Introduction

Namespace packages offer the huge benefit of being able to distribute parts of a large system in small, self-contained pieces. However, they can be somewhat clunky to navigate, since you end up with a large list of eggs in your egg cache, and then a seemingly endless series of directories you need to open to actually find the contents of your egg.

This recipe sets up a directory structure that mirrors the actual python namespaces, with symlinks to the egg contents. So, instead of this...:

egg-cache/
    my.egg.one-1.0-py2.7.egg/
        my/
            egg/
                one/
                    (contents of first egg)
    my.egg.two-1.0-py2.7.egg/
        my/
            egg/
                two/
                    (contents of second egg)

...you get this:

omelette/
    my/
        egg/
            one/
                (contents of first egg)
            two/
                (contents of second egg)

You can also include non-eggified python packages in the omelette. This makes it simple to get a single path that you can add to your PYTHONPATH for use with specialized python environments like when running under mod_wsgi or PyDev.

Typical usage with Zope and Plone

For a typical Plone buildout, with a part named "instance" that uses the plone.recipe.zope2instance recipe and a part named "zope2" that uses the plone.recipe.zope2install recipe, the following additions to buildout.cfg will result in an omelette including all eggs and old-style Products used by the Zope instance as well as all of the packages from Zope's lib/python. It is important that omelette come last if you want it to find everything:

[buildout]
parts =
    ...(other parts)...
    omelette

...

[omelette]
recipe = collective.recipe.omelette
eggs = ${instance:eggs}
products = ${instance:products}
packages = ${zope2:location}/lib/python ./

(Note: If your instance part lacks a 'products' variable, omit it from the omelette section as well, or the omelette will silently fail to build.)

Supported options

The recipe supports the following options:

eggs: List of eggs which should be included in the omelette.
location: (optional) Override the directory in which the omelette is created (default is parts/[name of buildout part])
ignore-develop: (optional) Ignore eggs that you are currently developing (listed in ${buildout:develop}). Default is False
ignores: (optional) List of eggs to ignore when preparing your omelette.
packages: List of Python packages whose contents should be included in the omelette. Each line should be in the format [package_location] [target_directory], where package_location is the real location of the package, and target_directory is the (relative) location where the package should be inserted into the omelette (defaults to Products/).
products: (optional) List of old Zope 2-style products directories whose contents should be included in the omelette, one per line. (For backwards-compatibility -- equivalent to using packages with Products as the target directory.)

Windows support under Python 2.7

Using omelette on Windows with Python 2.7 requires the junction utility to make links. Junction.exe must be present in your PATH when you run omelette.

Since Python 3.2 this is no longer needed.

Using omelette with eggtractor

Mustapha Benali's buildout.eggtractor provides a handy way for buildout to automatically find development eggs without having to edit buildout.cfg. However, if you use it, the omelette recipe won't be aware of your eggs unless you a) manually add them to the omelette part's eggs option, or b) add the name of the omelette part to the buildout part's tractor-target-parts option.

Using omelette with zipped eggs

Omelette doesn't currently know how to deal with eggs that are zipped. If it encounters one, you'll see a warning something like the following:

omelette: Warning: (While processing egg elementtree) Egg contents not found at
/Users/davidg/.buildout/eggs/elementtree-1.2.7_20070827_preview-py2.4.egg/elementtree.  Skipping.

You can tell buildout to unzip all eggs by setting the unzip = true flag in the [buildout] section. (Note that this will only take effect for eggs downloaded after the flag is set.)

Running the tests

Just grab the recipe from git and run:

tox -p auto

Known issue: The tests run buildout in a separate process, so it's currently impossible to put a pdb breakpoint in the recipe and debug during the test. If you need to do this, set up another buildout which installs an omelette part and includes collective.recipe.omelette as a development egg.