"Alex Hunsaker" <badalex@gmail.com> writes:
> Right I agree this is a non-issue. For that matter if I really wanted
> to muck with it I could just set
> process_shared_preload_libraries_in_progress = true in my _PG_init
> function. And I guess if anyone thinks thats a problem we can mark
> the flag as static and only export a function for reading the value.
Yeah, I thought about that and decided to leave it as a variable ---
if anyone actually has a good reason to do it, they have an (ugly)
workaround available this way. We're only trying to catch errors
of omission, not prevent C-level code from subverting the system
if it wants to.
regards, tom lane