Updated contributing documentation to contain the new release process and updated information about branches using DVCSes.

git-svn-id: http://code.djangoproject.com/svn/django/trunk@9071 bcc190cf-cafb-0310-a4f2-bffc1f526a37
This commit is contained in:
Jacob Kaplan-Moss 2008-09-19 19:44:10 +00:00
parent 060919afa1
commit de9acabf12
3 changed files with 290 additions and 62 deletions

View File

@ -184,6 +184,8 @@ Patch style
An exception is for code changes that are described more clearly in plain
English than in code. Indentation is the most common example; it's hard to
read patches when the only difference in code is that it's indented.
Patches in ``git diff`` format are also acceptable.
* When creating patches, always run ``svn diff`` from the top-level
``trunk`` directory -- i.e., the one that contains ``django``, ``docs``,
@ -400,10 +402,10 @@ translated, here's what to do:
* Join the `Django i18n mailing list`_ and introduce yourself.
* Create translations using the methods described in the :ref:`i18n
documentation <topics-i18n>`. For this you will use the ``django-admin.py
makemessages`` tool. In this particular case it should be run from the
top-level ``django`` directory of the Django source tree.
* Create translations using the methods described in the
:ref:`i18n documentation <topics-i18n>`. For this you will use the
``django-admin.py makemessages`` tool. In this particular case it should
be run from the top-level ``django`` directory of the Django source tree.
The script runs over the entire Django source tree and pulls out all
strings marked for translation. It creates (or updates) a message file in
@ -597,11 +599,11 @@ Our policy is:
version. Assume documentation readers are using the latest release, not the
development version.**
Our prefered way for marking new features is by prefacing the features'
Our preferred way for marking new features is by prefacing the features'
documentation with: ".. versionadded:: X.Y", followed by an optional one line
comment and a mandatory blank line.
General improvements, or other changes to the APIs that should be emphasised
General improvements, or other changes to the APIs that should be emphasized
should use the ".. versionchanged:: X.Y" directive (with the same format as the
``versionadded`` mentioned above.
@ -690,6 +692,11 @@ repository:
first commit the change to library Y, then commit feature X in a separate
commit. This goes a *long way* in helping all core Django developers
follow your changes.
* Separate bug fixes from feature changes.
Bug fixes need to be added to the current bugfix branch (e.g. the
``1.0.X`` branch) as well as the current trunk.
* If your commit closes a ticket in the Django `ticket tracker`_, begin
your commit message with the text "Fixed #abc", where "abc" is the number
@ -809,7 +816,6 @@ method as above::
./runtests.py --settings=settings markup
Requesting features
===================
@ -837,23 +843,80 @@ our repository; see below.
Branch policy
=============
In general, most development is confined to the trunk, and the trunk
is kept stable. People should be able to run production sites against the
trunk at any time.
In general, the trunk must be kept stable. People should be able to run
production sites against the trunk at any time. Additionally, commits to trunk
ought to be as atomic as possible -- smaller changes are better. Thus, large
feature changes -- that is, changes too large to be encapsulated in a single
patch, or changes that need multiple eyes on them -- must happen on dedicated
branches.
Thus, large architectural changes -- that is, changes too large to be
encapsulated in a single patch, or changes that need multiple eyes on them --
will have dedicated branches. See, for example, the `i18n branch`_. If you
have a change of this nature that you'd like to work on, ask on
`django-developers`_ for a branch to be created for you. We'll create a branch
for pretty much any kind of experimenting you'd like to do.
This means that if you want to work on a large feature -- anything that would
take more than a single patch, or requires large-scale refactoring -- you need
to do it on a feature branch. Our development process recognizes two options
for feature branches:
We will only branch entire copies of the Django tree, even if work is only
happening on part of that tree. This makes it painless to switch to a branch.
1. Feature branches using a distributed revision control system like
Git_, Mercurial_, Bazaar_, etc.
If you're familiar with one of these tools, this is probably your best
