Created attachment 648926 [details] [diff] [review]
IIUC, the result of all this wrapper-purity work in typed arrays is that we don't need this goofy proto-chain walk in getTypedArray:
- for the js::Class hooks (like obj_getProperty), 'obj' is the typed array (even with wrapped objects, do you know of any exceptions bholley?)
- for the natives, CallNonGenericMethod does the right thing
- JSPropertyOp was the remaining case where the given obj is the 'receiver', not the object holding the property, but sfink magnanimously converted them to be accessor properties so the generic accessor mechanism sorts it out
Any other corner cases I'm missing? Passes jit tests which I believe hold tests for all the issues we've seen in the last few months. Try'ing now.
Comment on attachment 648926 [details] [diff] [review]
Thanks! I meant to file a followup bug to do exactly this, but never got around to it. Probably because I didn't understand the dependency chain, which is a bad excuse.