Linked bindings initialization

[kaero]

Hi!

We've used linked bindings to compile single statically linked node binary for deployment, so i've a question about difference in the initialization process for built-in bindings, linked bindings and native extensions as dynamic libraries.

As far as i can see, linked binding initialization works more like built-in native modules init: module can be called using process._linkedBinding() and initialization function receive v8::String as second argument instead of v8::Object instance of Module.

My question is: is it good idea to hack linked binding initialization to make to behave more like require + DLOpen (create instance of Module and pass it to initialization function) or it's reserved for some internal needs? Maybe where is another, less "private" way to have native extensions as compiled-in modules?

[kaero]

@indutny Hi, Fedor! Can you help me to find anyone interested in this discussion? Is it right place for it?

[indutny]

@kaero I think it is a plain bug, and it should be receiving a module object instead of String.

[indutny]

Please feel free to file an issue for this.

[indutny]

And don't forget to cc me there ;) I'm going to take a look at it.

[kaero]

btw, it's only part my question. Second one is about usage of process._linkedBinding() which looks like private because of leading underscore in the name ;) Is it ok to use it?

[indutny]

I think the second question could be answered in a following way: please file an additional issue and cc me ;)

It probably just needs a documentation listing its signature and action, but it should be quite safe to use. Also, if it will get a doc - we will consider removing underscore.