Issue #21722 has been updated by ivoanjo (Ivo Anjo).
@ivoanjo (Ivo Anjo) I don't know if this would work, but what about using rb_postponed_job_trigger to delay adding the object into a WeakMap? (there would also probably need to be a temporary buffer used to mark the object until it is added)
That's a good question. I know when we started off to make this work we were still trying to support old Rubies where `rb_gc_force_recycle` was even a thing so that factored into the design. (We've since abandoned all hope there and support 3.1+ only for heap profiling). There was also not the redesigned `WeakMap`, etc. You're very right in calling this out -- there's a number of assumptions we need to invalidate about why we picked the current design. I'll try it out and report back! I suspect it may take me a bit as the "let's use the object id" assumption is baked pretty deeply in the current design. 😅 ---------------------------------------- Feature #21722: Expose rb_gc_mark_weak API for use in extensions https://bugs.ruby-lang.org/issues/21722#change-115456 * Author: ivoanjo (Ivo Anjo) * Status: Open ---------------------------------------- In https://bugs.ruby-lang.org/issues/21710 it came up that 1. On top of [deprecating _id2ref](https://bugs.ruby-lang.org/issues/15408) on Ruby 4.0, [it's a bad idea to be using object_id from the NEWOBJ tracepoint](https://bugs.ruby-lang.org/issues/21710#note-8) 2. `rb_gc_mark_weak` which would be the alternative for an extension that needs weak reference-like behavior [is not available for extensions](https://bugs.ruby-lang.org/issues/21710#note-9) So I've opened this ticket to request exposing `rb_gc_mark_weak` so it can be used by extensions? --- The [Datadog Ruby profiler](https://github.com/datadog/dd-trace-rb) is currently using object_id and id2ref to implement its "heap profiling" -- that is, we have a NEWOBJ tracepoint, and from time to time (e.g. not for every object), we select an object, and track its lifetime by keeping its id and checking from time to time if it's still alive. We're using this approach instead of: * The FREEOBJ event => Reduced overhead, as we don't need to be called for every object (+ not needing to deal with corner cases of when FREEOBJ may not be called for an object) * WeakMap => Weakmap APIs are Ruby-level and can't be used from the NEWOBJ tracepoint. (Edit I previously claimed the issue was needing the GVL as well which was not accurate -- we have the GVL inside the tracepoint anyway) For our purposes, it would be OK if this API is not "official" -- e.g. if it's one of those that gets exposed as a public symbol but not documented and no promises made for future Ruby releases. -- https://bugs.ruby-lang.org/