Zeugswetter Andreas SB <ZeugswetterA@wien.spardat.at> writes:
> Actually I am not sure whether the column = NULL syntax is even defined
> or allowed in SQL92 (e.g. Informix interprets the NULL as column name in
> this context and errs out).
Strictly speaking, SQL92 would require you to writefoo = CAST (NULL AS type-of-foo)
However, we allow unadorned NULL in other contexts as a shorthand for
the CAST notation, so it's inconsistent of us to say that in this
context it means something different.
The real problem with accepting this Microsoftism is that it's a trap
for unwary programmers. Case 1: someone who's not studied SQL in detail
might experiment with examples involving "foo = NULL" and jump to
reasonable but entirely incorrect conclusions about how comparisons
involving NULL operate. Case 2: someone who *has* studied SQL, and is
also aware that we accept unadorned NULLs, will also draw the wrong
conclusions about what this construct will do. Bottom line: this kluge
surprises everyone except those who already know it exists. I don't
like systems that surprise their users in inconsistent ways.
regards, tom lane
