The diversity of server-side programming languages provides great possibilites for development of applied solutions and presents a visible advantage of PostgreSQL over many other RDBMS.<br /> <br /> In this talk, we will inspect the internals of procedure languages. We will compare them in terms of functionality (based upon feature matrix) and performance in extreme and moderate cases (lots of data, lots of calculations, "the city test"). Afterwards, we'll attempt to find a explanation for the results we'll receive.<br /> <br /> An insane attempt of an alternative PL/pgSQL implementation (only partial, of course) will be undertaken in order to understand how it can be sped up without changing its underlying logic.<br /> <br /> We will discuss possibilities of the improvement in performance that could be gained by means of JIT-compilation of SQL queries (and of all server-side code, as well) by looking at the example of VitesseDB.<br /> <br /> The next subject of the talk is the PL/Perl language as an example of SQL-like server-side language. What advantages does it have compared to PL/pgSQL and how one have to pay for them? A couple of example from the speaker's experience and guidelines on how to use PL/Perl properly.<br /> <br /> In conclusion, we will look at how user-defined data types are supported in procedural language, once again, using PL/Perl as an example, taking into consideration the upcoming features of PostgreSQL 9.5. How to add a new type? We will attempt to implement support for JSON & JSONB in PL/Perl.<br /> <br /> This talk is aimed at application programmers who employ PostgreSQL.