Yes. I'm curious why the compiler didn't catch that.
2 ``` proparallel => 'u', prorettype => 'void', proargtypes => 'text text text text bool’, ```
Should we define proargtypes as “name name name name bool”? As the doc change mentions type “name” for the first 4 parameters. Maybe you want to avoid that because Name is a structure. Anyway, I don’t have a strong opinion here.
The existing pg_restore_relation_stats() and pg_restore_attribute_stats() use the text type for both parameters. This was done mostly to avoid typecasts that could potentially fail in a pg_restore script, and partly to make things easier on someone writing the function call. The -clear variants of those functions followed the same convention of text type for those paramers, and so this is just doing the same.
3 - extended_stats_funcs.c ``` * Portions Copyright (c) 1996-2026, PostgreSQL Global Development Group ```
This is a brand new file, so the copyright year should be just 2026. If you search for “2025-2026”, you will get a bunch of files, they should be created in 2025.
That may have been done because some of the code that will end up here is being moved from other files rather than being purely new. My revised patch is leaving it as-is for now.
4 - Given comment 1, no test fails. I think that’s because all test cases use inherited = false. Maybe we need a test case for inherited = true.
Added.
In addition to issues 1 and 4 I added some inheritance-specific tests per Paquier's off-list request, and the result is attached.
