On 2018/12/19 10:24, Amit Langote wrote:
> which is what I was thinking. Instead of specifically preventing
> partitioned tables, or foreign tables, or views, we could assert that only
> relations having heap files are passed.
Sorry, that's not what I'd said in my last email. I'd said we should add
guards so that it becomes a no-op for unsupported relkinds, which is not
the same thing as the above, so maybe we shouldn't do that. We should fix
the callers so that heap_sync is called only for heap relations. So, the
patch posted by Luis is on a good path.
Thanks,
Amit
