Browse Source

Commit first draft of release process documentation.

git-svn-id: svn:// b42882b7-edfa-0310-969c-e2dbd0fdcd60
zwelch 13 years ago
2 changed files with 123 additions and 0 deletions
  1. +1
  2. +122

+ 1
- 0
doc/manual/main.txt View File

@@ -18,6 +18,7 @@ check the mailing list archives to find the status of your feature (or bug).
follow when submitting patches to the project.
- The @subpage bugs page contains the content of the BUGS file, which
provides instructions for submitting bug reports to the maintainers.
- The @subpage releases page describes the project's release process.

@ref primer provide introductory materials for new developers on various
specific topics.

+ 122
- 0
doc/manual/release.txt View File

@@ -0,0 +1,122 @@
/** @page releases Release Processes

This page provides an introduction to the OpenOCD Release Proceses:
- @ref releaseswhy
- @ref releaseswho
- @ref releaseswhen
- @ref releaseshow

@section releaseswhy Why Produce Releases?

The OpenOCD maintainers should produce releases periodically.
he reasons for several reasons that should be given in detail, before
explaining who and how the processes occur.

At any time, a "source archives" can be produced by running 'make dist'
in the OpenOCD project tree. With the 0.2.0 release, this command will
produce openocd-\<version\>.{tar.gz,tar.bz2,zip} archives, which will be
suitable for being released when produced properly.

When released for users, these archives present several important
advantages when contrasted to using the Subversion repository:

-# They allow others to package and distribute the code to users.
-# They build easier for developers, because they contain
a working configure script that was produced by the Release Manager.
-# They prevent users from trying a random HEAD revision of the trunk.
-# They free developers from answering questions about trunk breakage.

Hopefully, this shows several good reasons to produce regular releases,
but these release processes were developed with some additional design
goals in mind. Specifically, the releases processes should have the
following properties:

-# Produce successive sets of release archives cleanly and consistently.
- Implementable as a script that automates the critical release steps.
-# Prevent human operators from doing it wrong, as much as possible.
-# Allow scheduling and automation of release process milestones.

The current release processes are documented in the following sections.
They attempt to meet these design goals, but there are many improvements
remaining to be made toward automating the process.

@section releaseswho OpenOCD Release Manager

OpenOCD archive releases will be produced by an individual filling the
role of <i>Release Manager</i>. This individual determines the schdule
(@see releaseswhen) and executes the release processes for the
community. Each release requires one individual to fulfill this role,
and these processes should survive any such transition gracefully.

@section releaseswhen OpenOCD Release Schedule

The OpenOCD release process must be carried out on a periodic basis
in order to realize the benefits outlined above (@see releaseswhy).

Starting with the 0.2.0 release, the OpenOCD project should produce a
new minor release each month, with a major release once per year. Bug
fix releases could be provided more frequently; however, these should
not be a priority for the Release Manager until the processes have been
fully automated, to use resources most efficiently.

If T is the time of the next release, then the following milestones
describe the release milestones for each new release cycle.

- T minus one month: start of new development cycle
- T minus two weeks: announce pending trunk closure to new work
- T minus one week: close trunk to new work, begin testing phase
- T minus two days: call for final bug fixes
- T minus one day: produce -rc packages and distribute to testers
- T minus one hour: produce final packages and post on-line

The process of scheduling release milestones should be community driven,
though the Release Manager should attempt to follow these guidelines.
Specifically, missing features that were scheduled for a release should
be dropped, rather than delaying the release cycle to wait for them.

@section releaseshow Release Process: Step-by-Step

The exact process likely requires a few releases to work out the bugs,
as it will take some experience before a script can be developed and
tested that does everything safely and robustly. Even then, some steps
require clear user intervention -- and not only by the release manager.

-# Produce final patches to the trunk (or release branch):
- add NEWS item to describe the release changes? (not ready for 0.2.0)
- the community should try to help produce this material
- can be used to automatically post "blurbs" about the project.
- bump library version if our API changed (not yet required)
- bump package version
-# Produce and verify the binary packages:
-# Start with a clean working copy, used for producing releases only.
-# produce a ChangeLog for the release (using svn2cl).
-# bootstrap, configure, and build the package.
-# run 'make distcheck' to produce the distribution archives.
-# run 'make maintainer-clean'; verify the repository is empty again.
-# Branch or tag the required tree in the Subversion repository:
- For a major/minor release from the main trunk, branch and tag it:
-# svn cp .../trunk .../branches/${BRANCH_VERSION}
-# svn cp .../branches/${BRANCH_VERSION} .../tags/${PACKAGE_VERSION}
- For a bug-fix or final release from a release branch, only tag it:
-# svn cp .../branches/${BRANCH_VERSION} .../tags/${PACKAGE_VERSION}
- where:
- BRANCH_VERSION - is x.0.0-trunk or x.y.0-trunk
- PACKAGE_VERSION - is x.y.z
-# Upload packages and post announcements of their availability:
-# Release packages into files section of berliOS project site.
-# Post announcement e-mail to the openocd-development list.
-# After the community has checked their sanity, we can post "blurbs":
-# Post NEWS update to and other trackers.
-# Submit big NEWS updates to news feeds (e.g. Digg, Reddit, etc.).

Totally-automated packaging and distribution of OpenOCD requires more
patching (post-0.2.0), but the final script(s) should be able to manage
most steps in these processes. The steps requiring user input can be
guided by an "assistant" that walks the Release Manager through the
process from beginning to end, performing basic sanity checks on their
various inputs (e.g. the NEWS blurb).

/** @file
This file contains the @ref releases page.