Created attachment 571905 [details] [diff] [review] NOTENEW There's no up-front indicator of the stacky start of a JSOP_NEW like there is JSOP_CALL* for JSOP_CALL. We already annotate every argument to a call with a bytecode, so this patch just adds one to indicate the start of a new expression -- this makes it possible to track opened calls as a stack.
Created attachment 571907 [details] [diff] [review] NOTENEW Forgot the null NOTENEW handling in the IonBuilder to prevent aborts!
Comment on attachment 571907 [details] [diff] [review] NOTENEW Oh noes, I just found that |null()| just pushes null and calls it, so calls are not as book-endy as I thought they were. Need to re-evaluate this approach.
This could be a reason for making decisions at the point of the call, rather than the callee push.
Status: ASSIGNED → RESOLVED
Last Resolved: 7 years ago
Resolution: --- → INVALID
You need to log in before you can comment on or make changes to this bug.