Draft for unit-tests

[cmaglie]

This PR aims to provide unit tests that can be run locally on a PC.

This patch is inspired by Aentiger's (😢) work on unit tests for the ArduinoCore-API.
The Serial connection is mocked with a bidirectional Stream.

Currently, only a single call test is implemented, which is very rough and can be optimized. However, my goal is to create a pilot implementation that works.

It uses CMake for building, and the catch.hpp library to run assertions on tests. It should also produce coverage, but I haven't investigated yet how to extract the data and upload it to codecov.
Further instructions can be found in the README in the PR.

[eigen-value]

Hi @cmaglie that's a super useful feature!
I'm experiencing some issues I wanna share, but first of all here's my setup:

• WSL2 Ubuntu 22.04
• cmake 4.0.2
• gcc 11.4.0

I tried to modify the tests a little and:

• If I try to multiply -2.0*3.0 I get:

/home/lucio/Arduino/libraries/Arduino_RPClite/extras/test/src/RPC/test_RPCClient.cpp:27: FAILED:
  REQUIRE( expected[i] == got[i] )
with expansion:
  '�' == '@'
with messages:
  Test case [RPCClient::call] failed at line 63
  Array size: expected: 27, got: 27
  Index 10: expected 192, got 64

• If I use the MALFORMED_CALL_ERR (0xFD) as error code instead of 1 in the missing parameter case I get:

/home/lucio/Arduino/libraries/Arduino_RPClite/extras/test/src/RPC/test_RPCClient.cpp:99: FAILED:
  REQUIRE( rpc.lastError.code == 0xFD )
with expansion:
  -3 == 253

It looks like a sign conversion problem maybe related to the typecast the StreamMock class is doing (?)

The library has no issues regarding error codes or float sign

[cmaglie]

• If I try to multiply -2.0*3.0 I get:

/home/lucio/Arduino/libraries/Arduino_RPClite/extras/test/src/RPC/test_RPCClient.cpp:27: FAILED:
  REQUIRE( expected[i] == got[i] )
with expansion:
  '�' == '@'
with messages:
  Test case [RPCClient::call] failed at line 63
  Array size: expected: 27, got: 27
  Index 10: expected 192, got 64

This is actually the correct outcome since you changed 2.0 to -2.0 the response obtained is different from the expected.

   Test case [RPCClient::call] failed at line 63
   Array size: expected: 27, got: 27
   Index 10: expected 192, got 64

since line 63 is:

    COMPARE_ARRAYS(expected, got);

The error indicates that the byte at index 10 is 64 instead of 192.
The strangest part is probably:

> /home/lucio/Arduino/libraries/Arduino_RPClite/extras/test/src/RPC/test_RPCClient.cpp:27: FAILED:
>   REQUIRE( expected[i] == got[i] )
> with expansion:
>   '�' == '@'

This is an artifact due to the implementation of the COMPARE_ARRAY part. I admit I sketched this part quickly, but it can be improved, for example changing line 27 from:

    REQUIRE(expected[i] == got[i]);

to

    REQUIRE(int(expected[i]) == int(got[i]));

improves the message:

/home/megabug/Workspace/Arduino_RPClite/extras/test/src/RPC/test_RPCClient.cpp:27: FAILED:
  REQUIRE( int(expected[i]) == int(got[i]) )
with expansion:
  64 == 192
with messages:
  Test case [RPCClient::call] failed at line 63
  Array size: expected: 27, got: 27
  Index 10: expected 64, got 192

• If I use the MALFORMED_CALL_ERR (0xFD) as error code instead of 1 in the missing parameter case I get:

/home/lucio/Arduino/libraries/Arduino_RPClite/extras/test/src/RPC/test_RPCClient.cpp:99: FAILED:
  REQUIRE( rpc.lastError.code == 0xFD )
with expansion:
  -3 == 253

It looks like a sign conversion problem maybe related to the typecast the StreamMock class is doing (?)

Here you are comparing rpc.lastError.code that is an integer with value -3 with 0xFD that is an integer with value 253.
It is correct that this will trigger an error, because you're trying to compare -3 with the Msgpack encoding of -3 that is the unsigned 0xFD.

The correct comparison should be:

REQUIRE(rpc.lastError.code == -3)

That is exactly what you will write client-side.

[cmaglie]

I've rebased this branch on top of the latest main.