st 2. 2. 2022 v 15:18 odesílatel Julien Rouhaud <rjuju123@gmail.com> napsal:
Hi,
On Tue, Feb 01, 2022 at 08:35:00PM +0100, Pavel Stehule wrote: > > I tested your patch, and it looks so it is working without any problem. All > tests passed. > > There is just one question. If printalias = true will be active for all > cases or just with some flag?
Sorry, as I just replied to Bharath I sent the wrong patch. The new patch has the same modification with printalias = true though, so I can still answer that question. The change is active for all cases, however it's a no-op for any in-core case, as a query sent by a client should be valid, and thus should have an alias attached to all subqueries. It's only different if you call get_query_def() on the result of pg_analyze_and_rewrite(), since this code doesn't add the subquery aliases as those aren't needed for the execution part.
ok.
I checked this trivial patch, and I don't see any problem. Again I run check-world with success. The documentation for this feature is not necessary. But I am not sure about regress tests. Without any other code, enfosing printalias will be invisible. What do you think about the transformation of your extension to a new module in src/test/modules? Maybe it can be used for other checks in future.