<enebo[m]>
kares: 2022-05-25T12:43:07.316-05:00 [main] INFO TraceType : Exception raised: NoMethodError : undefined method `value_for_database' for [:async, false]:Array
<enebo[m]>
Have no idea if this is real or significant (debugging newest kwargs changes) but I turned on exception logging while running rails:test_sqlite3 and I see this fairly often
<enebo[m]>
The unit tests are broad so we maybe never do this for real but I thought it looked interesting
<kares[m]>
Looks weird turn on backtrace logging as well to examine where's it coming from
<kares[m]>
indeed might be worth an investigation, I am only on phone now
<enebo[m]>
This could just be my branch too. I should go back to master and see if I still see it
<enebo[m]>
wow this may not only be real but common
<enebo[m]>
our log() in core.rb is getting kwargs for async and passing them into binds param and then not loving not getting actual binds
<enebo[m]>
We might even be throwing 1 exception per query :)
<enebo[m]>
Well it does happen but it never happens in our benchmark (and specifically one which would use binds)
<enebo[m]>
I am confused why this is eating the exception though. The bad binds value is getting processed by being wrapped into a lambda and saved. Later on it will notice we have some binds to validate and then it executes that lambda. I see nothing which would catch a NoMethodError and continue on its merry way
<enebo[m]>
I would say we definitely have a mismatch on log vs current Rails so there's that