a devdep if it's going to catch more, I just don't know anything about nw-gyp at all

[springmeyer]

Okay, new pull request testing node-webkit at #241 (idea is we can watch the travis build there to make sure it fails).

[kkoopa]

The declaration of AsyncProgress_ comes after this, maybe that's why it's whining? The symbol is undefined at that point.