Hi Tim, you may want to have a look at
Preserving Instance State during Refactorings in Live Environments https://hal.archives-ouvertes.fr/hal-02541754/document G > El 13 feb 2021, a las 9:46, Tim Mackinnon <tim@testit.works> escribió: > > I never thought of that implication - if ivars are proper objects (is that > now applied?) would this issue go away (an ivar would just get a new parent)? > > This is a very interesting problem, as its a type of transaction, and I never > though of it that way - its easy to think of it as just dumb code - but in > effect it isn’t. > > Tim > > On Fri, 12 Feb 2021, at 4:30 PM, Sean P. DeNigris wrote: >> Wonderful to have progress on this important topic - thank you! >> >> Sorry I haven't been following closely (maybe you addressed it already), but >> pushing up instance variables has a dangerous limitation - instances lose >> any data held in that var. I guess it's because it's deleted from the >> subclass prior to adding to the superclass to avoid duplicating. One >> solution would be to add a var to the superclass with a mangled name, copy >> the data for all instances, remove the var from the subclass, and then >> rename the mangled var. >> >> >> >> ----- >> Cheers, >> Sean >> -- >> Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html >>
