Leading zeroes

[jennybc]

I wonder if leading zeroes should cause a column to come in as character? I know I can force the variable to be character and will do so in my current example.

But I suspect that numbers with leading zeroes should be treated as character more often than not. Examples: social security numbers, zip codes, phone numbers with country exit codes, numbers written written for alphalexical sorting, and agent identifiers in the 00 Section of MI6.

[hadley]

Seems reasonable to me. Reprex:

collectorGuess("00001", default_locale())

[hadley]

Hmmm, this is easy enough to do for integers, but what's the rule for doubles? Should 001.1234 be parsed as a double? What about 0.1? Is the rule that if there's an initial zero, it must be followed by a period?

[jennybc]

I think if it just worked for integers that would still be really nice. All my real-life examples are unambiguously integer (last x digits of student IDs, player numbers such as 00 or 06, etc.).

[hadley]

You have to think about doubles because 00001 is both an integer and a double, so if I fix it so readr doesn't think it's an integer, it'll just get detected as a double, which isn't a big improvement.

[hadley]

This is what I have so far

collector_guess("0")
#> [1] "integer"
collector_guess("0.")
#> [1] "double"
collector_guess("0001")
#> [1] "character"