Build rclone binaries from source

[joserebelo]

• Adds git submodule with rclone v1.51.0
• Builds rclone from source with build-rclone.py, adapted from syncthing-android

This + #39 should get F-Droid support very close.

Not sure if this approach is what you had in mind, but it should work, at least for now.

[claassistantio]

All committers have signed the CLA.

[x0b]

(edited because I accidentally closed this while writing)

There is a reason why you were never asked to merge your pull request here - while the technical aspects can be reworked, the "legally compatible" is more tricky. The file you based this on build-syncthing.py, is from a MPL 2 licensed repository, and MPL 2 is incompatible with MIT (it is one-way compatible, but in the wrong direction).

The CLA tries to make this clear,

Which third party licenses are okay?

Any work licensed under license terms of which a combined work can still be released under the MIT license.

but it seems we should provide a list of licenses that are actually okay (It can not require anything more than attribution or retaining a copyright notice).

However, since this was asked in #47, I ported my build process to a gradle file that can build the required binaries using go and the clang from the Android NDK.

I have to admit that F-Droid is very far down the roadmap for a few reasons. For one, F-Droid and other non-play distribution channels require enabling unknown sources, which would otherwise protect against malicious apps - and most Android devices don't receive updates, which means that an installed app can easily obtain more privileges, and that on a device that a lot of people carry around almost everywhere they go.

This also affects the speed at which users update - it took months to until users updated from vulnerable app versions, and I can understand that - I don't update my F-Droid that often either. The only reason why I won't cease third party distribution is because that would only support the Google Play store monopoly.

The most important thing for me however is time - working on the app is only one of my hobbies, meaning I have to prioritize what I work on. How many of the (potential) app users are F-Droid users, and of them, how many can't install apps from Google Play? F-Droid does not provide user statistics, but it is hard spending time on this, when there are other things that the other >95% of users will benefit from. Especially if that version also does not supply logging, which is crucial for fast bug triage - without this, I couldn't do anything other than bug hunting.

However, this only means that I personally won't spend much time on this - I welcome any contribution, and would very much appreciate if someone takes the role of a F-Droid distribution maintainer. This also applies to one-time contributions - If this PR (together with #39) would not require such extensive rework, merging it would be a no-brainer.

[joserebelo]

While I understand your points towards not investing time on an F-Droid build, it would be a shame for such a project not to be available there.

About the license, I am sorry. I completely forgot about potential licensing issues, it's not something I had yet ran into, since I don't usually port code across open repositories like this. I will close this for now, and while I don't have a lot of time to work on it in the immediate future, I would be willing to work towards the F-Droid distribution, this time in a more license-complying and possibly technically improved manner.