run_command output to console if capture: false

[julianneswinoga]

Currently long-running programs using run_command do not have any output until they have finished. This can give the indication that the configure step is (wrongfully) stuck. Having console output during a long-running run_command also aides in debugging.

This patch outputs stdout and stderr to console when capture: false, instead of previously just throwing it away (subprocess.DEVNULL).

From Popen docs:

With the default settings of None, no redirection will occur.

I also added a unit test, allplatformstests.py::test_run_command_output, to check this behavior using test cases/common/33 run program. I modified the test case a little to make the stderr output more explicit.

[eli-schwartz]

Currently long-running programs using run_command do not have any output until they have finished. This can give the indication that the configure step is (wrongfully) stuck. Having console output during a long-running run_command also aides in debugging.

This patch outputs stdout and stderr to console when capture: false, instead of previously just throwing it away (subprocess.DEVNULL).

This doesn't handle long-running commands that do capture the stdout, then.

Maybe a better option is to log that we are running a command?

[julianneswinoga]

Yeah, I think the better solution is to have an option to show output as it's running (like custom_target(console: true)) but I also felt adding in that support was a lot of change for my first MR here 🙂

Since the current implementation just throws away the output I felt that this was a good feature that doesn't change the existing functionality.

[julianneswinoga]

@eli-schwartz Would you prefer that I teach run_command about the console keyword rather than this MR?

[julianneswinoga]

Closed in favor of #14104