- 1 Development Process Goals
- 2 Documentation
- 3 Branches
- 4 Dependencies
- 5 Developing New Features
- 6 Fixing Bugs
- 7 Release Version Numbering
- 8 Updating Translations
Development Process Goals
The goal of this process is to limit the number of potential bugs that enter the stable release branch(es) due to changes made to the sources. We want to make sure that changes made to releases have been audited and vetted BEFORE they are commited to the release branches. At the same time we want to try to keep the main development branch as stable as practical.
In particular, we want to prevent "bad releases". Those are releases where a bugfix inadvertantly breaks other code.
Documentation should start before coding ...
New or reworked features should also be documented as it turns out, some bugs are a problem of communication between user and developer. If you do not have the resources to update the documentation yourself, you should at least assign a Bugzilla entry to the documentation team with the essential information.
Branches in Git are cheap, therefore we should use branches to our advantage. There are always two "interesting" branches in use:
- master -- the main development branch
- maint -- the current stable release branch.
When close to a major release there will be a third "interesting" branch:
- 'unstable' -- the pre-release branch for the upcoming major release
Older release series are labelled with the release number, e.g. 2.4. Those branches are normally closed to new work shortly after a new maint branch is created from unstable at the beginning of a new release series.
Developers may also have other branches in use for development or bugfixing. This is described later in this document. These will generally not be pushed to the main repository.
Changing dependencies should usually happen on the master branch.
- Places to change (minimal) versions
- Some version numbers are used in Templates for easier updates of links etc.
Developing New Features
All new development work should target the master branch. This includes new features and enhancements. Small changes can be commited directly to master. Larger features or architectural changes should happen on a feature-branch; once that feature is working it can get merged back into master.
Bug fixes should be built against the maint branch. See Fixing Bugs below for details.
At any particular time, master should be "as stable as possible". We use feature branches to limit the risk that a feature never completes. If development of a feature were done solely in master and that feature never got completed it could leave master in a state where it doesn't compile, or it fails the regression tests.
If you are at all questioning whether to develop directly in master or in a branch, do it in a branch.
As a general rule, development branches should not be pushed back to the main repository. If one needs to publish a development branch, fork the Github repository to your personal account and publish the branch there.
Developers with the requisite privs may push minor changes directly to master or to maint as appropriate without a feature branch.
Use your judgement about whether your major change should be done directly on master/maint or whether you should use a feature branch. We trust you. Just remember, smaller changesets are easier to audit. When possible try to ensure that each commit can compile; that makes it much easier to bisect when tracking down a regression.
Backwards-Incompatible Schema Changes
To make the policy clearer, backwards-incompatible schema changes in an unstable release, e.g. 2.5, require a change to the previous stable release, e.g. 2.4, to provide a read facility for the new schema.
Often it is easier to add a new KVP which will be simply ignored by older versions than to change the schema. But then it would become harder to get the database normalized for the SQL backend.
LibGnuCash API Stability
Historically GnuCash has been developed as a stand-alone library with no concern for stability or backwards compatibility of the core API in spite of simultaneously insisting that the only safe way to operate on GnuCash data is to write code using the somewhat published API in C, Scheme, or Python. Over the first 20 years or so API has mostly been added and seldom pruned, so at as we get into converting more interesting components to C++ it's time to clean up and de-duplicate the API.
As far as we're able to tell no-one has ever written a consumer of the GnuCash C API outside of GnuCash itself, so during the conversion we're not going to worry about it. Everything that isn't Gui or Report will be rewritten in C++ and the C++ API will be new, and once it's internally stable we can begin to treat it like a real library with API guarantees. The C API necessary to interact with the Gtk GUI will remain as long as we retain the latter but there will be no stability guarantee for that part.
The bindings are another matter entirely. Much of the existing C API is also exported to Scheme and Python via SWIG. We know of many cases where users have written custom reports and a few cases of users writing Python extensions. Those users deserve reasonable stability in the API, hence the following policy.
