Visible to the public Finally, Safely-extensible and Efficient Language-integrated Query

TitleFinally, Safely-extensible and Efficient Language-integrated Query
Publication TypeConference Paper
Year of Publication2016
AuthorsSuzuki, Kenichi, Kiselyov, Oleg, Kameyama, Yukiyoshi
Conference NameProceedings of the 2016 ACM SIGPLAN Workshop on Partial Evaluation and Program Manipulation
Date PublishedJanuary 2016
PublisherACM
Conference LocationNew York, NY, USA
ISBN Number978-1-4503-4097-7
KeywordsCollaboration, compositionality, Computing Theory, EDSL, human factors, language-integrated query, LINQ, Metrics, pubcrawl, relational database security, Resiliency, safe coding standards, SQL, tagless-final
Abstract

Language-integrated query is an embedding of database queries into a host language to code queries at a higher level than the all-to-common concatenation of strings of SQL fragments. The eventually produced SQL is ensured to be well-formed and well-typed, and hence free from the embarrassing (security) problems. Language-integrated query takes advantage of the host language's functional and modular abstractions to compose and reuse queries and build query libraries. Furthermore, language-integrated query systems like T-LINQ generate efficient SQL, by applying a number of program transformations to the embedded query. Alas, the set of transformation rules is not designed to be extensible. We demonstrate a new technique of integrating database queries into a typed functional programming language, so to write well-typed, composable queries and execute them efficiently on any SQL back-end as well as on an in-memory noSQL store. A distinct feature of our framework is that both the query language as well as the transformation rules needed to generate efficient SQL are safely user-extensible, to account for many variations in the SQL back-ends, as well for domain-specific knowledge. The transformation rules are guaranteed to be type-preserving and hygienic by their very construction. They can be built from separately developed and reusable parts and arbitrarily composed into optimization pipelines. With this technique we have embedded into OCaml a relational query language that supports a very large subset of SQL including grouping and aggregation. Its types cover the complete set of intricate SQL behaviors.

URLhttps://dl.acm.org/doi/10.1145/2847538.2847542
DOI10.1145/2847538.2847542
Citation Keysuzuki_finally_2016