Re: NIST test defects

In article <D90A5A6C612A39408103E6ECDD77B8294CDE05(at)voyager(dot)corporate(dot)connx(dot)com>,
Dann Corbit <DCorbit(at)connx(dot)com> wrote:
>The Ocelot project has uncovered 23 defects in the NIST test:
>http://www.ocelot.ca/nist.htm
>
>All of the defects appear to be identified correctly as defects. I
>suggest (therefore) that the NIST verification effort consider the above
>report.

Many (most?) of the defects are addressed in the updates that are
available.

http://www.itl.nist.gov/div897/ctg/sql-testing/upd600.ted

For example,

! 13.10 Format of <update statement: searched> specifies <table name>,
! not <table reference>. Remove the "AS X" syntax and correct the
! outer reference.
sub *yts797.* //
/AS X//
/X.A/TV.A/
!
! 5.2 <reserved word> list includes DEC.
! Replace column name DEC with DEK, not a reserved word.
! This column is not used in any programs.
sub *cts5sch2.sql / DEC /
/ DEC DECIMAL/ DEK DECIMAL/

The .ted file is processed via a tool that comes with the NIST
distribution.

I would assume that whomever is running the NIST tests read the entire set
of documentation and applied the above updates.

That's not to say that update fixes all issues. But at least the handful
that I sampled off of the ocelot page were already addressed in at least
one location. Again, as pointed out in the ocelot document, it was not
unusual for the same class of error to occur in more than one location, and
I would not be surprised that some errors corrected in one location still
exist in others.

mrc
--
Mike Castle dalgoda(at)ix(dot)netcom(dot)com www.netcom.com/~dalgoda/
We are all of us living in the shadow of Manhattan. -- Watchmen
fatal ("You are in a maze of twisty compiler features, all different"); -- gcc