update to nan v2 (adds support for iojs 3)

[defunctzombie]

Nan v2 required more code changes but also adds support for iojs 3 and
hopefully provides a more stable API which will not have to change as
much going forward.

[defunctzombie]

Travis build is failing on iojs v3 but this seems to be unrelated to this repo and caused by errors from the v8 code. Maybe a stricter compiler or some other changes. It builds on my OSX machine so it could also be linux specific. If anyone has a linux machine locally, please give this PR a try on a few versions of node and iojs.

[ruimarinho]

Working fine here, iojs v3.0.0 on OSX (El Cap). Thank you!

[jbergstroem]

@defunctzombie iojs requires gcc 4.8.x or newer (v8 is c++11 only nowadays). People in this issue seems to have gone through the same pains.

Changes LGTM.

[defunctzombie]

Sigh. Why doesn't travis use the correct gcc when you specify iojs :(

Seems like we need to file a new ticket with them saying iojs-v3 support is incomplete because no native modules can build.

[jbergstroem]

@defunctzombie thinking it might rather be choice of distro than jenkins being conservative. Anyway, seems to be a few one-liners to getting 4.8.x running at travis in that issue.

[kkoopa]

While this compiles for now, you still want to change all uses of v8::Value::ToXXX and v8::Value::XXXValue to Nan::To<XXX> https://github.com/nodejs/nan/blob/master/doc/converters.md#api_nan_to, since the former will be removed from V8 in coming versions.

I would also recommend Nan::Utf8String as a drop-in replacement of v8::String::Utf8Value, which handles invalid UTF-8 in a consistent way.

[defunctzombie]

@kkoopa can you give an example from this codebase of Nan::To use. The nan docs are not so clear to me how I would update the following:

const ssize_t rounds = info[0]->Int32Value();

My first stab at it was

const ssize_t rounds = Nan::To<ssize_t>(info[0]).FromJust();

but this doesn't build.

[kkoopa]

Only the listed conversions are available. While int64_t is the closest to ssize_t, the original used int32_t.

[defunctzombie]

ssize_t depends on architecture iirc

So what would be the API to do the above conversion. Examples/docs are
lacking and/or unclear to me how when I would use these and how to work
with maybe objects.

Do I need to consult v8 maybe object docs?

On Monday, August 17, 2015, Benjamin Byholm [email protected]
wrote:

Only the listed conversions are available. While int64_t is the closest
to ssize_t, the original used int32_t.

—
Reply to this email directly or view it on GitHub
#324 (comment)
.

[kkoopa]

They won't say more than what's written at https://github.com/nodejs/nan/blob/master/doc/maybe_types.md#api_nan_maybe

Use Nan::Maybe<>::FromJust if you are certain that it is valid, otherwise, use Nan::Maybe<>::FromMaybe with a suitable default value.

The above conversion is done as
const ssize_t rounds = Nan::To<int32_t>(info[0]).FromMaybe(0);

[defunctzombie]

Thanks, I will give that a try.