Created attachment 196728 [details] [review]
context: Add optional signal emitted after a GC finishes

Embedders may want to perform garbage collection profiling; support
this with an optional signal emitted after a collection finishes.

I added a boolean to avoid calling JS_SetGCCallback() if not used,
because there's no way for that API to chain.

Created attachment 196729 [details] [review]
object: Run JS_updateMallocCounter() when creating GObjects

We need to ensure the garbage collector runs *sometime*.  It's easily
possible to make tiny proxies for large native objects with gjs.  Now,
Mozilla calls JS_updateMallocCounter() in a few places, namely the
Canvas stuff and when creating XPC proxies (roughly their GObject
equivalent).

This patch aims to match the latter one.

Created attachment 196730 [details] [review]
context: Add gjs_context_gc() wrapper

Simple wrapper for JS_GC() avoids simple embedders having to drop to
JSAPI.

Review of attachment 196730 [details] [review]:

Makes sense.

Attachment 196728 [details] pushed as 20cfbeb - context: Add optional signal emitted after a GC finishes
Attachment 196729 [details] pushed as 72bc430 - object: Run JS_updateMallocCounter() when creating GObjects
Attachment 196730 [details] pushed as 059f6a1 - context: Add gjs_context_gc() wrapper

What javascript library is being used that contains the JS_updateMallocCounter() function? One of the few google hits for this function was actually a report to remove this from the public API (https://bugzilla.mozilla.org/show_bug.cgi?id=525372).

Would it otherwise be somewhat ok to wrap the call in a #ifdef JS_updateMallocCounter block?