On Mon, Sep 3, 2018 at 5:05 AM Andres Freund <andres@anarazel.de> wrote:
> On 2018-08-27 22:40:17 -0400, Tom Lane wrote:
> > Thomas Munro <thomas.munro@enterprisedb.com> writes:
> > > We could probably improve that situation by making syscache lookups
> > > (and probably other things too) fail when called from _PG_init() in
> > > regular backends so that extension authors are made aware of this
> > > hazard, or perhaps go the other way and change the order we do things
> > > in parallel workers.
> >
> > Hmm. There's an argument to be made for the latter: we don't really
> > want stuff failing in parallel workers if it works fine normally.
>
> Yea, I guess there's an argument to be made for that.
>
>
> > On the other hand, it seems clear to me that we *don't* want extensions to
> > be doing stuff like syscache lookups in _PG_init(), because that would
> > prevent them from working as shared_preload_libraries entries.
>
> Yea, that doesn't seem great.
>
>
> > And on the third hand, intentionally breaking code that used to work
> > isn't likely to win us many friends either. So I'm not sure that your
> > first option is really tenable. Perhaps we could get away with doing
> > it in HEAD and not back-patching ... but that does little for existing
> > problems.
>
> I wonder if we could make it warn in all branches and error out hard in
> master? That way hopefully most extensions would be fixed long before
> the next release comes out, and people don't load new libraries at that
> crazy a rate...
Any other views? It sounds like we do want a patch like the one I
posted earlier. The question is whether we also want a new warning,
and in later releases an error. Do we really want to crack down on
extension libraries that wouldn't work in shared_preload_libraries?
What about hypothetical libraries that are smart enough to avoid
making such calls when they detect that they're running from
shared_preload_libraries -- you'd generate false warnings when loaded
the regular way.
--
Thomas Munro
http://www.enterprisedb.com