Re: st: Use of subinword in Stata 13 vs. Stata 12

[vwiggins@stata.com (Vince Wiggins, StataCorp)]

Dirk Enzmann <dirk.enzmann@uni-hamburg.de> writes that the subinword()
function, when the string to be substituted for is empty, behaves
differently in version 12 than it does in version 13. (See details of
Dirk's post below my signature.)

In version 12, an empty string is always returned.  In version 13, the
original string is returned untouched.  As Robert Picard
<picard@netbox.com> notes, the Stata 12 behavior in wrong.  An
empty string does not match any of the characters in the original
string and therefore there are no substitutions to be made.  That is what
Stata 13 does.

Asking to match on an empty string is, in one sense, a poorly formed
expression.  Such expressions still occur all the time in programming,
as a degenerate case.

We will look into updating the subinword() function in Stata 12.


-- Vince 
   vwiggins@stata.com



------------ Original message ---------------------------

When trying to use -subinword()-, I get different results in Stata 13 as 
compared to Stata 12 (or 11, or 10). Example:

* --- Start Stata commands: ---------------

local vlist = "trunk"
local generate = ""
di subinword("`vlist'","`generate'"," ",.)

* --- End Stata commands. -----------------

Using version 13 the -display- command in the third line shows

trunk

(which I expect), whereas version 12 (or 11 or 10) shows

nothing. Why does the replacement of the occurance of "" in "trunk" 
yields ""? And why does version 13 yield "trunk"?
 
<end>
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/