ets-labs / python-dependency-injector Goto Github PK

Adding of Google Analytics for documentation on RTD

We need to create a ScopedProvider. The idea is of this provider that it behaves like a singleton, but has not only one scope.

ScopedProvider should support not only one scope and should have functionality for clearing all the scopes.

File formats: .gliffy, .gxml

Fix the crutch with injection.value:

class Injection(object):

    """Base injection class."""

    def __init__(self, name, injectable):
        """Initializer."""
        self.name = name
        self.injectable = injectable

    @property
    def value(self):
        """Return injectable value."""
        if hasattr(self.injectable, '__is_objects_provider__'):  # TODO: Fix this
            return self.injectable()
        return self.injectable

File formats: .vsd, .vsx, .vsdx

Integration of Coveralls

Python lib - coveralls-python

Expectations from current integration are:

• Make unit tests coverage measuring on every commit.
• Store history of unit tests coverage with attached commits information.
• Badge about unit tests coverage from shields.io should be placed into main README.md.

Acceptance criterias:

• Tests on Python 3.4 passed.
• Badge with supported version added to README.md

https://readthedocs.org/projects/objects/

The idea is not to create overridden list for every Provider instance. This could be done just in override time.

Acceptance criterias:

• Memory gain need to be calculated.

Creation of Objects internals class diagram

Need to create documentation structure

It seems a little bit confusing

Acceptance criterias:

• PyLint, flake8 and pep257 are run on every commit.
• Results should be easy to overview.
• History of results is saved.

https://pypip.in/ is not reachable anymore, need to remove or replace its badges from README

• PyLint Rating not less than 9.0

Need to create Callable provider. Current Function provider is a static provider of an function object.

Idea of Callable is to provide an callable function with some predefined dependencies.

Need to add Objects-specific errors hierarchy.

Integration of TravisCI and tox.

Expectations from current integration are:

• Make unit tests running automatically on every commit.
• Tests should be run on Python 2.7.
• Configuration should be extensible to setup unit testing in other environments (Python 2.6, Python 3, PyPy, etc...)
• Badge about build's status should be placed into main README.md.

• Split concept example into declarative part with some asserts and usage part.
• Retain only concept example in main REAME.md
• Add README.md for examples.
• Add link to examples/README.md in main README.md

Acceptance criterias:

• Tests on Python 2.6 passed.
• Badge with supported version added to README.md

• Implement publish command in setup.py
• Remove manage.py from requirements-maintain.txt
• Remove manage.py

Make project's shields using https://pypip.in/

Acceptance criterias:

• Tests on Python 3.3 passed.
• Badge with supported version added to README.md

File formats: .graffle, .graffle.zip

There are some kind of situations with settings injections that should be handled in special way:

- Most of the systems make settings global usable (Django settings, Flask app.config etc...). This is not a best practice.
- Settings could be defined after the objects catalog is defined. So best what `objects` can do now is:

class Catalog(AbstractCatalog):

    config = ExternalDependency(instance_of=dict)

    consumer = NewInstance(SomeClass,
                           InitArg('config', config))


Catalog.config.satisfy(Object(dict(some_key='some_value')))

The idea is to make special type of provider for such cases:

class Catalog(AbstractCatalog):

    config = Config()

    consumer = NewInstance(SomeClass,
                           InitArg('some_setting', config.some_key))


Catalog.config.satisfy(Object(dict(some_key='some_value')))

Acceptance criterias:

• Tests on PyPy passed.
• Badge with supported version added to README.md

Acceptance criterias:

• Tests on PyPy3 passed.
• Badge with supported version added to README.md

Add __slots__ attribute for all classes to get more effective memory usage.

Acceptance criterias:

• Tests on Python 3.2 passed.
• Badge with supported version added to README.md.

There are some kind of situations with settings injections that should be handled in special way:

• Most of the systems make settings global usable (Django settings, Flask app.config etc...). This is not a best practice.
• Settings could be defined after the objects catalog is defined. So best what objects can do now is:

class Catalog(AbstractCatalog):

    config = ExternalDependency(instance_of=dict)

    consumer = NewInstance(SomeClass,
                           InitArg('config', config))


Catalog.config.satisfy(Object(dict(some_key='some_value')))

The idea is to make special type of provider for such cases:

class Catalog(AbstractCatalog):

    config = Config()

    consumer = NewInstance(SomeClass,
                           InitArg('some_setting', config.some_key))


Catalog.config.satisfy(Object(dict(some_key='some_value')))

Example:

• http://www.jointjs.com/demos/umlcd

AutoLayout:

• http://www.daviddurman.com/automatic-graph-layout-with-jointjs-and-dagre.html
• http://jointjs.com/rappid/docs/layout/directedGraph