Support for members with accessors in class_

[yonatan-mitmit]

The PR contains the following changes:

1. Support for const members pointers (as an overload of set_const). This requires special handling as the setter cannot compile and the readonly runtime check wasn't sufficient
2. Support for members with special accessors (a class containing a property which has itself special setter/getter). The more natural behavior is that X.y = z will invoke X.y.setZ(). Previous syntax forced using (n JS X.y.z or X.y.setZ() )
3. Support for "chainable setters" these return T& instead of void and support a "fluent" syntax X.setY().setZ(). In JS they are treated as properties and thus are not chainable, but it doesn't require library changes to get wrapped.

P.S.
Thanks for the awesome library

[yonatan-mitmit]

The tests above fail on old versions of V8. (I'm testing against 7.8.279.23)
Without these changes, the code doesn't compile on latest.
What's the version you want supported?

[pmed]

Hi @yonatan-mitmit

Thank you very much for the contribution! I will review and merge it.

I would prefer to stay with the elder V8 version, as possible. Currently Travis CI runs tests for different V8 versions starting from 6.3, in development I use 7.5.

It seems that V8 has some changes in its API, so the movement to the newer V8 is inevitable. In this case a minimal V8 version should be a reasonable value, e.g. the same as used in Node.js LTS version.

Best regards
Pavel

[yonatan-mitmit]

Added ifdefs for the V8 version. NewFromUtf8 that returns a Local (And not MaybeLocal) was deprecated in 7.6

[pmed]

Hi @yonatan-mitmit,

thanks for the pointing out the issue with v8::String::NewFromUtf8 in v8pp::throw_ex(). It turned out that such an overloaded function returning MaybeLocal exists for a long time.

I've fixed this in context of issue #131, it works with V8 version 7.9. I need to add the newer version to Travis build matrix.

[cr-mitmit]

Thanks the fix.
Can you see re the rest of the PR (Support member's access operator as properties?)

Thanks!

[pmed]

Hi @yonatan-mitmit

Can you see re the rest of the PR (Support member's access operator as properties?)

Sorry for the review delay, I had no spare time. The proposed changes look good 👍

Could you please also add a test case, in order to demonstrate how such a property would be bind in C++ and used in JavaScript?

Best regards,
Pavel