It's great that you found it. I don't think I ever used that setting yet, but I have switched to ['forcedName'] for some things, sometimes. Perhaps it is related to some other changes in the overall build for whether the names are 'preserved' for dynamic access based on what other classes use with 'known' member references (guessing only).
On Tue, Jul 2, 2019 at 9:03 AM Harbs <[email protected]> wrote: > Adding js-dynamic-access-unknown-members fixed my problem, but I don’t > know why I needed to add it to TLF all of the sudden. > > FWIW, the problem was the properties in the object lookup in > SpacingLimitPropertyHandler being minified. > > > On Jul 1, 2019, at 11:18 PM, Greg Dove <[email protected]> wrote: > > > > By property, you mean getter/setter accessor? > > If yes, I'm assuming that you've already inspected the js-debug > annotations > > to verify that they have not changed? (still have @lends with @export for > > each ) > > Other than that, I think I'd have to find a way to repro it to help > figure > > it out. I'll keep an eye open for anything like that today. > > > > > > On Tue, Jul 2, 2019 at 8:01 AM Harbs <[email protected]> wrote: > > > >> I’ve been doing a lot of that. > >> > >> Right now, I’m trying to figure out why an object has renamed properties > >> when it didn’t used to. > >> > >> One thing I discovered is that initializing properties causes Google > >> Closure to mess up renaming much more than uninitialized ones. Not sure > why > >> that is. > >> > >> Harbs > >> > >>> On Jul 1, 2019, at 10:19 PM, Greg Dove <[email protected]> wrote: > >>> > >>> 'The latests compiler changes have really taken a LOT of my time' > >>> > >>> Sorry to hear that, I have been there myself in the past. Can I check: > is > >>> this still related to Vector? > >>> The docs also explain ways to switch off the other newer things (such > as > >>> the implicit casts: js-complex-implicit-coercions, for example), in > case > >>> that makes a difference. > >>> > >>> > >>> > >>> > >>> On Tue, Jul 2, 2019 at 6:53 AM Harbs <[email protected]> wrote: > >>> > >>>> Yes. Switching off the index checking seems to help. (partially. I’m > >> still > >>>> trying to track down why my app isn’t working) > >>>> > >>>> The latests compiler changes have really taken a LOT of my time… :-( > >>>> > >>>>> On Jul 1, 2019, at 9:46 PM, Greg Dove <[email protected]> wrote: > >>>>> > >>>>> ' Yes. Not sure how to reproduce t for you.' > >>>>> > >>>>> OK, thanks for confirming. Hopefully switching off the index checking > >>>> works > >>>>> for you for now. I will see if I can repro that either later today or > >>>>> tomorrow. > >>>>> > >>>>> > >>>>> > >>>>> On Tue, Jul 2, 2019 at 6:44 AM Harbs <[email protected]> wrote: > >>>>> > >>>>>> Yes. Not sure how to reproduce t for you. > >>>>>> > >>>>>>> On Jul 1, 2019, at 8:37 PM, Greg Dove <[email protected]> wrote: > >>>>>>> > >>>>>>> Are you seeing the index checking code being generated with > >>>>>>> '-js-vector-emulation-class=Array' set for all your builds? If so, > >>>>>> that's > >>>>>>> a bug, and I'd be keen to repro it and fix. > >>>>>> > >>>>>> > >>>> > >>>> > >> > >> > >
