Block OR BoolExpr from pushdown

[svenklemm]

RestrictInfo with OR need special handling which we currently do not do which can lead to segfaults in planner if unhandled.

Fixes #6912

[akuzm]

What exactly is broken? I reproduced it but I have no idea what's going on... Theoretically, the make_restrictinfo should take care of creating the proper orclause as well.

[svenklemm]

What exactly is broken? I reproduced it but I have no idea what's going on... Theoretically, the make_restrictinfo should take care of creating the proper orclause as well.

Not yet sure what is going wrong there but the or branches are not wrapped in restrictinfos

[svenklemm]

It fixes the segfault but won't it create some perf regression?

Potentially yes. Ideally we can reenable this in followup PR after we fix OR processing.

[akuzm]

Potentially yes. Ideally we can reenable this in followup PR after we fix OR processing.

Let's go with the thing we discussed on Slack -- avoiding nested AND boolexprs by applying eval_const_expressions after modify_expression in pushdown_quals. I checked that it fixes this case.

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 80.83%. Comparing base (59f50f2) to head (5b9ffa0). Report is 164 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #6917      +/-   ##
==========================================
+ Coverage   80.06%   80.83%   +0.77%     
==========================================
  Files         190      199       +9     
  Lines       37181    37195      +14     
  Branches     9450     9704     +254     
==========================================
+ Hits        29770    30068     +298     
- Misses       2997     3246     +249     
+ Partials     4414     3881     -533     

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.