Closed Bug 561368 (interp_jit_semantics) Opened 10 years ago Closed Last year
Interpreter and JIT execution semantics must be identical
Tracking bug. Interpreter (wordcode and abc) and JIT (debugger or non, with any combination of optimization modes or runmodes) must always behave identically, except for performance/memory/stack use. Unfortunately, this is currently not the case, and must be fixed before we can employ any execution policies that dynamically switch between modes.
Priority: -- → P2
Target Milestone: --- → flash10.2
Target Milestone: flash10.x - Serrano → flash10.2.x-Spicy
Adding Dan Sc. All skips and expected failures for interp in testconfig must be scrutunized to see if there are other bugs that belong in the depends list.
I verified the skips and expected failures in testconfig.txt. Only bugs 456852 and 551587 show differences between jit and interp in the acceptance tests. The other skips and expected failures are for other reasons.
Target Milestone: flash10.2.x-Spicy → flash10.x-Wasabi
OSR for all platforms isn't happening until Nigel, so this moves out too.
Target Milestone: flash10.x-Wasabi → flash11-Nigel
Target Milestone: Q1 12 - Brannan → Q2 12 - Cyril
Status: ASSIGNED → RESOLVED
Closed: Last year
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.