Jácome Cunha

n/a

This technical report describes the research goals and results of the SpreadSheet as a Programming Paradigm research project. This was a project funded by Funda{\c c}ão para a Ciencia e Tecnologia – FCT: the Portuguese research foundation, under reference FCOMP-01-0124-FEDER-010048, that ran from May 2010 till July 2013. This report includes the complete document reporting the results achieved during the project execution, which was submitted to FCT for evaluation on October 2013. It describes the goals of the project, and the different research tasks presenting the deliver- ables of each of them. It also presents the management and result dissemination work performed during the project's execution. The document includes also a self assess- ment of the achieved results, and a complete list of scientific publications describing the contributions of the project. Finally, this document includes the FCT evaluation report.

Although spreadsheets can be seen as a flexible programming environment, they lack some of the concepts of regular programming languages, such as structured data types. This can lead the user to edit the spreadsheet in a wrong way and perhaps cause corrupt or redundant data. We devised a method for extraction of a relational model from a spreadsheet and the subsequent embedding of the model back into the spreadsheet to create a model-based spreadsheet programming environment. The extraction algorithm is specific for spreadsheets since it considers particularities such as layout and column arrangement. The extracted model is used to generate formulas and visual elements that are then embedded in the spreadsheet helping the user to edit data in a correct way. We present preliminary experimental results from applying our approach to a sample of spreadsheets from the EUSES Spreadsheet Corpus.

Spreadsheets are notoriously error-prone. To help avoid the introduction of errors when changing spreadsheets, models that capture the structure and interdependencies of spreadsheets at a conceptual level have been proposed. Thus, spreadsheet evolution can be made safe within the confines of a model. As in any other model/instance setting, evolution may not only require changes at the instance level but also at the model level. When model changes are required, the safety of instance evolution can not be guarded by the model alone. We have designed an appropriate representation of spreadsheet models, including the fundamental notions of formulæand references. For these models and their instances, we have designed coupled transformation rules that cover specific spreadsheet evolution steps, such as the insertion of columns in all occurrences of a repeated block of cells. Each model-level transformation rule is coupled with instance level migration rules from the source to the target model and vice versa. These coupled rules can be composed to create compound transformations at the model level inducing compound transformations at the instance level. This approach guarantees safe evolution of spreadsheets even when models change.

Spreadsheets are notoriously error-prone. To help avoid the introduction of errors when changing spreadsheets, models that capture the structure and inter-dependencies of spreadsheets at a conceptual level have been proposed. Thus, spreadsheet evolution can be made safe within the confines of a model. As in any other model/instance setting, evolution may not only require changes at the instance level but also at the model level. When model changes are required, the safety of instance evolution can not be guarded by the model alone. Coupled transformation of models and instances are supported by the 2LT platform and have been applied for transformation of algebraic datatypes, XML schemas, and relational database models. We have extended 2LT to spreadsheet evolution. We have designed an appropriate representation of spreadsheet models, including the fundamental notions of formulæ, references, and blocks of cells. For these models and their instances, we have designed coupled transformation rules that cover specific spreadsheet evolution steps, such as extraction of a block of cells into a separate sheet or insertion of columns in all occurrences of a repeated block of cells. Each model-level transformation rule is coupled with instance level migration rules from the source to the target model and vice versa. These coupled rules can be composed to create compound transformations at the model level that induce compound transformations at the instance level. With this approach, spreadsheet evolution can be made safe, even when model changes are involved.

Spreadsheets can be viewed as a highly flexible end-users programming environment which enjoys wide-spread adoption. But spreadsheets lack many of the structured programming concepts of regular programming paradigms. In particular, the lack of data structures in spreadsheets may lead spreadsheet users to cause redundancy, loss, or corruption of data during edit actions. In this paper, we demonstrate how implicit structural properties of spreadsheet data can be exploited to offer edit assistance to spreadsheet users. Our approach is based on the discovery of functional dependencies among data items which allow automatic reconstruction of a relational database schema. From this schema, new formulas and visual objects are embedded into the spreadsheet to offer features for auto-completion, guarded deletion, and controlled insertion. Schema discovery and spreadsheet enhancement are carried out automatically in the background and do not disturb normal user experience.

This paper presents techniques and tools to transform spreadsheets into relational databases and back. A set of data refinement rules is introduced to map a tabular datatype into a relational database schema. Having expressed the transformation of the two data models as data refinements, we obtain for free the functions that migrate the data. We use well-known relational database techniques to optimize and query the data. Because data refinements define bidirectional transformations we can map such database back to an optimized spreadsheet. We have implemented the data refinement rules and we have constructed tools to manipulate, optimize and refactor Excel-like spreadsheets.

Algebraic theories for modeling components and their interactions offer abstraction over the specifics of component states and interfaces. For example, such theories deal with forms of sequential composition of two components in a manner independent of the type of data stored in the states of the components, and independent of the number and types of methods offered by the interfaces of the combinators. General purpose programming languages do not offer this level of abstraction, which implies that a gap must be bridged when turning component models into implementations. In this paper, we present an approach to prototyping of component-based systems that employs so-called type-level programming (or compile-time computation) to bridge the gap between abstract component models and their type-safe implementation in a functional programming language. We demonstrate our approach using Barbosa's model of components as generalized Mealy machines. For this model, we develop a combinator library in Haskell, which uses type-level programming with two effects. Firstly, wiring between components is computed during compilation. Secondly, the well-formedness of the component compositions is guarded by Haskell's strong type system.