niner

> What about moving the var up a level and initializing to NULL? That would mean that in the error case it would contain the value of the previous iteration....

This seems to left-pad the shorter string. Docs and design docs just say to pad the string. But somehow I'd have expected it to right-pad. How do we know which...

> My subsequent changes assume that the pointer generated by the expression > `STABLE(type)` can't change (outside of the GC) - is it correct that > `st` is never reassigned?...

At least, whatever stable we read at this point, it will still point at allocated memory and writing to debug_name is safe. Considering how rebless and setdebugtypename are used, I'd...

I can confirm that this does indeed break packaging MoarVM. I have commited the proposed fix. I would like to keep this issue open however to discuss how we could...

Installing into /usr is actually already covered by the OBS builds in https://build.opensuse.org/project/show/home:niner9:rakudo-git (for which we just lack the Github notifications).

> If they can call now, I wonder if also `deoptallpoint` applies? Wouldn't the same be true then for assign and decont which can also call?

MVM_SPESH_DISABLE always also disables the JIT, as the JIT is just a stage of spesh processing. It's a bit confusing that in your backtrace the offset and data arguments are...

@vrurg https://github.com/MoarVM/MoarVM/commit/2abb469945e08ea1b677f49ecc403be7e23a93a2 fixes the segfault but there's still a bug in the raku-smartmatch dispatcher. It uses dispatcher-replace-arg-literal-obj to replace a native int argument with an obj and that is forbidden.

@cygx you still working on this?