takashikkbn / gmail.com wrote:
> r65473 didn't work either. So I reverted that again on r65474.
> 
> I think doing 2 separate complicated things are hard like this, so the refactoring change (migration to ccan) and behavior change ("rb_mjit_unit can either exist in unit_queue or active_units, but not both") should be done separately.

I think the original problem is related to GC, my original patch did not work
and got stuck because it left in_jit in a bad state (and leaked FILE *f).
This is a refined version which works with and without my ccan/list change:

https://80x24.org/spew/20181031191919.26547-1-e / 80x24.org/raw

The ccan/list patch seems fine, but I'll commit the above one, first.

Unsubscribe: <mailto:ruby-core-request / ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>