Daan
Daan
> > This would be my concern as well, although I don't think it's a common thing to have 2 independent versions of a custom allocator in the same process....
@maxbachmann : yes, on unix etc. there is no issue for multiple instances either;
Thanks for the PR -- looks good :-)
Thanks for the quick update -- I'll take a look soon. These things always turn out to have surprising edge cases or portability concerns.
Thanks so much!
Apologies for the late reply but it is fixed now -- the bug was in two place, a missing substitution and a missing check for polymorphic values due to annotations....
Ah I see. I changed the default behavior just before the release -- ouch. Can you try if it is different if you set `MIMALLOC_ALLOW_LARGE_OS_PAGES=1` ? Currently, there are 3...
Ah I see. I will look into this later
- if you use `-no-builtin-malloc` flag in gcc, I believe it turns off all the `__builtin_xxx` functions related to malloc so there won't be a `malloc` matched with `__builtin_free` (but...
Strange. Not sure what is the cause since clang installed. However, Koka invokes `clang-cl` on windows and it might use the Windows Universal C runtime -- maybe that is not...