On Fri, Jan 25, 2002 at 08:39:40AM -0800, Jeff Zucker wrote:
> Tim Bunce wrote:
> >
> > Please make SQL92 the default as far as possible.
>
> That is *always* my aim but I welcome people monitoring me on it since I
> may make mistakes and since the standard is not always clear.
>
> > That means all
> > identifiers are case insentivive unless enclose in double quotes.
>
> Ooops, you are correct, I misremembered what I had read, but I would
> have checked before implementing. (Data & Darwin p. 33)
>
> > And the comment style is C-style.
>
> Sorry, I disagree. From Date & Darwin, _A Guide to the SQL Standard
> Fourth Edition_, p. 30:
>
> A comment consists of two immediately adjacent hyphens ("--"),
> followed by a sequence of zero or more characters (not necessarily
> SQL language characters), terminating with a newline marker.
>
> and from the full (not draft) SQL92 spec I purchased from ansi.org,
> section 5.2:
>
> <comment> ::= <comment introducer [ <comment character>... ] <newline>
>
> <comment introducer> ::= <minus sign><minus sign>[<minus sign>...]
Ah, you can see I've spent too much time with Oracle :-)
Thanks.
Tim.
p.s. Of course the SQL standards team should be ashamed of creating
a syntax that risks breakage with things like:
"update foo set bar=bar-$value"
(If you can't see it, consider what happens if $value is negative."
