* [#615] Refactor lightwalletd client
This moves the lightwalletd client code to a separate Gradle module, but does not yet do much clean up the public API (e.g. hiding generated GRPC objects). That will be a followon change via #673
I’ve prototyped a safer version of the API that would be implemented for #673 for two specific calls: obtaining the current block height and obtaining the light wallet info. These were the easiest endpoints to update and they provide a useful exploration for the future changes needed.
* Fix benchmarking for networking module
- Moved to fixture and build type check components to the new networking module, so it's accessible from all needed parts of the projects
- Changed fixture return type to fit to all needed usages
* Align with previous review comment
* Fix wrong merge
* Add basic sanity test
- To satisfy tests requirements on emulators.wtf service
* Remove grpc dependency from sdk-lib module
* Repackage cash.z.wallet.sdk.rpc to cash.z.wallet.sdk.internal.rpc
* Fix BuildConfig path
* Update demo-app paths to rpc
* Fix broken grpc services locations
- Our aim here is to change only the local services location (package name), the server location can't be changed due to backward compatibility.
* Enhance GRPC Response model
* Adopt enhanced GRPC response model
- Adopted in a few endpoints
* Enhance Connection failure type
* Add simple fixture test
* Refactor fetchTransaction() to work with Response
- Refactored this server call to adopt our new Response mechanism
- GrpsStatusResolver.resolveFailureFromStatus() now accepts Throwable input
- Added Response.Failure.toThrowable() to satisfy subsequent server calls results processing
- A few other minor refactoring changes
* Remove commented out log
* Unify return type of collection returning methods
- Switched fetchUtxos() to return Sequence instead of List
- Added a check on empty tAddress input
* fetchUtxos returns flow
- Switched fetchUtxos() to return Flow of Service.GetAddressUtxosReply
- Internally it calls getAddressUtxosStream() instead of getAddressUtxos() from GRPC layer
* Update unsafe suffix documentation
* Address several minor change requests
* Remove code parameter
- Removed from the locally triggered failures with default codes.
* Rename local variable
* Switch from empty response to exception
- Our server calls now rather throw IllegalArgumentException than return an empty response
- Removed commented out log message
- Updated javadocs
* Update proto files
- Plus related api changes adoption
* Unify new clients instances name
* AGP 7.4.0 changes
- packagingOptions -> androidComponents in sdk-lib and lightwallet-client-lib modules
* SDK version bump
* Response resolver tests
* Release build upload timeout increase
* Release build upload timeout increase fix
* Revert timeout
- As Github has some infrastructure troubles and we need to wait
* Add migrations documentation
* Sort packaging filters
* Remove private field from public documentation
* Hide private variables
* Remove package from Android Manifest
* Throw exception instead of custom error
- This step unify our approach to validation on client side across all server calls
* Replace setAddresses index const with number
* Fix indentation in proto file
---------
Co-authored-by: Honza <rychnovsky.honza@gmail.com>
The Android and Rust code have previously managed joint custody of the derived data database. With more complex migrations now required, we need to make the Android side read-only. To achieve that, the Android side will remove Room and instead rely on more primitive SQLite APIs for read-only access.
As part of implementing this change, database management throughout the SDK is being refactored.
There will be multiple representations of the data:
- Database representation (and some Entity representations in the places that Room hasn’t been removed yet). These representations are not as type safe and don’t match Kotlin best practices in all cases.
- Once #615 is implemented there will also be network representations
- Type safe models, which often appear in the public API The Database and Network representations will be converted to and from the type safe model representation.
This updates our templates to be more consistent with the Secant app repository.
Note that the checklist item for code coverage was removed for now, until we improve the test infrastructure for the SDK.
This also addresses a security reminder, by explicitly stating that one shouldn't run the demo app from a pull request until after reviewing the code changes.
Carter Jernigan
Honza Rychnovsky