The virtual column mystery...

[@robertstefanov]

Hi Connor, Looks like a feature introduced after 12.1 and removed in 19.7. You can achieve the same slowness with just one row: create table t as select rownum c1 from dual connect by level

[@DatabaseDude]

Thanks for that - i'll pass it on to the team. Very interesting

[@berndeckenfels]

Another example would be a where on c1, so that only one row needs to be returned, so it's mildly slow (select c1, Virtcol where c1=1) 0,1 seconds

[@kridhick]

What about the cache? Is that anonymous plsql block retieving already computed value? I could be wrong here. Anyhow, Thank you for the exploration. Eagerly waiting for the results. Just curious.

[@AMOLPALAV44]

Did you get the answer of this mystery?

[@DatabaseDude]

Yes we did. Check out this other video I did on a thing called "row shipping". kzbin.info/www/bejne/oJ3GfIewn5Knn6s Notice the multiple passes through rows? Turns out it would impact the execution count on that virtual column. This is fixed in 23c.

[@AMOLPALAV44]

@@DatabaseDude thank you

[@DominiqueComte]

I’m curious, did you get an answer ?

[@DatabaseDude]

Not to date - its been logged as a bug though.