At the beginning of the eighties, if you had to operate on persistent data with a complex structure, you had to go through a cicle of reading data from storage, converting it in main memory format, operate on the main memory format, convert back to storage format, save the converted data. Somebody had to do something about this.
1. Introduction: Galileo and Fibonacci
At the beginning of the eighties, if you had to operate on persistent data with a complex
structure, you had to go through a cicle of reading data from storage, converting it in main
memory format, operate on the main memory format, convert back to the storage format, save.
At the Computer Science Department of Pisa University, Antonio Albano began the Galileo
project with the aim of designing a new approach that could end all that hussle: a modern and
robust language and system where you could just operate on complex persistent data as if it
where a local variable of your code, without any need to convert, to load, to save [
This sounds ambitious, but Galileo was actually much more than this. First of all, it abandoned
the relational data model that was dominating in that period in favour of a much more expressive
“semantic data model”. Then, it exhibited a static and strong type system, which means that
every type error was detected at compile time, with no exception. It was a mostly functional
programming language, meaning that side-efects were limited to a statically delimited amount
of updatable locations, paving the way for simpler approaches to automatic verification of
program properties and to powerful code optimization techniques [
The static type system of Galileo was inspired by the language ML, designed by Robin Milner
in the seventies [
The first paper on Galileo appears in 1983 [
In the meanwhile, the Galileo project was moving along new lines. First of all, the quest for a
better integration of object-oriented features with strong subtyping [
An important contribution of the project was the design of new linguistic mechanisms, that were absent in traditional object-oriented languages but were important to allow Galileo to be used for conceptual models. In this context, the main contributions of the project were the notions of roles, virtual objects, and named associations.
Roles and virtual objects were introduced in order to give objects the ability to change thier class
during their lifetime, and to extend to strongly typed objects the capabilities that views provide
to relational databases [
While Galileo was evolving in order to incorporate these mechanism, its type system was showing its limits, which derived from the initial choice of trading parametric polymorphism for subtyping. Advances in the study of type theory, some of which arrived from the Galileo group itself, proved now that it was possible to have a language supporting the two mechanisms at the same time [13, 14].
The combination of parametric polymorphism and subtyping is dificult but is extremely powerful. Around 1991, the Galileo group, while continuing its work on new versions of the language and system Galileo [15], began the definition of a new Database Programming Language, called Fibonacci, which would inherit all the ideas and innovations introduced by Galileo, but would embed them into an extremely rich type system, able to support new mechanisms for code reuse and modularization [16, 17].
The SEBD community followed this stream of innovations, since the first Fibonacci papers were presented in 1993 and in 1994 to SEBD, before the publication of the Fibonacci definition in VLDB Journal in 1995 [18, 19].
Fibonacci was the apex of the Galileo project, and its conclusion. The century was closing down, new research themes were attracting the group interest, such as semi-structued data, XML data, mobile ambients, graph databases, data warehouses, peer-to-peer databases. In the world of commercial databases many new systems were now object-relational, mainstream languages, such as Java, were starting to combine subtyping with parametric polymorphism, some NoSQL system were trying to address the impedence mismatch problem, the group decided to move onwards.
We do not yet understand why so many programmers have still to load data, convert, operate, convert back, save the converted data, forty years after Antonio Albano and others studied many diferent approaches to free the programming world from this legacy, but, at least, we tried to make our part. (Eds.), 17th International Conference on Very Large Data Bases, September 3-6, 1991, Barcelona, Catalonia, Spain, Proceedings, Morgan Kaufmann, 1991, pp. 565–575. URL: http://www.vldb.org/conf/1991/P565.PDF, Best Paper Award of the European Program Committee. [13] P. Curien, G. Ghelli, Coherence of subsumption, minimum typing and type-checking in F<=, Math. Struct. Comput. Sci. 2 (1992) 55–91. URL: https://doi.org/10.1017/S0960129500001134. doi:10.1017/S0960129500001134. [14] G. Ghelli, Termination of system F-bounded: A complete proof, Inf. Comput. 139 (1997) 39–56. URL: https://doi.org/10.1006/inco.1997.2662. doi:10.1006/inco.1997.2662. [15] A. Albano, G. Antognoni, G. Baratti, G. Ghelli, R. Orsini, Il Galileo95, in: A. Albano, S. Salerno, F. Arcelli, M. Gaeta, S. Rizzo, G. Vantini (Eds.), Atti del Terzo Convegno Nazionale su Sistemi Evoluti per Basi di Dati, SEBD 1995, Ravello (Costiera Amalfitana), Italy, 28-30 Giugno 1995, 1995, pp. 247–272. [16] A. Albano, G. Ghelli, R. Orsini, Fibonacci: A programming language for object databases,
VLDB J. 4 (1995) 403–444. URL: http://www.vldb.org/journal/VLDBJ4/P403.pdf. [17] G. Ghelli, F. Nanni, G. Puglielli, A. Albano, Tipi e moduli nel linguaggio Fibonacci, in: P. Atzeni, L. Cabibbo, M. Panti (Eds.), Atti del Sesto Convegno Nazionale Sistemi Evoluti per Basi di Dati, SEBD 1998, Ancona, Italy, 23-25 Giugno 1998, 1998, pp. 377–397. [18] A. Albano, R. Bergamini, G. Ghelli, R. Orsini, An introduction to the database programming language Fibonacci, in: D. Saccà (Ed.), Convegno SEBD’93, Proceedings of the Conference on Advanced Database Systems (Atti del Convegno Nazionale su Sistemi Evoluti per Basi di Dati, SEBD) 1993, Hotel Capo Suvero, Gizzeria, Italy, 14-16 June, 1993, Mediterranean Press (via S. Pellico, 13 - Tel. 0984-465645 - 87030 Rende (CS)), 1993, pp. 247–266. [19] A. Albano, C. Brasini, M. Diotallevi, G. Ghelli, R. Orsini, A guided tour of the Fibonacci system, in: S. Bergamaschi, C. Sartori, P. Tiberio (Eds.), Atti del Secondo Convegno Nazionale su Sistemi Evoluti per Basi di Dati, SEBD 1994, Rimini, Italy, 1994, Editrice Esculapio Progetto Leonardo, via U. Terracini, 30, 40131 Bologna, 1994, pp. 371–394.