J clause now requiring (parentheses) around statements to recognize variables

[mlandry22-h2o]

After upgrading to 1.9.7, a few statements that previously ran successfully now create errors due to unrecognized columns.

A workaround seems to be to wrap the statement in parentheses.

A pair of examples.

## case 1
allNumeric<-all(DT[,!is.na(as.numeric(FieldName))])   ## original
# Error in eval(expr, envir, enclos) :
# object 'FieldName' not found

allNumeric<-all(DT[,(!is.na(as.numeric(FieldName)))])   ## workaround

## case 2
dupeType[removalIndex>0,rowId-(2*removalIndex-1)]   ## original
dupeType[removalIndex>0,(rowId-(2*removalIndex-1))]   ## workaround

[MichaelChirico]

@mattdowle I guess this is a side-effect of the new auto-.SD-with-FALSE?

Catching ! as jsub[[1]] and assuming it was column exclusion?

DT = data.table(FieldName = c("1", "2", "3", "four", "five"))

#works fine
DT[ , is.na(as.numeric(FieldName))]
#fails, as above
DT[ , !is.na(as.numeric(FieldName))]

I'm surprised this wasn't caught by any unit tests!

[mattdowle]

Yes indeed. Recent change last few days. Thanks. Was not intended. On it ...