Fetching contributors…
Cannot retrieve contributors at this time
33 lines (26 sloc) 1.63 KB

Extending PGXN client

PGXN Client can be easily extended, either adding new builtin commands, to be included in the !pgxnclient package, or writing new scripts in any language you want.

In order to add new builtin commands, add a Python module into the pgxnclient/commands containing your command or a set of logically-related commands. The commands are implemented by subclassing the !Command class. Your commands will benefit of all the infrastructure available for the other commands. For up-to-date information take a look at the implementation of builtin simple commands, such as the ones in

If you are not into Python and want to add commands written in other languages, you can provide a link (either soft or hard) to your command under one of the libexec directories. The exact location of the directories depends on the client installation: distribution packagers may decide to move them according to their own policies. The location of one of the directories, which can be considered the "public" one, can always be known using the command pgxn help --libexec. Note that this directory may not exist: in this case the command being installed is responsible to create it. Links are also looked for in the :envvar:`PATH` directories.

In order to implement the command :samp:`pgxn {foo}`, the link should be named :samp:`pgxn-{foo}`. The :program:`pgxn` script will dispatch the command and all the options to your script. Note that you can package many commands into the same script by looking at argv[0] to know the name of the link through which your script has been invoked.