Adding CI from GitHubActions for tests

[Alex0vSky]

Conversation from #23:

Can I do checks in CI from GitHubAction? I understand this a little.

Yes, it's a good idea, we require CI for writing and executing tests.

About writing tests, from the top of my head:

1. EXPECT_MODULE_FUNC_CALL and ON_MODULE_FUNC_CALL
   for the WinApi function starting from zero arguments up to 12 from for example CreateWindowExA().
2. Handling initialization errors
3. Handling hook errors, set and unset
4. GTest conflict workaround/bypass
5. Others such as RESTORE_MODULE_FUNC and VERIFY_AND_CLEAR_MODULE_FUNC_EXPECTATIONS

It would be enough?

[smalti]

It would be enough?

Yes, I think it's a good set of tests, and we can also add:

• Tests for Windows ApiSet DLL redirection, RoInitialize (used in gmock-win32-sample) or another function
• Tests that use INVOKE_REAL_MODULE_FUNC and REAL_MODULE_FUNC

[Alex0vSky]

Here are the fist CI logs:
https://github.com/smalti/gmock-win32/actions/runs/7767518532/job/21184418906
See "Build and run all tests in loop" node.
Used: 2Mb cache for "gmock+gtest" and ~4 minutes of time.

• Let's add the "Optimization" combination?
• And I propose to get rid of the "Debug" combination; there is no influence of the "_DEBUG" macro in the library code.

As soon as we make a decision, I will delete the discussion #29.
And we’ll also add an elementary "badge" to the readme.

There are still many questions... Auto releases, SemVer, behavior on error/failure, smart "badge"...

[Alex0vSky]

And I propose to get rid of the "Debug" combination; there is no influence of the "_DEBUG" macro in the library code.

Goodbye "_DEBUG", see you soon.

Let's add the "Optimization" combination?

Done in #38. Execution time: ~9 minutes.

And we’ll also add an elementary "badge" to the readme.

Done in #37. The path is configured correctly for this repo, and not for my fork.

[smalti]

Done in #38. Execution time: ~9 minutes.

Thanks!

Goodbye "_DEBUG", see you soon.

+1

[Alex0vSky]

My discussion/question #29 is closed, since we have reached a common opinion and there is no negative reaction to the first results of CI over time.
Closed due to success.

One from last action to close this thread lead in new issue #39.
We can then decide what to do if an error/failure occurs in the tests.