You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
makeConvertKit will need additional parameters makeHostForGuestRemotable and makeHostForGuestPromise. These create host wrappers that forward to the corresponding guest objects just like makeGuestForHostRemotable and makeGuestForHostVow make guest wrappers that forward to the corresponding host object.
The magic step is how the existing mechanisms should (UNTRIED) repopulate the bijection on replay, since the host wrapper that will need to wrap a guest object is already in the log, for example as the argument of a checkCall. When the guest performs this call again during replay, the equate step of the replay will reassociate these in the bijection, assuming neither is already in the bijection mapped to something else. If it is, then this is a behavior difference that causes replay to fail. But BEWARE it is a subtle one that may be hard to debug and upgrade to fix.
The extra log opcodes this will need are the duals of the current opcodes. The current codes are doFulfill, doReject, doReturn, doThrow, checkCall.
The new opcodes would be checkFulfill, checkReject, checkReturn, checkThrow, doCall.
If #9299 has already happened (likely), then it will have added checkSend,
requiring this one to similarly add doSend.
The text was updated successfully, but these errors were encountered:
See #9097
makeConvertKit
will need additional parametersmakeHostForGuestRemotable
andmakeHostForGuestPromise
. These create host wrappers that forward to the corresponding guest objects just likemakeGuestForHostRemotable
andmakeGuestForHostVow
make guest wrappers that forward to the corresponding host object.The magic step is how the existing mechanisms should (UNTRIED) repopulate the bijection on replay, since the host wrapper that will need to wrap a guest object is already in the log, for example as the argument of a
checkCall
. When the guest performs this call again during replay, theequate
step of the replay will reassociate these in the bijection, assuming neither is already in the bijection mapped to something else. If it is, then this is a behavior difference that causes replay to fail. But BEWARE it is a subtle one that may be hard to debug and upgrade to fix.The extra log opcodes this will need are the duals of the current opcodes. The current codes are
doFulfill
,doReject
,doReturn
,doThrow
,checkCall
.The new opcodes would be
checkFulfill
,checkReject
,checkReturn
,checkThrow
,doCall
.If #9299 has already happened (likely), then it will have added
checkSend
,requiring this one to similarly add
doSend
.The text was updated successfully, but these errors were encountered: