|type checking SQL expressions email@example.com (Jean Morissette) (2005-03-20)|
|Re: type checking SQL expressions firstname.lastname@example.org (George Neuner) (2005-03-24)|
|Re: type checking SQL expressions Brian.Inglis@SystematicSW.ab.ca (2005-03-25)|
|Re: type checking SQL expressions email@example.com (Gene Wirchenko) (2005-03-27)|
|Re: type checking SQL expressions Brian.Inglis@SystematicSW.ab.ca (Brian Inglis) (2005-03-31)|
|Re: type checking SQL expressions firstname.lastname@example.org (Jean Morissette) (2005-04-11)|
|Re: type checking SQL expressions email@example.com (George Neuner) (2005-04-16)|
|From:||George Neuner <firstname.lastname@example.org>|
|Date:||16 Apr 2005 11:13:40 -0400|
|References:||05-03-076 05-03-100 05-03-111 05-03-123 05-04-034|
On 11 Apr 2005 00:21:52 -0400, Jean Morissette
>Thanks for your responses, but I think that you have misunderstand me.
>By 'type system' I was refering to formal typing rules. By exemple,
>Trigoni has defined the typing rules of OQL in this paper: ?Towards a
>Formal Type System for ODMG OQL? at
>It's really easy to check the type of each node in a parse tree when
>the typing rules are defined.
We understand what you want - the problem is it doesn't exist. The
SQL committee are not language theorists and, AFAIK, they never
described a formal type system for the query language. There are
probably been some academic studies that have produced informal
descriptions, but a quick search in the ACM library and on Google
didn't turn up anything that looked useful.
There has been a lot written about type systems for database languages
but most is aimed at developing new languages and very little is
relevent to SQL as it exists now.
Jan Van den Bussche wrote a paper called "Polymorphic type inference
for the relational algebra". It is a general type theoretic treatment
of the algebra which is not aimed at SQL per se but it may address
some of your questions and point you in the right direction.
An index of his papers is at
Return to the
Search the comp.compilers archives again.