On 2016-10-03 15:40:12 -0400, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
> > On 2016-10-03 14:49:20 -0400, Tom Lane wrote:
> >> ... ignoring unresolved symbols in shlibs is the default
> >> on Linux, and while you can make it throw errors, that just leads to
> >> errors for all the references into the core backend. Not very helpful.
> >> AFAICS, GNU ld lacks any equivalent to macOS' -bundle_loader switch,
> >> which is what we'd need to make this usable.
>
> > Hm. I wonder if it's actually possible to link against the main backend,
> > when compiling as a position-independent-executable...
>
> I tried that, didn't work.
Too bad :(. The only way I could see checking this properly would be
to LD_PRELOAD the .so against postgres, at build time. That should then
error out if there's undefined symbols; without the issue that _PG_init
might error out for some reason.
Regards,
Andres