fix(swingset): do not record GC syscalls in the transcript #3148
Conversation
| @@ -26,7 +26,13 @@ export function makeTranscriptManager( | |||
| }; | |||
| } | |||
|
|
|||
| const gcSyscalls = new Set(['dropImports', 'retireImports', 'retireExports']); | |||
There was a problem hiding this comment.
I think I'd label this unrecordedSyscalls or something like that. Today it's GC but conceivably in the future there could be other things that fall into this category.
There was a problem hiding this comment.
Good point, but I think it's slightly more likely to move a different direction: we don't record them during the initial execution, but we still collect something about them during replay (to deal with possible early dropImports). So "unrecorded" won't quite capture it.. the code on the if X.include(syscallType) side will want to recognize GC syscalls and do something special with them, record/not-record/collect. So my hunch is that we'll continue to mark this particular set as "GC syscalls", but the mapping (or not) of "GC syscall" to "don't record" will be made explicit at the point of use.
Consensus mode will depend upon GC being deterministic, but solo mode does not. Solo mode requires GC be "sufficiently deterministic", which means a finalizer may or may not run in any given crank. To support this, we must not record the GC-related syscalls (dropImport, retireImport, retireExport) in the transcript. When replaying a transcript, we ignore these syscalls as well. closes #3146 refs #2615 refs #2660 refs #2724
warner
force-pushed
the
3146-no-gc-transcript
branch
from
67c416c
to
cfb5a33
Compare
Consensus mode will depend upon GC being deterministic, but solo mode does
not. Solo mode requires GC be "sufficiently deterministic", which means a
finalizer may or may not run in any given crank.
To support this, we must not record the GC-related syscalls (dropImport,
retireImport, retireExport) in the transcript. When replaying a transcript,
we ignore these syscalls as well.
closes #3146
refs #2615
refs #2660
refs #2724
refs #3106