Alternative to creating references to non-objects

[mika-fischer]

nodejs/node#15289 removed the possibility to create references to values that are not objects.

My use-case is the following. I want to store arguments given to my function persistently. Then some time later I want to call a callback function with those parameters. In the past I just used a vector<Reference<Value>>. Now I don't quite understand how I'm supposed to do it...

As an example, see this test: https://github.com/mika-fischer/napi-thread-safe-callback/blob/e03dda83fb1d2a63fb8332406e2436fc18a63fba/test/tests.cpp#L36-L51

[gabrielschulhof]

How about storing the arguments in a JS array and then creating a reference to the array. Below is the pseudocode skeleton: ```c napi_create_array(); for (index = 0; index < argc; index++) { napi_set_element(array, index, argv[index]); } napi_create_reference(array, &array_ref); ```

…

On Fri, Sep 29, 2017 at 11:49 AM, Mika Fischer ***@***.***> wrote: nodejs/node#15289 <nodejs/node#15289> removed the possibility to create references to values that are not objects. My use-case is the following. I want to store arguments given to my function persistently. Then some time later I want to call a callback function with those parameters. In the past I just used a vector<Reference<Value>>. Now I don't quite understand how I'm supposed to do it... As an example, see this test: https://github.com/mika- fischer/napi-thread-safe-callback/blob/e03dda83fb1d2a63fb8332406e2436 fc18a63fba/test/tests.cpp#L36-L51 — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#149>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA7k0R_0u2jIKmgFqS4_joy9vjAFLR7vks5snK8cgaJpZM4PoYFz> .

[mika-fischer]

OK, fair enough, that's slightly inconcenient, but not too bad...

But just so I understand: There's now no way at all to just hold onto say a string literal across calls into C++?

[gabrielschulhof]

There is. You can convert it to a `char *` and then pass the `char *` to `napi_wrap()`. In general, you need to convert to a native type, store the data in a heap-allocated buffer, and use `napi_wrap()`.

…

On Fri, Sep 29, 2017 at 1:17 PM, Mika Fischer ***@***.***> wrote: OK, fair enough, that's slightly inconcenient, but not too bad... But just so I understand: There's now no way at all to just hold onto say a string literal across calls into C++? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#149 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA7k0X7gRpG7-BD63rsy-k1FGKdGbZkcks5snMPPgaJpZM4PoYFz> .

[mika-fischer]

Sure, I meant keeping the JS string without converting to C++. That seems impossible now, right?

And while converting a string to C++ by itself is not that bad, in cases like this where you get passed a Value and you don't care what it is, you just want to hold on to it for later use, things are a lot more complicated than before if you don't want to use this boxing trick...

[gabrielschulhof]

It is indeed more complicated. Nevertheless, it is necessary, because not all engines support making references to non-object types.

…

On Fri, Sep 29, 2017 at 1:51 PM, Mika Fischer ***@***.***> wrote: Sure, I meant keeping the JS string without converting to C++. That seems impossible now, right? And while converting a string to C++ by itself is not that bad, in cases like this where you get passed a Value and you don't care what it is, you just want to hold on to it for later use, things are a lot more complicated than before if you don't want to use this boxing trick... — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#149 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA7k0Yh1YomF_IR2IaoYrIa4Kv9JZAUWks5snMupgaJpZM4PoYFz> .

[mika-fischer]

Ah I see, that makes sense then. Thanks for the explanations!