Re: pgsql: Add more SQL/JSON constructor functions

On Fri, Sep 6, 2024 at 1:34 PM Amit Langote <amitlangote09(at)gmail(dot)com> wrote:
> On Fri, Sep 6, 2024 at 12:07 PM Amit Langote <amitlangote09(at)gmail(dot)com> wrote:
> > On Thu, Sep 5, 2024 at 9:58 PM Amit Langote <amitlangote09(at)gmail(dot)com> wr
> > Pushed.
>
> Reverted 0002-0004 from both master and REL_17_STABLE due to BF failures.
>
> 0002-0003 are easily fixed by changing the newly added tests to not
> use EXPLAIN VERBOSE to test deparsing related changes, so will re-push
> those shortly.

Done.

> 0004 perhaps doesn't play nicely with LLVM compilation but I don't yet
> understand why.

Attached is an updated patch that takes care of the issue. The bug
was that llvm_compile_expr() didn't like that jump_error, jump_empty,
and jump_end could all point to the same step. In the attached,
jump_empty / jump_error are left to be -1 if ON ERROR, ON EMPTY steps
are not added, instead of making them also point to the step address
that jump_end points to. ExecEvalJsonExprPath() are also updated to
check if jump_error or jump_empty is -1 and return jump_end if so.

--
Thanks, Amit Langote