Andrea Giammarchi

if that's too much or too heavy there's the lighter way (hopefully lighter than the `[obj, undefined]` array): ```js const dropHandler = { dropped: new WeakSet, get(obj, name) { if...

P.S. latest way (which I like more) allows you to check: ```js Module.finalizationRegistry.unregister(wrapper); if (!dropHandler.dropped.has(obj)) { dropHandler.dropped.add(obj); _Py_DecRef(obj); } ``` this guards against possible race conditions between the wrapper and...

btw, since we are here, I see the same issue in here https://github.com/pyodide/pyodide/blob/4860848a72ba793d96e41d89ef15f995804e9f70/src/core/pyproxy.ts#L1066 it looks to me the assumption was that `register(observed, info, keyToUnregister)` has `info` as an array that...

last comment ... if you change your finalization registry callback to accept just `obj` and you keep track of the already decreased counted `obj` references in your code, the Python...

to whom it might concern, as this is a common topic these days, I've created an extremely thin utility which goal is to help simplifying everything for various uses cases,...

FYI we're trying to see if an intentionally broken artifact could fix the current state but I am worried if you fix this bug instead such artifact will suddenly break...

P.S. I can confirm the artifact works via `import('https://cdn.jsdelivr.net/npm/@pyscript/core@latest')`

> it's a known limitation of the default files *unpkg* simply redirects to the entry point which works as expected so this seems to be more a choice (avoid redirect)...

P.S. `+esm` breaks in all possible ways and we're creating our own optimized dist folder that should never need extra changes on top ... but we got it solved with...

https://github.com/WebReflection/hyperHTML/issues/413#issuecomment-1497659474