Backport of Bitcoin Core PR14795
Details
Details
Diff Detail
Diff Detail
- Repository
- rABC Bitcoin ABC
- Branch
- backport-pr14795
- Lint
Lint Passed - Unit
No Test Coverage - Build Status
Buildable 12831 Build 25732: Build Diff build-without-wallet · build-clang-tidy · build-clang-10 · build-diff Build 25731: arc lint + arc unit
Event Timeline
Comment Actions
[Bot Message]
One or more PR numbers were detected in the summary.
Links to those PRs have been inserted into the summary for reference.
Comment Actions
The test plan does not test the supposed change, nor does it test the wildcard behavior that differs from the original PR.
Comment Actions
I missed the wildcard when fixing the cherry-pick conflict. I believe I fixed that, and just tested it with the following command:
./test/functional/test_runner.py interface_*
Regarding the test plan, I need to search for a test that takes parameters. I don't know yet if there is any. Maybe I'll need to backport Core PR 12510 for that. I will look into that issue tomorrow.
| test/functional/test_runner.py | ||
|---|---|---|
| 274 | Can we replace this hideous hack by some proper use of https://docs.python.org/3/library/glob.html ? | |
Comment Actions
I'm having a closer look at test_runner.py, and it seems our code has diverged quite a lot with Bitcoin Core. I'm sorry, should have started with that before naively trying to apply the backport.
The test detection and extra command line arguments are done very differently. In our version, we always run all variations of tests that accept params. I'm not sure that allowing to pass extra arguments to test_runner.py makes sense for us. I should probably scrap the entire diff.
When you try passing extra parameters to a test that is listed as one of the tests with options, our script always runs multiple tests, so the create_cache.py script is called first, and it chokes on the unrecognized extra argument that is passed to him.
In the bitcoin core version of the script, when you pass an extra argument it necessarily means that you are running a single test, so the create_cache.py step is avoided.