We operate the "Fork & Pull" model explained at
https://help.github.com/articles/using-pull-requests
You should fork the project into your own repo, create a topic branch there and then make one or more pull requests back to the OpenStreetMap repository. Your pull requests will then be reviewed and discussed.
To understand the osm2pgsql code, it helps to know some history on it. Osm2pgsql was written in C in 2007 as a port of an older Python utility. In 2014 it was ported to C++ by MapQuest and the last C version was released as 0.86.0. In its time, it has had varying contribution activity, including times with no maintainer or active developers.
Very few parts of the code now show their C origin, most has been transformed to modern C++. We are currently targeting C++17.
Osm2pgsql uses semantic versioning.
Bugs and known issues are fixed on the main branch only. Exceptions may be made for severe bugs.
Code must be written in the K&R 1TBS style with 4 spaces indentation. Tabs should never be used in the C++ code. Braces must always be used for code blocks, even one-liners.
Names should use underscores, not camel case, with class/struct names ending in
_t
. Template parameters must use all upper case.
Header files should be included in the following order, each group in their own block:
- The corresponding .hpp file (in .cpp files only)
- Other osm2pgsql header files
- Header files from external libraries, each in their own block
- C++/C C++ standard library header files
There is a .clang-format configuration available and all code must be run through clang-format before submitting. You can use git-clang-format after staging all your changes:
git-clang-format src/*pp tests/*pp
clang-format 7 or later is required.
Comments in code should follow the Doxygen convention using backslashes (not @-signs) for commands.
User documentation is available on the website. The source of the web page is in its own repository (https://github.com/openstreetmap/osm2pgsql-website).
The man pages for osm2pgsql and
osm2pgsql-replication can be built from source
with make man
.
They need pandoc and argparse-manpage for the conversion. These tools can be installed with:
sudo apt-get install pandoc python3-argparse-manpage
Results should be checked into the repository.
Osm2pgsql must compile and pass all tests at least on Linux, OS X and Windows. Tests run on Github action to make sure that it does.
On Linux the latest stable versions of popular distributions and the stable version before that are supported if possible.
All maintained versions of PostgreSQL are supported.
osm2pgsql is tested with two types of tests: Classic tests written in C++ and BDD (Behavior Driven Development) tests written in Python.
The code comes with a suite of tests. They are only compiled and run when
BUILD_TESTS=ON
is set in the CMake config.
Tests are executed by calling ctest
. You can call ctest
with -L NoDB
to
only run tests that don't need a database.
Most of these tests depend on being able to set up a database and run osm2pgsql
against it. This is most easily done using pg_virtualenv
. Just run
pg_virtualenv ctest
pg_virtualenv
creates a separate postgres server instance. The test databases
are created in this instance and the complete server is destroyed after the
tests are finished. ctest also calls appropriate fixtures that create the
separate tablespace required for some tests.
When running without pg_virtualenv
, you need to ensure that PostgreSQL is
running and that your user is a superuser of that system. You also need to
create an appropriate test tablespace manually. To do that, run:
sudo -u postgres createuser -s $USER
sudo mkdir -p /tmp/psql-tablespace
sudo chown postgres.postgres /tmp/psql-tablespace
psql -c "CREATE TABLESPACE tablespacetest LOCATION '/tmp/psql-tablespace'" postgres
Once this is all set up, all the tests should run (no SKIPs), and pass (no FAILs). If you find something which seems to be a bug, please check to see if it is a known issue at https://github.com/openstreetmap/osm2pgsql/issues and, if it's not already known, report it there.
If you have failing tests and want to look at the test database to figure out
what's happening, you can set the environment variable OSM2PGSQL_KEEP_TEST_DB
to anything. This will disable the database cleanup at the end of the test.
This will often be used together with the -s
option of pg_virtualenv
which
drops you into a shell after a failed test where you can still access the
database created by pg_virtualenv
.
Tests in the tests/bdd
directory use behave,
a Python implementation of a behaviour-driven test framework. To run the
BDD tests you need to have behave and psycopg2 installed. On Ubuntu run:
sudo apt-get install python3-psycopg2 python3-behave
There are ctest directives to run the tests. If you want to run the tests manually, for example to run single tests during development, you can switch to the bdd test directory and run behave directly from there:
cd osm2pgsql/tests/bdd
behave -DBINARY=<your build directory>/osm2pgsql
Per default, behave assumes that the build directory is under osm2pgsql/build
.
If your setup works like that, you can leave out the -D
parameter.
To make this a bit easier a shell script run-behave
is provided in your
build directory which sets those correct paths and calls behave
. If run
with -p
as first option it will wrap the call to behave
in a call to
pg_virtualenv
for your convenience. All other command line parameters of
run-behave
will be passed through to behave.
To run a single test, simply add the name of the test file, followed by a column and the line number of the test:
behave flex/area.feature:71
If you need to inspect the database that a test produces, you can add
-DKEEP_TEST_DB
and behave won't remove the database after the test is
finished. This makes of course only sense, when running a single test.
When running under pg_virtualenv, don't forget to keep the virtual environment
as well. You can use the handy -s
switch:
pg_virtualenv -s behave -DKEEP_TEST_DB flex/area.feature:71
It drops you into a shell when the behave test fails, where you can use
psql to look at the database. Or start a shell in the virtual environment
with pg_virtualenv bash
and run behave from there.
The BDDs automatically detect if osm2pgsql was compiled with Lua and
proj support and skip tests accordingly. They also check for the test
tablespace tablespacetest
for tests that need tablespaces.
BDD tests hide print statements by default. For development purposes they
can be shown by adding these lines to tests/bdd/.behaverc
:
color=False
stdout_capture=False
stderr_capture=False
log_capture=False
To create coverage reports, set BUILD_COVERAGE
in the CMake config to ON
,
compile and run the tests. Then run make coverage
. This will generate a
coverage report in coverage/index.html
in the build directory.
For this to work you need a coverage tool installed. For GCC this is gcov
,
for Clang this is llvm-cov
in the right version. CMake will automatically
try to find the correct tool. In any case the tool gcovr
is used to create
the report.
- Decide on a new version. (See semantic versioning.)
- Update version in CMakeLists.txt, look for
project
function. - Build man page (
make man
) and copy it toman/osm2pgsql.1
. - Tag release with release notes in commit message and upload the tag to Github.
- Fill out release notes on Github.
- Copy Windows binaries and source tarball to osm2pgsql.org.
- Add release info to osm2pgsql.org.
- Publish release notes as News article on osm2pgsql.org.
The current maintainers of osm2pgsql are Sarah Hoffmann and Paul Norman.