UDF for API pull

[jerrykern]

I'm writing a UDF to pull data from the U.S. SEC's EDGAR database using their API. The database has two types of data, however, string and numerical. Should I write separate UDFs to pull and return the separate types of data, or write one UDF that returns type variant? I'm pretty sure I could do either, but am having trouble thinking through pros and cons.

 

[ranman256]

I do 99% variables as variant to catch nulls.

If i need to do calculations, i convert it internally to the correct number type to get correct returns. (like double or long int.)