Canoe is a simple log analyzer that watches logs and performs actions on them in real time. It's an old concept, but a new tool.

Canoes are the processes created via filters (filters) that are sent down a route (actions).

When you think of how you monitor log files it's likely that you sling together a huge pipeline of shell commands--something like this

tail -F t.log | grep '^|ERROR' --color=always | egrep -v '(ignore1|ignore2|ignore3)'

Which will highlight ERROR and filter out all of the ignored things. It can get much more complicated, especially when you try to take into consideration multiple lines and context.

In canoe, think of filters as grep. Well, grep with an always available -C <n>

Filters can be composed, and should be for complicated tasks.

Routes are what happens after messages are filtered. Lets say that you wanted to recreate the above pipeline. To do so you'd create a canoe out of a FilterOutWords, and send it down a ColoredOutputRoute, with a regex to color the word 'ERROR'.

Of course, all of this is done through a configuration file, so it's relatively easy to do.