On Sunday 09 August 2009 05:21:48 Jeff Davis wrote:
> * If the hook can implement XML, should we refactor the XML support (and
> COALESCE, etc.) to use the hook for the sake of consistency? If the hook
> is not good enough for those features, that might indicate a problem.
Well, for 8.4, I proposed to rewrite xmlconcat, which is currently part of
that hardcoded XML support, into a variadic function. That was shot down for
some unclear backwards compatibility reason. (I guess, someone might have
created their own xmlconcat function in a public schema and would now be
surprised that it's actually callable?!?) With that in mind, what chances of
success will a plan have that proposes to reimplement a bunch of core
functionality like COALESCE in user space?
Another example that was mentioned during PGCon and that these hooks may or
may not be useful for is somehow de-hardcoding various SQL-standard
parentheses-less functions such as current_timestamp (thus opening the door
for implementing Oracle's sysdate in userspace), but it's again unclear to me
whether that would not be objected to if those functions became subject to the
schema search path.