|Right, and that's what it turned out Alan was trying to do in the first |place: Find a phrase that may contain delimiters, set off by delimiters |on either side of it. I would have changed "word" to "phrase" except |that I was leaving as much of his code intact as possible to make it |easy to compare. | |My solution is equivalent to his, just without the need to choose a safe |character to substitute for the target phrase. Note that neither will |find a phrase that starts or ends with a word delimiter that abuts |another word. In Alan's example of searching for A(I), he wouldn't find |it in the context A(I)B(J). To handle cases like that, you'd test the |first and last characters of the phrase, and omit the appropriate VERIFY |stage when either is in the list of word delimiters.
Correct, I'm not really looking for words in the usual sense, or phrases exactly. I cannot simply "split everything up into words", because what I am searching for may contain word-separators. Thank you for your solution, which is better than mine. And thank you for pointing out the example of A(I)B(I). I don't think that's valid FORTRAN, but it is valid REXX. I haven't figured out what to do with it, but I would like to be able to find A(I) and/or B(I) in this case, too. Alan Ackerman Alan (dot) Ackerman (at) Bank of America (dot) com
