The API of ocaml-gettext is really reduced. It is made on purpose. The design is heavily based on modules and functors. There is no real reason for this design, it was just really useful at the time the code was written.

The library supposes that all the textdomain that will be used during translation, are declared before using it. It is a real constraint, but it enables more optimisation. Moreover, it allows having a more "functional" use.

First of all, a parameter t should be defined. This parameter holds all the required value to initialise ocaml-gettext. In particular, it contains information about :

This parameter is build and updated through internal functions. You don't have direct access to it.

The parameter t is not directly used for translation. It must be converted into a parameter t' which is a real function to access translation. The transformation from t to t' is handled through a function realize. The parameter t' is used in low level translation.

All the work of the library is done in the function realize. This function is not provided in the base package. It is build out of real implementation of ocaml-gettext ( such as gettext-camomile or gettext-stub ). This function could handle all the parameters in different ways. Concerning gettext-camomile, it builds a translation table for all the files found which correspond to a declared textdomain.

Since it should be very difficult to pass a parameter t or t' in all functions that should use translation, we provide a more simple way to use the library. The top level functions use a global reference to store the parameters t and t'. This helps to integrate ocaml-gettext more easily into existing application.