Skip to content
Switch branches/tags

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Huskie - the powerful logger

Make logging through processes.

Why Huskie??

Named after my trusty Husqavana chainsaw. How I love it.


(huskie-log "something" "logger-name")

will send "something" to the "logger-name" which will be automatically mapped to a file in "/tmp/logger-name"

The automatic file mapping is done with `huskie-log-default-directory' which is a variable you can let-bind or simply change.

(let ((huskie-log-default-directory "/home/nictest"))
  (hukie-log "just a test" "my-log"))

will send "just a test" to the log process "my-log" and create a file "/home/nictest/my-log"

Log Files

The mapping between filenames and lognames can be defined:

(huskie-bind-logname->filename "nictest" "/tmp/my-log")
(huskie-log "test100!" "nictest")

will send "test100!" to the file "/tmp/my-log" though a logging process called "nictest" will be created.

The filename used MUST be a file, it is specifically checked for fileness and huskie fails if it's not.

However, huskie does NOT check that the file is under any sane root; indeed, it cannot do so because you might want to save log files in all sorts of odd places.

I am very interested to hear what people think about this so do create an issue on the GitHub if you have a viewpoint about this.


All the logging is done through async processes. A default process is used unless something specific exists for the logname. Setting a script can be done with `huskie-set-script'.

The script MUST include a single %s to accept the filename. Failure to do so will result in an error.

The script MUST be a shell script.

   "while read line; do echo $line > %s ; done")
(huskie-log "test 200" "nictest-log")

will send "test 200" through "nictest" via the syslog logger program.

You can avoid the use of the filename through shell tricks:

   "while read line; do logger -t my-log $line ; done # %s")
(huskie-log "test 200" "nictest-log")

this will send the logging through the system logger, the filename is just ignored.

The shell script could totally alter the filename passed into it.

Here's an example where all the things that can be adjusted, are:

   "while read line; do echo $line > %s ; done")
(huskie-bind-logname->filename "test-log" "/tmp/my-log")
(huskie-log "test 200" "test-log")

this causes all lines to go through the script, which is simple (but could of course be quite complicated) but causes the file created to be /tmp/my-log.


Using the shell form:

while read line ; do ... ; done

is an excellent way of ensuring you have lines going through loggers. This is why huskie does it this way. Other solutions are possible of course, but this is probably the best.

Log scripts to create dates and such are likely good candidates for the shell script.


Deriving log file destinations from the log text itself is a VERY BAD idea. Huskie makes no attempt to thwart this. If you do it you are a fool.


logging chainsaw






No releases published


No packages published