Revision 18 as of 2010-08-06 02:19:36

Clear message

Preparing a release is a fairly straight-forward process despite appearances to the contrary.

Setting up the tools

The python-storm package is made to be built with AutoPPA. AutoPPA eases the process of building packages for all supported releases of Ubuntu. You need to create a ~/.autoppa.conf file with contents similar to:

email = Username <>
ppa = username
branch = /home/username/src/projects/storm/packaging
repository = /home/username/src/autoppa-builds
releases = dapper hardy karmic lucid maverick

The branch above should be a branch of lp:~storm/storm/packaging. It's best to bzr checkout the branch, to ensure that changes are uploaded back to the master copy. You'll also need something similar to this in your ~/

fqdn =
method = ftp
incoming = ~username/ppa/ubuntu/
login = anonymous
allow_unsigned_uploads = 0

AutoPPA is available from Launchpad.

Are you ready?

  1. Make sure all bugs in the current active milestone on Launchpad

    are in the Fix Committed state.

  2. Review the Bazaar log and make sure the NEWS file contains information about all the changes made since the last release.

When all the bug fixes and new features are ready in trunk it's time to create the release.

  1. Edit storm/ and set version to the new version.

  2. Set the release date in the NEWS file.
  3. Merge the changes from lp:storm into lp:~storm/storm/packaging.

Preparing a release-candidate package

  1. Build the package in your own PPA so that you can test them before publishing them in the official Storm PPA:

    autoppa build storm 0.18-1~storm1

This will upload sources to your PPA for the Ubuntu releases specified in ~/.autoppa.conf. When the python-storm packages build you should test them in pristine environments, to make sure they work properly. Amazon's EC2 is good for this kind of testing. If you run into problems you can build the package locally to debug issues (run in the packaging branch): fakeroot debian/rules binary

Any changes made for the build will have been committed to the packaging branch.

Preparing a release-candidate tarball

  1. Run make release to prepare a tarball (created in the dist directory).

  2. This will create dist/storm-0.17.tar.bz2. Unpack it and make sure everything is in place: the README, NEWS, TODO and LICENSE files, the PKG-INFO and other distutils setup files, the debian directory with package source, the storm and tests directories with Python code, the Makefile and the storm.egg-info directory.

Test the packages

Test the new packages. If everything looks good move onto the next step. Otherwise, fix the problems and build a new version of the package. Bump the stormN part of the version string to stormN+1 and rebuild the package and tarball.

Creating and publishing the new tarball and packages

If the tarball and package are in order you can copy the package to the official Storm PPA and upload the tarball to Launchpad.

  1. Find the new packages in your PPA and copy them to the Storm Team PPA to publish them.
  2. Upload the tarball to Launchpad using the 'Create a release' function from the milestone page. Let Launchpad close the milestone as part of creating the release. Use the announcement contents to fill in the 'Release notes' and 'Detailed changelog' sections.
  3. After the release is created use the 'Add file' function to upload the tarball. Be sure to also upload a valid GPG signature

    for it. Use 'Storm <version>' for the description of the file.

Build new API documentation

  1. Run make doc in the packaging branch to build new API documentation.

  2. Copy it to a public location and update the IRC topic. Make sure the link on the front page of the wiki is correct. Contact jkakar if you want to use the existing location and he will update the documentation for you.

Announce the new release

  1. Make an announcement on Launchpad to publicise the release. Use

    the text 'Storm <version> is out!' for the headline and the text 'The latest stable release of Storm with new features and bug fixes is available.' for the summary. Set the URL to the milestone, such as:

  2. Change the topic on the IRC channel to reference the new version.
  3. Write a nice release announcement for the Launchpad page and send a copy of it to the Storm mailing list to announce the release. The announcement from 0.16 is below and can be used as a template for new releases. In general, the announcement should (1) provide information about where to get the latest packages and tarball, (2) highlight the top 3-5 changes, including information about each one, and finally (3) the contents of the NEW file should be included.

