Ported to NAN 1.2.0

[kkoopa]

This addresses #252. Uses NAN 1.2.0. Tests pass on 0.10.28 and 0.11.13.

[springmeyer]

wow wow wow @kkoopa this is incredible. thank you!

A few questions right off:

• mixed use of NanNew<string>("string") and just NanNew("string"). Is there a reason for this?.
• v8::String::NewSymbol is not used by NanNew<String> right? If I'm understanding right I assume this is because NewSymbol is not recommended anymore given nodejs/node-v0.x-archive@f674b09?
• (args.This()); -> (args.Holder());. Thanks for making these changes - from my understanding (mostly from ObjectWrap/NODE_SET_PROTOTYPE_METHOD/node::SetPrototypeMethod is unsafe nodejs/node-v0.x-archive#6690) its always best to use Holder - is that right?
• ->ToObject() -> .As<Object>(). So .As works with older node and is best practice? I've noticed that either could be used but never paid close attention.
• Color::constructor->HasInstance(obj) -> NanNew(Color::constructor)->HasInstance(obj) - can you tell me more about why this is needed?

[kkoopa]

• No reason, just mix of manual porting and automated search and replace.This particular overload of NanNew will be introduced in 1.1, it is just a shorthand for a common use case.
• Correct, NanSymbol will be deprecated in 1.1 for those precise reasons and because String::NewSymbol is no more in node 0.11. NAN 1.0.0 mapped it to String::New for node 0.11, but we decided to deprecate and remove it instead.
• I don't really know myself. I changed those with ObjectWrap::Unwrap to use Holder(), but others still use This()
• As<T>() is slightly faster, as it skips some checks that T::Cast() performs
• Local<T> NanPersistentToLocal(const Persistent<T>&) was merged into Local<T> NanNew(const Persistent<T>&)

[springmeyer]

excellent, thank you - all sounds good. So, do you have a timeline on the next Nan 1.1.0 release?

[kkoopa]

@rvagg promised a release earlier this week, maybe he can find time to do it tomorrow?

If you would like to try it out right now, you can change it to depend on NAN master.

[springmeyer]

Tested against Nan master and it works perfectly. Will plan to merge this once a Nan release is out - but no rush on that. Thanks again!

[rvagg]

yes, I've just settled down in Melbourne for the weekend and this is near the top of my list. release coming soon!

[springmeyer]

Looks like 1.2.0 is out? nodejs/nan@c12da50. @kkoopa Any adjustments needed before merging?

[diorahman]

Tried to "merge" the last commit, with this PR to my own fork: https://github.com/rockybars/node-mapnik/compare/nan-1.2.0 but it couldn't pass at following tests:

  1) mapnik.VectorTile  should be able to clear data (sync):
     Error: serialization failed, possible race condition
  2) mapnik.VectorTile  should be able to clear data (async):
     Uncaught Error: serialization failed, possible race condition
  3) mapnik.VectorTile  should be able to add data:
     Uncaught Error: serialization failed, possible race condition
  4) mapnik.VectorTile  should be able to push an image tile directly into a vector tile layer without decoding:
     AssertionError: 11687 == 12146
  5) mapnik.VectorTile  should include image in getData pbf output:
     AssertionError: 11687 == 12146

The build job:
https://travis-ci.org/rockybars/node-mapnik/jobs/27408391

[kkoopa]

No adjustments needed, went ahead and changed the used version in package.json

All tests pass for me.

[springmeyer]

merged into master now :) Thank you!