Mark Shannon
Mark Shannon
I'm closing this for now. We can revisit splitting up `RETURN_VALUE` when the full refactoring I suggested above is ready. https://github.com/python/cpython/pull/125972 splits up `LOAD_CONST` and removes `RETURN_CONST` and shows slightly...
Actually, don't worry about the merge conflict. I can fix it in my PR
> Given https://github.com/python/cpython/pull/117123, and the performance regression, maybe it makes sense to revert https://github.com/python/cpython/pull/116206 while working on getting this right? Why not merge this PR? Your numbers show that it...
OOI, how much in absolute terms? A relative increase in memory isn't that meaningful. We expect short lived programs to have increased memory use as the collector will lag a...
I'm going to merge this, as it does fix the immediate problem. The memory use for mypy2 is up significantly, but the memory use for "tomli_loads", which also uses a...
https://github.com/python/cpython/pull/117213 should fix it
When allocating a shared version number we need to check that not only is the code object the same, but that the globals, builtins and (number of) defaults is the...
The defaults can be changed via `code.replace` changing either the tuple of defaults or code object passed to `MAKE_FUNCTION`, but not the other. I think @brandtbucher had an example.
The example I was thinking of is [this](https://github.com/python/cpython/issues/117045#issuecomment-2008307301) which is just an assert. We might want to optimize based on the defaults in future, so we might as well check...
I need to add checks for builtins.