Subject | Re: [Firebird-Architect] Standard Conformance |
---|---|
Author | Jim Starkey |
Post date | 2003-06-16T15:20:38Z |
peter_jacobi.rm wrote:
interoperability.
The SQL standard is a counter-example that allows mutually incompatible
implementations
to conform to a single standard. Identifiers, for example, can be upper
case, lower case,
mixed case, case sensitive, and case insensitive. The cynical would say
that it is less
than a standard than an organized barrier to entry into the database
market as each
major vendor has arranged for official blessing of their individual
dialect. If you want
the full catalog of the standard's deficiencies, see the Java
DatabaseMetaData class.
All that said, no database system should exhibit arbitrary differences
from the standard.
But since nothing any database implementation can do with regards to
standard
compliance can make the goal of interoperability realizable, the effort
has more the
flavor of good faith than good deeds.
I designed DSRI and OSRI to conform to DEC's database language, SQL, and
QUEL,
none of which were then standardized. Of the three, only SQL remains.
Firebird,
in my not very humble opinion, should make more than a good faith effort
to track the
standard where possible, quickly repair major shortcomings such as
multi-level
name space, but not lose any sleep over the issue.
>I kindly ask someone who can speak for theThe goal of almost all standards -- screw threads, languages, APIs -- is
>direction of Firebird development, to comment on
>the issue of Standard Conformance in current (1.5)
>and future versions of Firebird:
>
>Being relatively new in doing SQL in earnest (and
>coming from C++), I'm somewhat disturbed by the
>general non standardization of actual SQL implementations.
>In C++ (despite the problem of 'extensions', which
>tempt the programmers to write none-portable code) all
>compiler vendors at least declare to plan to support the
>ISO standard, and even Microsoft comes close to
>doing so in VC7.
>
>
interoperability.
The SQL standard is a counter-example that allows mutually incompatible
implementations
to conform to a single standard. Identifiers, for example, can be upper
case, lower case,
mixed case, case sensitive, and case insensitive. The cynical would say
that it is less
than a standard than an organized barrier to entry into the database
market as each
major vendor has arranged for official blessing of their individual
dialect. If you want
the full catalog of the standard's deficiencies, see the Java
DatabaseMetaData class.
All that said, no database system should exhibit arbitrary differences
from the standard.
But since nothing any database implementation can do with regards to
standard
compliance can make the goal of interoperability realizable, the effort
has more the
flavor of good faith than good deeds.
I designed DSRI and OSRI to conform to DEC's database language, SQL, and
QUEL,
none of which were then standardized. Of the three, only SQL remains.
Firebird,
in my not very humble opinion, should make more than a good faith effort
to track the
standard where possible, quickly repair major shortcomings such as
multi-level
name space, but not lose any sleep over the issue.