option since it doesn't require any support or buy-in from the Django
core developers.
However, do keep in mind that Django will continue to use Subversion for
the foreseeable future, and this will naturally limit the recognition of
your branch. Further, if your branch becomes eligible for merging to
trunk you'll need to find a core developer familiar with your DVCS of
choice who'll actually perform the merge.
If you do decided to start a distributed branch of Django and choose to make it
public, please add the branch to the `Django branches`_ wiki page.
2. Feature branches using SVN have a higher bar. If you want a branch in SVN
itself, you'll need a "mentor" among the :ref:`core committers
<internals-committers>`. This person is responsible for actually creating
the branch, monitoring your process (see below), and ultimately merging
the branch into trunk.
If you want a feature branch in SVN, you'll need to ask in
`django-developers`_ for a mentor.
Developers working on a branch should periodically merge changes from the trunk
into the branch. Please merge at least once a week. Every time you merge from
the trunk, note the merge and revision numbers in the commit message.
.. _git: http://git.or.cz/
.. _mercurial: http://www.selenic.com/mercurial/
.. _bazaar: http://bazaar-vcs.org/
.. _django branches: http://code.djangoproject.com/wiki/DjangoBranches
Branch rules
------------
We've got a few rules for branches born out of experience with what makes a
successful Django branch.
DVCS branches are obviously not under central control, so we have no way of
enforcing these rules. However, if you're using a DVCS, following these rules
will give you the best chance of having a successful branch (read: merged back to
trunk).
Developers with branches in SVN, however, **must** follow these rules. The
branch mentor will keep on eye on the branch and **will delete it** if these
rules are broken.
* Only branch entire copies of the Django tree, even if work is only
happening on part of that tree. This makes it painless to switch to a
branch.
* Merge changes from trunk no less than once a week, and preferably every
couple-three days.
In our experience, doing regular trunk merges is often the difference
between a successful branch and one that fizzles and dies.
If you're working on an SVN branch, you should be using `svnmerge.py`_
to track merges from trunk.
* Keep tests passing and documentation up-to-date. As with patches,
we'll only merge a branch that comes with tests and documentation.
.. _svnmerge.py: http://www.orcaware.com/svn/wiki/Svnmerge.py
Once the branch is stable and ready to be merged into the trunk, alert
`django-developers`_.
@ -953,47 +1016,6 @@ file. Then copy the branch's version of the ``django`` directory into
.. _path file: http://docs.python.org/lib/module-site.html
.. _official-releases:
Official releases
=================
Django's release numbering works as follows:
* Versions are numbered in the form ``A.B`` or ``A.B.C``.
* ``A`` is the major version number, which is only incremented for major
changes to Django, and these changes are not necessarily
backwards-compatible. That is, code you wrote for Django 6.0 may break
when we release Django 7.0.
* ``B`` is the minor version number, which is incremented for large yet
backwards compatible changes. Code written for Django 6.4 will continue
to work under Django 6.5.
A minor release may deprecate certain features in previous releases. If a
feature in version ``A.B`` is deprecated, it will continue to work in
version ``A.B+1``. In version ``A.B+2``, use of the feature will raise a
``PendingDeprecationWarning`` but will continue to work. Version
``A.B+3`` will remove the feature entirely. Major point releases will
always remove deprecated features immediately.
* ``C`` is the micro version number which, is incremented for bug and
security fixes. A new micro-release will always be 100%
backwards-compatible with the previous micro-release.
* In some cases, we'll make release candidate releases. These are of the
form ``A.BrcN``, which means the ``Nth`` candidate release of version
``A.B``.
An exception to this version numbering scheme is the pre-1.0 Django code.
There's no guarantee of backwards-compatibility until the 1.0 release.
In Subversion, each Django release will be tagged under `tags/releases`_. If
it's necessary to release a bug fix release or a security release that doesn't
come from the trunk, we'll copy that tag to ``branches/releases`` to make the
bug fix release.
Deciding on features
====================

