Hm, on further thought, I think I want the following:
*) Tests run on CI should spew every piece of plausibly useful information available. The cost for missing useful information is high.
*) Enabling a report =log should make just that log's messages output, not change global log verbosity.
*) Some sort of MIR_CLIENT_DEBUG environment variable to set the global client log verbosity.
I don't really mind if hugely verbose logging is not default on local test runs; it's cheap to turn back on.
Furthermore, I think that we *can* output messages to stdout/stderr by default for clients; although they're likely to be run by users from a terminal, users who do that are unlikely to mind some log messages. And if they do, they already hate glib/gtk, which will reliably spew tens to hundreds of lines of messages for lots of apps in normal operation.
We should *also* give clients the ability to hook in their own logger; they can then silence us if they really care.
Alternatively, as Ryan suggests, we could log to the journal :).
Hm, on further thought, I think I want the following:
*) Tests run on CI should spew every piece of plausibly useful information available. The cost for missing useful information is high.
*) Enabling a report =log should make just that log's messages output, not change global log verbosity.
*) Some sort of MIR_CLIENT_DEBUG environment variable to set the global client log verbosity.
I don't really mind if hugely verbose logging is not default on local test runs; it's cheap to turn back on.
Furthermore, I think that we *can* output messages to stdout/stderr by default for clients; although they're likely to be run by users from a terminal, users who do that are unlikely to mind some log messages. And if they do, they already hate glib/gtk, which will reliably spew tens to hundreds of lines of messages for lots of apps in normal operation.
We should *also* give clients the ability to hook in their own logger; they can then silence us if they really care.
Alternatively, as Ryan suggests, we could log to the journal :).