Avoiding Guava

[futurice]

Avoiding Guava

[staltz]

We don't see Guava as bringing enough value into the code base, so given its size, we prefer to avoid it.

We'll write about multi-dexing soon in the future.

[jaggernod]

Multidexing on older platform is going to be quite slow.
We need to decide how fast should a debug build be. If we want to build multiple flavors of a big project then we might as well grab a coffee or rather 3 course dinner.

But I agree that getting rid of a big utility library just due to 65k problem seems rather silly.

[jonikarppinen]

For what it's worth, admittedly there's no real consensus about Guava on Android among Futurice developers.

I for one used Guava in one of our customer projects and was completely happy with that choice (in that case the app was not huge so we were not close to 65k limit).

So I wouldn't use a statement as strong as "avoid Guava" myself, and depending on the needs, I'd consider using Guava also in the future, going multidex if necessary. (But if I'd be mostly using stuff like Lists.newArrayList(), that hardly justifies adding the lib. :-) )

On top of that, I think Proguard will strip out any methods you don't end up actually using there

But you won't use ProGuard on e.g. debug builds, so you do need take take the largeish method count into consideration. Anyway, agreed that we should cover multidexing in the guide. (Maybe I'll do it if I find the time to look into it.)

[AlexanderKaraberov]

I agree with @pimterry that Guava is not an issue. It offers a lot of goodies, which simplify development and I used for a lot of projects - the code become clearer with it (if use it reasonably of course). Maybe when we will have Java 8 Streams we can avoid Guava.
With the help of ProGuard and now Multidex support 65K will not be an issue.
I posted a question more than two years ago on stackoverflow: http://stackoverflow.com/questions/14978699/is-it-a-good-idea-to-use-google-guava-library-for-android-development about the same issue.

[peter-tackage]

I didn't really see the value in Guava, given the pressure it put on method limits. However, a few months back I saw a talk that demoed an app that showed it's strengths and I was fairly impressed. I think the key is that if you use Guava, use it thoroughly and get real value from it; it's a good, reliable library.

With the somewhat recent modularization of Google Play Services, the pressure on method limits has eased somewhat, making the benefits of using Guava less of a trade-off.

[peter-tackage]

Closing this discussion due to inactivity.