As requested on #16976
No breaking changes
As requested on #16976
No breaking changes
As requested on #16976
No breaking changes
59 | @@ -60,7 +60,7 @@ after_build: 60 | - ps: clcache -z 61 | #- 7z a bitcoin-%APPVEYOR_BUILD_VERSION%.zip %APPVEYOR_BUILD_FOLDER%\build_msvc\%platform%\%configuration%\*.exe 62 | test_script: 63 | -- cmd: src\test_bitcoin.exe -k stdout -e stdout 2> NUL 64 | +- cmd: src\test_bitcoin.exe -l test_suite -k stdout -e stdout 2> NUL
that is going to print the boost test messages always, even if everything passes
I am not sure about any log limits for appveyor, but this seems a bit verbose to me
Also, it would be nice to run the tests in parallel (like we do with automake make check -j xx)
Jup, it does print it unconditionally: https://ci.appveyor.com/project/DrahtBot/bitcoin/builds/29582061#L1000
re-run ci
Doesn't work (nothing is printed):
src\test_bitcoin.exe -l test_suite -k stdout -e stdout 2> NUL
Command exited with code -1073741819
cc @sipsorcery
Looks like #17573 may be causing test_bitcoin.exe to seemingly do nothing. Currently tracking down why.
Running the test_bitcoin.exe with the addition of -l test_suite adds just under 1000 lines of logs to the appveyor output. I suspect it has a negligible impact on the appveyor job performance.
My concern would be for contributors needing to look at the logs the burden is increased. I don't think it gives us a lot of bang for our buck as most of the failures are at the build stage or the python functional tests. If a way could be found to only output the logs if there's a test_bitcoin.exe failure that would be useful.
re-run ci
Going to close. @sipsorcery maybe you'd like to pick this up if you think it's an improvement.