The Storm team is proud to announce Storm 0.16!

The new release includes a number of new features:

 * Storm's C extensions are enabled by default
 * Comparable expressions have new string comparison methods
 * The default object cache size has been changed
 * Set expressions use fewer stack frames during compilation
 * MySQL reserved words are correctly handled

This release includes official packages for all supported releases
of Ubuntu.  They are available in the Storm team's PPA:

You can find the release files at:

You can always get the latest source code from Launchpad:

 bzr branch lp:storm

Finally, you can join us in the #storm channel on
and on the Storm mailing list:

Read on for more...

Storm's C extensions are enabled by default

Storm has had C extensions for a long time, but until now they've
been disabled by default.  The extensions have a significant
improvement on performance and are now used by default.  The
extensions can be disabled by defining a STORM_CEXTENSIONS=0
environment variable.  When not available, Storm will automatically
fall back to the equivalent Python versions of optimized code.

Comparable expressions have new string comparison methods

Comparable expressions (such as Column and Alias) provide new
startswith(), endswith() and contains_string() methods.  These
methods perform prefix, suffix and substring comparisons using LIKE.
Strings used with these methods are automatically escaped.

The default object cache size has been changed

Storm 0.15 included an announcement about the default cache size
being changed from 100 items to 1000 items.  Due to a bug this
information was actually incorrect.  This is now fixed and should
result in improved performance.  Please report any problems or
concerns about this change to the mailing list.

Set expressions use fewer stack frames during compilation

The set expression constructor flattens the first argument, if it's
part of the same type.  The resulting expression tree uses fewer
stack frames during compilation, reducing the chance of hitting
Python's recursion limit.

MySQL reserved words are correctly handled

In addition to the standard SQL92 reserved words, MySQL has it's own
set reserved words.  Storm is now aware of them and will perform
escaping correctly, making it easier for users that have table and
column names that conflict with this list.

Detailed changelog


 - The set expression constructor will now flatten its first argument
  if it is of the same type.  The resulting expression tree uses less
  stack when compiling so reduces the chance of hitting Python's
  recursion limit (bug #242813).
 - Add startswith(), endswith() and contains_string() methods to
  Comparable.  These methods perform prefix, suffix and substring
  checks respectively using the LIKE operator, taking care of escaping
  for you (bug #387840).
 - C extensions are enabled by default.  Define the
  STORM_CEXTENSIONS=0 environment variable to disable them (bug #410592).
 - The README file contains information about Storm's license and
  detailed instructions on setting up a development environment
  suitable for running the entire test suite.
 - 'make doc' uses Pydoctor to generate API documentation.
 - Integration tests for Django now work with Django 1.1.

Bug fixes:

 - Remove a leak when mutable variables (ListVariable or PickleVariable
  instances) are collected before store.flush, leaving hooks behind them.
 - The ResultSet min, max and sum methods now work correctly when the
  result set is empty and the column has allow_none=False set.
  Previously this resulted in a NoneError (bug #457801).
 - MySQL reserved words are handled properly (bug #433833).
 - Test loading code has been simplified.  Support for py.test has
  been removed in this process, as it was not functioning correctly
  and didn't fit into the PyUnit framework (bug #331905).
 - Remote diverged and remote deleted references now use a weak
  (Python) reference to the local object.  This prevents a leak
  when the remote object stays in memory (bug #475148).
 - Check for invalidated state when returning the remote object of a
  relation: it fixes a bug if the local key of the Reference is the
  primary key (bug #435962).
 - The default Cache instance created for a Store honours Cache's
  default size.  Store's docstring has been updated to reflect
  this (bug #374180).

Prepare to develop the next version

  1. If it hasn't already been done, create a milestone for the next version and target bugs that needs to be delivered with it.
  2. Bump the version value in the storm/

  3. Edit the NEWS file and add a new section for the development version.