I'm working on a project utilising a similar config file syntax as rlog does.
The logic consequence was to join them both.
It's working, but rlog complains about unkown or malformed settings:

rlog - Malformed line in config file file:line. Ignored.
rlog - Unknown or illegal setting name in config file file:line. Ignored.

Having multiple config file for a single application is a nightmare for me.
So I decided to comment out corresponding lines in my version of rlog.

What about officially supporting multi function config files instead?

Regards,
Trickx

Currently, all configuration is done via environment variables. Sometimes, however, it may be useful to change the logging parameters at run time.

We already have a function to change the output stream (for example to a new log file), but it would be nice to also be able to change the other logging parameters.

As an extension of this: It might be good to be able to programmatically query the current parameters in an easy manner. The program could just read the environment variables, but rlog itself could provide a convenient facility for that.

Since 1.14 (some time ago) we got go modules. If one, in the present day, executes a go get on your repo, expecting it to execute the example.go file in the $GOPATH/pkg/mod/... it won't work as it complains about no go.mod file. How do you feel about adding that?

This will also add the general ability to configure rlog via config file.

Config files by default should be at $RLOG_CONFIG_ROOT/filename.conf

By default, RLOG_CONFIG_ROOT is assumed to be "/etc/rlog/"

filename is just the filename of the executable. The default of looking for filename.conf can be overwritten with the RLOG_CONFIG_FILE variable, which can be used to set an absolute path to a full filename. That way, multiple executables can utilize the same config file.

rlog will check the config file every N seconds for changes.

The config file simply contains whatever you would have as environment variable assignments:

RLOG_LOG_LEVEL=INFO
RLOG_TRACE_LEVEL=3
...

Environment variables provided at the start of a process take precedence over whatever is in the config file, unless the line in the config file starts with a "!", like so:

RLOG_LOG_LEVEL=INFO
!RLOG_TRACE_LEVEL=3
...

When rlog encounters a log message, it should check how long ago it last checked the config file. If it's more than RLOG_CONF_CHECK_INTERVAL seconds ago, then it should re-read the config file again.

In Python I can do something like

logger = logging.getLogger("foobar")

and foobar gets printed in every line logged by that object.

One use case where this is useful is in Lambdas: by initializing the logger with the lambda name, one can easily filter / see which lambda function emitted the line when using a log aggregation tool, such as CloudWatch or ElasticSearch.

I haven't found a way of obtaining the same with rlog.

Happy to contribute with a PR if needs be.

I see logs are printing in console and log file as well, is it possible to disable console logging and write logs only to log file with rlog?

I tried below in main.go file and it redirectsall the logs from main.go to log file only as expected.
newLogFile, err := os.OpenFile("logs/rlog-output.log", os.O_APPEND|os.O_WRONLY|os.O_CREATE, 0777)
if err == nil {
rlog.Info("About to change log output. Check /logs/rlog-output.log...")
rlog.SetOutput(newLogFile)
rlog.Info("This should go to the new logfile")
}

But, it won't write logs to log file only, If I use rlog functions in another go file.