watchr

Summary

Agile development tool that monitors a directory tree, and triggers a user defined action whenever an observed file is modified. Its most typical use is continuous testing, and as such it is a more flexible alternative to autotest.

Features

watchr is:

• Simple to use
• Highly flexible
• Evented ( Listens for filesystem events with native c libs )
• Portable ( Linux, *BSD, OSX, Solaris, Windows )
• Fast ( Immediately reacts to file changes )

Most importantly it allows running tests in an environment that is agnostic to:

• Web frameworks ( rails, merb, sinatra, camping, invisible, ... )
• Test frameworks ( test/unit, minitest, rspec, test/spec, expectations, ... )
• Ruby interpreters ( ruby1.8, ruby1.9, MRI, JRuby, Rubinius, ... )
• Package frameworks ( rubygems, rip, ... )

Usage

On the command line,

$ watchr path/to/script.file

will monitor files in the current directory tree, and react to events on those files in accordance with the script.

Scripts

The script contains a set of simple rules that map observed files to an action. Its DSL is a single method: watch(pattern, &action)

watch( 'a regexp pattern matching paths to observe' )  {|match_data_object| command_to_run }

So for example,

watch( 'test/test_.*\.rb' )  {|md| system("ruby #{md[0]}") }

will match any test file and run it whenever it is saved.

A continuous testing script for a basic project could be

watch( 'test/test_.*\.rb' )  {|md| system("ruby #{md[0]}") }
watch( 'lib/(.*)\.rb' )      {|md| system("ruby test/test_#{md[1]}.rb") }

which, in addition to running any saved test file as above, will also run a lib file's associated test. This mimics the equivalent autotest behaviour.

It's easy to see why watchr is so flexible, since the whole command is custom. The above actions could just as easily call "jruby", "ruby --rubygems", "ruby -Ilib", "specrb", "rbx", ... or any combination of these. For the sake of comparison, autotest runs with:

$ /usr/bin/ruby1.8 -I.:lib:test -rubygems -e "%w[test/unit test/test_helper.rb test/test_watchr.rb].each { |f| require f }"

locking the environment into ruby1.8, rubygems and test/unit for all tests.

And remember the scripts are pure ruby, so feel free to add methods, Signal#trap calls, etc. Updates to script files are picked up on the fly (no need to restart watchr) so experimenting is painless.

The wiki has more details and examples. You might also want to take a look at watchr's own scripts, specs.watchr, docs.watchr and gem.watchr, to get you started.

Install

gem install watchr

If you're on Linux/BSD and have the rev gem installed, Watchr will detect it and use it automatically. This will make Watchr evented.

gem install rev

You can get the same evented behaviour on OS X by installing ruby-fsevent.

gem install ruby-fsevent

See Also

• redgreen: Standalone redgreen eye candy for test results, ala autotest.
• phocus: Run focused tests when running the whole file/suite is unnecessary.
• autowatchr: Provides some autotest-like behavior for watchr
• nestor: Continuous testing server for Rails

Links

• code: http://github.com/mynyml/watchr
• docs: http://yardoc.org/docs/mynyml-watchr/file:README.rdoc
• wiki: http://wiki.github.com/mynyml/watchr
• bugs: http://github.com/mynyml/watchr/issues