inlist(): limit of 10 arguments for strings

[Andrea Discacciati]

I'm curious to know why the inlist() function has a limit of 10 arguments for strings. Note that the number of arguments can be up to 255 for reals.

If I have a list of more than 10 strings, what are my options? Apart from:
-"breaking down" the list in two or more sub-lists and then using two or more inlist() functions joined by an "OR" condition;
-writing a very long list of "OR" conditions;
-looping over the elements of the list?

[Nick Cox]

You can map from distinct strings to a numeric variable using egen, group() and then use its values. See also http://www.stata.com/support/faqs/da...s-for-subsets/

[Andrea Discacciati]

Thank you, Nick.

I remain puzzled by the limit of only 10 strings, however.

[William Lisowski]

I cannot pass up an opportunity to demonstrate an approach using regular expressions. For what it's worth, I do agree that 10 seems like a low limit.

. input str8 s

             s
  1. gnxl
  2. gnx
  3. foo
  4. barbara
  5. end

. generate m = regexm(s,"^(gnxl|foo|bar|)$")

. list

     +-------------+
     |       s   m |
     |-------------|
  1. |    gnxl   1 |
  2. |     gnx   0 |
  3. |     foo   1 |
  4. | barbara   0 |
     +-------------+

[Nick Cox]

Why only 10 strings?

Only those who wrote the code at StataCorp can answer, but

* as strings can be very, very long in Stata, some limit is not that surprising

* perhaps this is a deliberate signal that you should be thinking about another approach if you want to specify more than a short list.

[wbuchanan]

I would suspect it is a combination of performance and memory management. With numeric values a sort can help to select the records fairly efficiently (e.g., selecting from two ordered lists isn't terribly demanding), but doing the same with strings (each of which requires a varying amount of memory) may not be as efficient.

[Andrea Discacciati]

William Lisowski Thank you very much!