View File

@ -21,3 +21,4 @@ the hood".
contributing
documentation
committers
release-process

View File

@ -0,0 +1,205 @@
.. _internals-release-process:
========================
Django's release process
========================
.. _official-releases:
Official releases
=================
Django's release numbering works as follows:
* Versions are numbered in the form ``A.B`` or ``A.B.C``.
* ``A`` is the *major version* number, which is only incremented for major
changes to Django, and these changes are not necessarily
backwards-compatible. That is, code you wrote for Django 6.0 may break
when we release Django 7.0.
* ``B`` is the *minor version* number, which is incremented for large yet
backwards compatible changes. Code written for Django 6.4 will continue
to work under Django 6.5.
* ``C`` is the *micro version* number which, is incremented for bug and
security fixes. A new micro-release will always be 100%
backwards-compatible with the previous micro-release.
* In some cases, we'll make alpha, beta, or release candidate releases.
These are of the form ``A.B alpha/beta/rc N``, which means the ``Nth``
alpha/beta/release candidate of version ``A.B``.
An exception to this version numbering scheme is the pre-1.0 Django code.
There's no guarantee of backwards-compatibility until the 1.0 release.
In Subversion, each Django release will be tagged under ``tags/releases``. If
it's necessary to release a bug fix release or a security release that doesn't
come from the trunk, we'll copy that tag to ``branches/releases`` to make the
bug fix release.
Major releases
--------------
Major releases (1.0, 2.0, etc.) will happen very infrequently (think "years",
not "months"), and will probably represent major, sweeping changes to Django.
Minor releases
--------------
Minor release (1.1, 1.2, etc.) will happen roughly every six months -- see
`release process`_, below for details.
These releases will contain new features, improvements to existing features, and
such. A minor release may deprecate certain features from previous releases. If a
feature in version ``A.B`` is deprecated, it will continue to work in version
``A.B+1``. In version ``A.B+2``, use of the feature will raise a
``PendingDeprecationWarning`` but will continue to work. Version ``A.B+3`` will
remove the feature entirely.
So, for example, if we decided to remove a function that existed in Django 1.0:
* Django 1.1 will contain a backwards-compatible replica of the function
which will raise a ``PendingDeprecationWarning``. This warning is silent
by default; you need to explicitly turn on display of these warnings.
* Django 1.2 will contain the backwards-compatible replica, but the warning
will be promoted to a full-fledged ``DeprecationWarning``. This warning is
*loud* by default, and will likely be quite annoying.
* Django 1.3 will remove the feature outright.
Micro releases
--------------
Micro releases (1.0.1, 1.0.2, 1.1.1, etc.) will be issued at least once half-way
between minor releases, and probably more often as needed.
These releases will always be 100% compatible with the associated minor release
-- the answer to "should I upgrade to the latest micro release?" will always be
"yes."
Each minor release of Django will have a "release maintainer" appointed. This
person will be responsible for making sure that bug fixes are applied to both
trunk and the maintained micro-release branch. This person will also work with
the release manager to decide when to release the micro releases.
Supported versions
==================
At any moment in time, Django's developer team will support a set of releases to
varying levels:
* The current development trunk will get new features and bug fixes
requiring major refactoring.
* All bug fixes applied to the trunk will also be applied to the last
minor release, to be released as the next micro release.
* Security fixes will be applied to the current trunk and the previous two
minor releases.
As a concrete example, consider a moment in time halfway between the release of
Django 1.3 and 1.4. At this point in time:
* Features will be added to development trunk, to be released as Django 1.4.
* Bug fixes will be applied to a ``1.3.X`` branch, and released as 1.3.1,
1.3.2, etc.
* Security releases will be applied to trunk, a ``1.3.X`` branch and a
``1.2.X`` branch. Security fixes will trigger the release of of ``1.3.1``,
``1.2.1``, etc.
.. _release-process:
Release process
===============
Django uses a time-based release schedule, with minor (i.e. 1.1, 1.2, etc.)
releases every six months, or more, depending on features.
After each previous release (and after a suitable cooling-off period of a week
or two), the core development team will examine the landscape and announce a
timeline for the next release. Most releases will be scheduled in the 6-9 month
range, but if we have bigger features to development we might schedule a longer
period to allow for more ambitious work.
Release cycle
-------------
Each release cycle will be split into three periods, each lasting roughly
one-third of the cycle:
Phase one: feature proposal
~~~~~~~~~~~~~~~~~~~~~~~~~~~
The first phase of the release process will be devoted to figuring out what
features to include in the next version. This should include a good deal of
preliminary work on those features -- working code trumps grand design.
At the end of part one, the core developers will propose a feature list for the
upcoming release. This will be broken into:
* "Must-have": critical features that will delay the release if not finished
* "Maybe" features: that will be pushed to the next release if not finished
* "Not going to happen": features explicitly deferred to a later release.
Anything that hasn't got at least some work done by the end of the first third
isn't eligible for the next release; a design alone isn't sufficient.
Phase two: development
~~~~~~~~~~~~~~~~~~~~~~
The second third of the release schedule is the "heads-down" working period.
Using the roadmap produced at the end of phase one, we'll all work very hard to
get everything on it done.
Longer release schedules will likely spend more than a third of the time in this
phase.
At the end of phase two, any unfinished "maybe" features will be postponed until
the next release. Though it shouldn't happen, any "must-have" features will
extend phase two, and thus postpone the final release.
Phase two will culminate with an alpha release.
Phase three: bugfixes
~~~~~~~~~~~~~~~~~~~~~
The last third of a release is spent fixing bugs -- no new features will be
accepted during this time. We'll release a beta release about halfway through,
and an rc complete with string freeze two weeks before the end of the schedule.
Bug-fix releases
----------------
After a minor release (i.e 1.1), the previous release will go into bug-fix mode.
A branch will be created of the form ``branches/releases/1.0.X`` to track
bug-fixes to the previous release. When possible, bugs fixed on trunk must
*also* be fixed on the bug-fix branch; this means that commits need to cleanly
separate bug fixes from feature additions. Each bug-fix branch will have a
maintainer who will work with the committers to keep them honest on backporting
bug fixes.
How this all fits together
--------------------------
Let's look at a hypothetical example for how this all first together. Imagine,
if you will, a point about halfway between 1.1 and 1.2. At this point,
development will be happening in a bunch of places:
* On trunk, development towards 1.2 proceeds with small additions, bugs
fixes, etc. being checked in daily.
* On the branch "branches/releases/1.1.X", bug fixes found in the 1.1
release are checked in as needed. At some point, this branch will be
released as "1.1.1", "1.1.2", etc.
* On the branch "branches/releases/1.0.X", security fixes are made if
needed and released as "1.0.2", "1.0.3", etc.
* On feature branches, development of major features is done. These
branches will be merged into trunk before the end of phase two.