This is a pre-production deployment of Warehouse. Changes made here affect the production instance of PyPI (pypi.python.org).
Help us improve Python packaging - Donate today!

Bootstrap a TurboGears app in a VirtualEnv

Project Description

Status and License

tg_bootstrap is simple glue code between virtualenv and TurboGears.

It is written by Arlo Belshee. It is licensed under an MIT-style permissive license.

You can install it with easy_install tg_bootstrap, or from the subversion repository with easy_install tg_bootstrap==dev.

What It Does

tg_bootstrap is a tool to create a TurboGears application in an isolated Python environment with a minimal number of system dependencies.

tg_bootstrap achieves the following:

  • Create a one-click development config script.
  • Create a one-click release deployment script.
  • Remove your TurboGears-built setup.py’s dependency on TurboGears.

These install scripts will set up your app on any machine (Linux, Mac, or Windows). The machine need only have a bare Python install (you will also need command-line svn on the path for your development machine). Your app will be completely isolated from any other Python install on your development machine. It will not share any packages either from or to your global system Python or any other Python install.

Both scripts will upgrade a previous version of your app if you install over the top of it. The release script installs your app as an egg. The development script checks out the source and install from there.

TurboGears generates a setup.py for your app when you create it. Unfortunately, this setup.py imports a module (finddata.py) from TurboGears itself. Thus, even though your setup.py script indicates that TurboGears is a dependency, it cannot be used to install TurboGears on a machine that doesn’t already have TurboGears.

To get around this problem, I have packaged TurboGears’ finddata.py in this bootstrapper. This way, you need only have tg_bootstrap installed to run your setup.py. All the real dependencies - the ones that version frequently - can be controlled by each app’s setup.py.

I will occasionally update the finddata.py that I include with tg_bootstrap. However, it doesn’t need to support everything that the main TurboGears finddata.py supports. After all, you just use this finddata.py for long enough to install TurboGears on a new development machine. After that, including during egg-based release installs, your setup.py uses TurboGears’ then-current finddata.py.

Usage

Generating Scripts

Assume you were a company with two different TurboGears applications. Here is a script which will generate 4 scripts: release and development install scripts for each product:

from tg_bootstrap import bootstrap

bootstrap.create_scripts(
  src_repository='http://svn.example.com/app1/trunk',
  proj='someapp',
  release_version='2.3.1b3')

bootstrap.create_scripts(
  src_repository='http://svn.example.com/otherapp/trunk',
  proj='otherapp',
  release_version='0.2.1')

Creating a New Project

To create a brand-new TurboGears project without mucking up your system Python (by, say, installing TurboGears into it), do the following:

  1. If necessary, create a subversion repository for your project(s).
  2. Add an empty directory for your new project’s source.
  3. Create a temporary virtualenv to work in. To do this by hand,
    • Download virtualenv.py.
    • Using your system python, run python virtualenv.py /tmp/sandbox/for/now
    • Activate that virtualenv. See the virtualenv documentation for platform-specific instructions.
    • In the virtualenv, run easy_install tg_bootstrap and easy_install TurboGears.
    • Note to Vista users: easy_install is not yet aware of the Vista security model. Thus, it creates an easy_install.exe application that doesn’t work on Vista! As a workaround, you can use easy_install-script.py, run via virtualenv’s python. Take care, as Python’s installer has associated .py files with your system python install. So if you have activated a virtualenv, then easy_install-script.py foo will install foo to your system Python, while python easy_install-script.py foo will install it to your virtualenv.
  4. In the active temporary virtualenv, check out your project’s empty directory.
  5. Use tg-admin in the virtualenv to create your project as normal.
  6. Update your project’s setup.py to use tg_bootstrap’s finddata.py (see below).
  7. Commit your project.
  8. If you haven’t already, add another directory to your repository to contain your bootstrapping scripts.
  9. Extend your installer script generator to contain the command for your new app. Generate and commit your install scripts.
  10. From any Python (system or virtualenv), run your development install script: python projname-develop.py /sandbox/for/development/of/new/app.
  11. Delete the temporary bootstrapping virtualenv, and begin working as normal in your development sandbox.

Installing From Generated Scripts

To A Development System

To start developing an existing project on a new machine, you need only:

  1. Make sure Python is installed on your machine.
  2. Make sure command-line svn is installed and on your path.
  3. Run python someproj-develop.py /sandbox/path/for/app

The script will create a virtualenv at /sandbox/path/for/app. Inside the virtualenv, it will check out your project’s source to /sandbox/path/for/app/src. It will then perform a development install of your project, including all dependencies, into that virtualenv.

If your project adds dependencies or you which to freshen your dependencies to the latest version, just run your app’s setup.py script from within the virtualenv:

python setup.py develop

To A Release System

To package and deploy a project release, you need only:

  1. Package your application as an egg.
  2. Either upload your egg to pypi or copy it to your deployment / test server.
  3. Update the release number in your script generation script and regenerate your scripts.
  4. Make sure Python is installed on your target server.
  5. Run python someproj-release.py /deployment/path/for/app

The release script will create a virtualenv at /sandbox/path/for/app. It will install your egg, and all its dependencies, into that virtualenv. At that point, you need only do whatever your system requires to make your app live.

The virtualenv documentation describes how to configure common systems to run an app from a virtualenv. It includes descriptions of shared persistent server mechanisms such as mod_python.

Note: virtualenv.py supports the --no-site-packages option. It defaults to including your system Python’s site packages in your virtualenv. However, keeping dependencies clean makes upgrades easier. Thus, install scripts generated with tg_bootstrap have the reverse default, and a --with-global-site-packages override option. If you use tg_bootstrap and don’t provide a command-line option, then your virtualenv will start clean and blank; it will not have access to your machine’s system python libraries.

The generated scripts accept all of virtualenv.py’s other options, unchanged.

Reducing Dependencies

To use this, simply change the line of your setup.py which reads:

from turbogears.finddata import find_package_data

to read:

try:
       from turbogears.finddata import find_package_data
except ImportError:
       from tg_bootstrap.finddata import find_package_data

The rest of setup.py is unchanged.

Changes & News

svn trunk

Nothing yet

0.4

  • Documentation
  • Initial version. Supports creating new projects without system dependencies.
Release History

Release History

This version
History Node

0.4.0

History Node

0.1.0

History Node

0.1.0dev-r9

Download Files

Download Files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

File Name & Checksum SHA256 Checksum Help Version File Type Upload Date
tg_bootstrap-0.4.0-py2.5.egg (16.9 kB) Copy SHA256 Checksum SHA256 2.5 Egg Jan 12, 2008
tg_bootstrap-0.4.0.zip (16.2 kB) Copy SHA256 Checksum SHA256 Source Jan 12, 2008

Supported By

WebFaction WebFaction Technical Writing Elastic Elastic Search Pingdom Pingdom Monitoring Dyn Dyn DNS Sentry Sentry Error Logging CloudAMQP CloudAMQP RabbitMQ Heroku Heroku PaaS Kabu Creative Kabu Creative UX & Design Fastly Fastly CDN DigiCert DigiCert EV Certificate Rackspace Rackspace Cloud Servers DreamHost DreamHost Log Hosting