Re: segmentation fault using currtid and partitioned tables

On Fri, May 22, 2020 at 07:32:57PM -0400, Alvaro Herrera wrote:
> I don't know, but this stuff is so unused that your patch seems
> excessive ... and I think we'd rather not backpatch something so large.
> I propose we do something less invasive in the backbranches, like just
> throw elog() errors (nothing fancy) where necessary to avoid the
> crashes. Even for pg12 it seems that that should be sufficient.

Even knowing that those trigger a bunch of elog()s which are not
something that should be user-triggerable? :)

Perhaps you are right though, and that we don't need to spend this
much energy into improving the error messages so I am fine to discard
this part. At the end, in order to remove the crashes, you just need
to keep around the two RELKIND_HAS_STORAGE() checks. But I would
rather keep these two to use ereport(ERRCODE_FEATURE_NOT_SUPPORTED)
instead of elog(), and keep the test coverage of the previous patch
(including the tests for the aggregates I noticed were missing).
Would you be fine with that?

> For pg13 and beyond, I liked Tom's idea of installing dummy functions
> for tables without storage -- that seems safer.

Not sure about that for v13. That would be invasive post-beta.
--
Michael