Thanks to both of you! Attila: the TypeEvaluator (which is the only thing that will cause such as map change AFAICT) was added after the continuation code in question, so I think it was actually correct back then ;-)
Hannes > Am 30.09.2016 um 18:29 schrieb Attila Szegedi <szege...@gmail.com>: > > +1; good work. I'm kinda embarrassed I didn’t think of this case… > > Attila. > >> On 30 Sep 2016, at 18:18, Marcus Lagergren <mar...@lagergren.net> wrote: >> >> +1. Good catch, Hannes! >> >> /M >> >>> On 30 Sep 2016, at 16:46, Hannes Wallnöfer <hannes.wallnoe...@oracle.com> >>> wrote: >>> >>> Please review 8166902: Nested object literal property maps not reset in >>> optimistic recompilation: >>> >>> Bug: https://bugs.openjdk.java.net/browse/JDK-8166902 >>> Webrev: http://cr.openjdk.java.net/~hannesw/8166902/webrev.00/ >>> >>> The original assumption was that the property map only has to be replaced >>> in the top level object literal when optimistic recompilation happens >>> within a nested object literal. However, recompilation may also create >>> different maps on parent object literals as more information about types >>> may be available. >>> >>> Thanks, >>> Hannes >> >
