Re: Virtual generated columns

On Feb 9, 2025 at 16:00 +0800, Alexander Lakhin <exclusion@gmail.com>, wrote:

Hi,

I've been investigating for a while and here are my findings.

During the parse stage, we set the Var->varnullingrels in the parse_analyze_fixedparams function. 
Later, when rewriting the parse tree in pg_rewrite_query() to expand virtual columns, we replace the expression column b with a new Var that includes a, since b is defined as a * 1. 
Unfortunately, we overlooked updating the Var->varnullingrels at this point.
As a result, when we enter search_indexed_tlist_for_var, it leads to a failure. 
While we do have another target entry with the correct varnullingrels, the expression involving the virtual column generates another column reference, which causes the error.
Currently, I don't have a solid fix. 
One potential solution is to correct the Vars at or after the rewrite stage by traversing the parse tree again using markNullableIfNeeded. 
However, this approach may require exposing the ParseState, which doesn't seem ideal.
It appears that the virtual column generation function during the rewrite stage does not account for the Var field settings, leading to the errors we are encountering.

--
Zhang Mingli
HashData