Sloy / android-dependency-injection-performance

[NOT MAINTAINED] Measure the performance of several Dependency Injection frameworks in different devices
313 stars 45 forks source link

documentation: should define what "custom" is #4

Open kenyee opened 5 years ago

kenyee commented 5 years ago

Not sure what this is w/o looking at code. Maybe a short paragraph on how it works in the main readme would be useful.

Sloy commented 5 years ago

You're right, it's not clear at all what it is. It's a quick alternative that we wrote for our production app while the performance issues in Koin were being resolved. I added it to the comparison just to check that nothing was wrong, but it wasn't meant to stay.

I'll give it a thought and I'll see how can I explain it in the readme without giving it too much importance.

coachroebuck commented 5 years ago

Respect ✊

The alternative you wrote happened to be faster than all of the available DI tools, with the exception of Dagger.

Sloy commented 5 years ago

Haha thanks @coachroebuck. But please don't ever use it for production :p It's fast because it's not safe and lack many basic features, We used it as temporary replacement for Koin with it's same API while the performance issues were being resolved.

But @arnaudgiuliani did a great job on performance for Koin 2.0, so it's best to use it instead (or a different one, if you wish).

Nevertheless it was a cool proof of concept :)

coachroebuck commented 5 years ago

hahaha! Thanks for that warning Sloy! I promise I won't hijack the code and use it in production. I don't want defects attached to my name being reported. I have an ego to uphold. hahaha!!

There have been some very heated debates regarding dependency injection at the office. We actually built our own dependency injection tool last year, but I have recently discovered flaws in our tool that I found disturbing. I honestly do not remember how I stumbled upon your repository, but it is awesome! You even provided a clean example for simple-minded people like me of how to using Dagger.

This is an excellent repository in which I not only demonstrated performance to my colleagues, I felt it was necessary to contribute. The data does not lie.