- Functions must not change signature (except in C++ which supports overloading). If one needs to do the same thing with different types, make a new function with a different name.
- Functions may only be deleted in the master branch. Functions to be deleted must first be marked deprecated in the maint branch at least 6 months before the major release.
- Deprecated functions should be rewritten as wrappers of their replacement functions. If the function is no longer needed in the C library the wrapper should be in the SWIG interface file (e.g. engine.i).
Marking a function as deprecated
- C: __attribute__((deprecated)) (gcc/clang) or __declspec(deprecated("explanation/redirection string")) (Visual Studio)
- C++: [[deprecated("explanation/redirection string")]] (Requires C++14, use the C method in maint)
- Python: warnings.warn("explanation/redirection string", DeprecationWarning, stacklevel=2)
- Scheme: (issue-deprecation-warning "explanation/redirection string")
Python Note: DeprecationWarnings are disabled by default, so developers of Python programs using GnuCash bindings should test by invoking python with -Wd or setting PYTHONWARNINGS=default in the shell environment.
In general bug fixes should always target the oldest active branch that features the bug. If the bug is present in the current stable version of gnucash, the bug should be fixed on maint (or a sub branch thereof). If the bug exists only in the development version of gnucash, the bug should be fixed on master only.
The exception here would be a bug in the stable branch for which the fix itself is a very complex or large body of code. Since this code has a higher risk of introducing new bugs in itself it should not be applied on the maint branch. Instead such a bug can only be fixed on the master branch (unstable version).
Since we frequently merge the maint branch into master, please do a test merge of maint into master in your local repository to ensure that your patch merges cleanly.
If it doesn't and you don't know what to do about it, say so on the bug report when you attach your patch so that a developer can help you asses the situation and decide how to resolve it.
Many times a bugfix is simple. In that case the bug can be fixed directly on the maint branch and the changeset can be merged forward to the master branch so future major releases carry it as well.
If the bugfix is sufficiently complicated then that bugfix warrants a bugfix branch. The developer can create a branch off the release branch (maint) to work on the bugfix, and once that's been tested that branch can be merged back into the release branch.
Release Version Numbering
GnuCash uses two version number schemes:
- For versions up through 2.7.8, each GnuCash version number consists of 3 numbers separated by a period, in format Fundamental.Major.Bugfix E.g. 2.4.11
- Beginning with version 3.0, released in April 2018, each version number consists of only 2 integers separated by a period,
meaning Major.Minor, e.g. 3.1.
The only fundamental change, from 1 to 2, involved changing most of the code from Scheme to C and upgrading the GUI from Gtk1 to Gtk2. Since we used this so infrequently we decided to switch to a two-number release.
Major releases are planned every 2 - 4 years. See the Release Schedule for the development teams forecast.
There are 2 types of Major releases, Stable and Unstable.
- Stable releases are intended for use by users with their live data. They have one or two-digit minor versions, e.g. 3.1 or 3.21.
- Unstable release numbers have three digit minor versions beginning with 900, e.g. 3.902.
- Unstable releases are the precursors to a stable release, and are intended only for testing, by either developers or interested community members who don't have the time or skills to build GnuCash from source. They are meant to expose the developers' work to a wide audience, so changes can be tested in a range of environments, before being included in a stable release.
- Unstable releases come late in a major development cycle. By the time unstable releases are issued, relevant longer term modifications should be nearly done (except for some bug fixing).
- Unstable releases are very short-lived, normally only six months or so, with new releases every month and a final pre-stable release two weeks before the stable release. If a significant problem surfaces in that last release, the developers will focus on that problem and do another pre-stable with the stable to follow two weeks later.
- Due to the possibility of data corruption, unstable releases should only be used on a copy of live GnuCash data.
Minor releases are made typically every 3 months; see the Release Schedule. Minor releases contain fixes to reported problems and small features that the development team consider low risk for incompatibility.
Minor releases should never break file compatibility with other releases in the same major version.