using sqr root minus one as error code

[tclviii-dyalog]

now that irrational numbers are here i figured i could use 1J1 as an error flag ''fill'' value
when i import historic oil prices from reuters and reuters returns a '#n/a' type XL error for
days that reuters didn't capture a price (such as not having an ''open'' price on some far in the
future oil contract that nobody traded that day, even though there is a ''settlement'' value)

i figured since 1J1 was a number, i could mult/divide/add/subtract two vectors, each of which
might have some missing days, and the ''errors'' would ''flow'' through as imaginary results

of course, i didn't count on two vectors BOTH having imaginary numbers, and the
sqr root -1(s) cancelling out when multiplied together

is there some flag for preserving the provenance of the result of an arithmetic operation
such that i can tell that a real number is the result of the operation on 2 imaginary numbers?

[DanB|Dyalog]

There is no such flag but you can always use the numbers themselves as flag by extracting the irrational part using 11o, i.e.

      {⍵,[1.1]0≠11○⍵}n×*○0j.5×n←⍳9
 0J01 1
¯2    0
 0J¯3 1
 4    0
 0J05 1
¯6    0
 0J¯7 1
 8    0
 0J09 1

[Dick Bowman]

I'm a bit puzzled, when you say "irrational number" - do you mean "complex number"?

Off the top of my head I wonder whether using zilde as the fill might be a useable alternative...

      ]displayr 1 2 ⍬ 4 + 5 ⍬ 7 8
┌4───────────────┐
│   ┌0─┐ ┌0─┐    │
│ 6 │ 0│ │ 0│ 12 │
│   └~─┘ └~─┘    │
└¯2──────────────┘

which is - loosely - what I've done in the past in similar contexts.

But I can see merit also in using complex numbers with real part for value and imaginary part as an indicator - depends on the application.