On Tue, 7 Jun 2005, Yuji SHIRASAKI wrote:
> I don't think the DBMS vendors did not pay serious notice of SQL standards.
>
> My impression is that they just did not have enough standards to implement
> the required features, so they needed to introduce their own syntax extension
> to the standard.
>
> The main reason of different SQL implementations is that the SQL standardization
> is behind the SQL implementation. Fortunately we are still behind the SQL
> standardization, so there is no reason to ignore the standard.
I'm sorry, I don't agree. It seems to me that many of the deviations from the SQL standard e.g. as described by Troels Arvin at http://troels.arvin.dk/db/rdbms/ could easily be fixed: in most cases there seems no reason why the vendor's pioneering syntax and the subsequently defined standard syntax could not both be supported. This is what happens in many cases with well-standardised procedural programming languages such as Fortran and C, where almost all compilers support the ISO Standard but this by no means prevents them from providing extensions which do not conflict with it. The commercial vendors of DBMS obviously have a big financial interest in getting their customers locked in to their product by getting them to take advantage, perhaps inadvertently, of non-standard features. In support of this may I point out that Postgres, the only product compared here where the authors have no financial interest in its success, has generally the highest compliance with the SQL Standard.
But to get back to the main point: I feel that it is unlikely that most users of ADQL will be any more expert in the finer points of Standard SQL than we have shown ourselves to be (i.e. not very), so that an extension such as SELECT INTO <table> which seems natural will not confuse them overmuch, nor are they likely to complain about our stretching of the standard syntax in such a small way.
-- Clive Page Dept of Physics & Astronomy, University of Leicester, Leicester, LE1 7RH, U.K.Received on 2005-06-07Z15:36:36