Open GoogleCodeExporter opened 9 years ago
If this is because the app is using AssistedInject, update to the latest SVN
revision and the performance hit will completely go away. The latest SVN
AssistedInject is faster than even the initial implementation of it (and crazy
faster than the second revision of it).
Original comment by sberlin
on 30 Jul 2010 at 6:45
16ms is too slow. We can make this much faster.
Original comment by limpbizkit
on 30 Jul 2010 at 6:46
Original comment by sberlin
on 23 Nov 2010 at 1:30
would love to see this get into 3.1/4.0
Original comment by m...@niskala.org
on 10 Jan 2014 at 8:25
While InjectorBuilder.build() is still slow, child injectors as a whole were
massively optimized back before the v3 release. It's insanely fast now so long
as you aren't injecting a Provider of an assisted arg or an Injector into your
assisted class.
Original comment by sberlin
on 10 Jan 2014 at 8:31
(and by 'child injectors were optimized', I actually meant "assisted inject was
optimized", which was the #1 use-case for child injectors in apps.)
Original comment by sberlin
on 10 Jan 2014 at 8:32
Cool. I'm hoping to use child injectors for separate activities in android,
which sounds like it isn't covered by those existing optimizations
Original comment by m...@niskala.org
on 10 Jan 2014 at 8:35
Interesting. Have you looked at Dagger? It's usually a better choice for
Android.
Original comment by sberlin
on 10 Jan 2014 at 8:48
Yes, I'm aware of dagger. Dagger is okay but it's not as powerful as
RoboGuice. With phones getting faster and faster (many are now faster than the
original macbook air), the flexibility of runtime dependency injection wins out
over compile-time solutions, IMO. As long as we keep optimizing :)
Original comment by m...@niskala.org
on 10 Jan 2014 at 8:52
Fair enough. FWIW, there's not a whole lot left here to optimize that I've
found. Patches are very, very welcome. :-)
Original comment by sberlin
on 10 Jan 2014 at 9:09
will take a look :)
Original comment by m...@niskala.org
on 10 Jan 2014 at 9:35
Original issue reported on code.google.com by
crazybob...@gmail.com
on 30 Jul 2010 at 5:30Attachments: