Changing the defaults about logxi HOT 5 OPEN

I'm not sure if this helps. The way I use logxi for package-private logging is to create a logger with a namespace.

//// models/init.go
var logger  = log.New("my-models")

//// controllers/init.go
var logger = log.New("my-controllers")

You can then control how each package logs

LOGXI=my-models=DBG,my-controllers=ERR

from logxi.

That's what we do too, but we want different defaults for LOGXI and LOGXI_FORMAT when running tests. Our makefile can set these, but the makefile is only useful when you want to run all the project's tests. When we are focussing on individual packages or tests, we're using "go test ..." and its tougher to manage a different set of environment variables just for those test runs. For our other configuration options (also configured via environment variables), we use a .env loader at runtime, but that happens too late for logxi...by the time we've run that, logxi has already configured itself.

Maybe I can just use the Suppress() function to turn off logs completely during tests.

from logxi.

I need to refactor how logxi reads environment vars. The unit tests are more complicated than they need to be because of it. Do you want to take a stab at refactoring it? I have limited time.

from logxi.

Thoughts on how to refactor? I'm a little torn on the best approach. There are good reasons to init logging in init() functions (mainly so logging can happen as early as possible in boot up). If I add a more robust way to reconfigure existing loggers, it seems I'll have to add at least one extra mutex into the logger calls, which will impact performance. Or move loggers to channels to avoid the lock, but again, not sure what the performance impact will be...

There's currently a "concurrent writer" which has a mutex...maybe I could move that mutex up the stack, to a "ConcurrentLogger"? That would avoid adding another mutex.

from logxi.

Another simpler approach would be passing defaults to log.New().

from logxi.