It's unused right now, Brendan claims fallout from bug 558451. Should it be used, or can it be removed?
We should be able to write a testcase (or find one from the bug that introduced it: bug 496790) and make it bark on a build before the patch-stack for bug 558451 landed. Checking... /be
I added JS_ASSERT(0) to the top of WrapEscapingClosure in jsfun.cpp and ran the test suite. No failures. I think it has been dead since JSOP_UPVAR was removed (bug 592202). JSOP_UPVAR was used in closures on the assumption that they could not escape. Now we emit JSOP_GETFCSLOT or JSOP_NAME in such cases, and those are harmless (though JSOP_NAME gives the wrong answer). If WrapEscapingClosure is dead, GetCallVarChecked is definitely dead, as are the _DBG opcodes. There may be more dead code here, but that's enough for one bug. Taking.
Created attachment 544069 [details] [diff] [review] WIP 1 This passes tests just fine, but I can't think of any sane reason for keeping the needsWrapper assertions around, so I'm going to delete those too.
Created attachment 544077 [details] [diff] [review] v2
Comment on attachment 544077 [details] [diff] [review] v2 Whew, that lightened the load. Should have cleaned this up when UPVAR went, thanks for doing it now. Rather than leave JSOP_UNUSED23 around, I'd just slide down the ops above. /be