Would it be worthwhile for programmers to provide key words, perhaps in the
command's help file, to be detected by -search- or -findit-? The key words
could be flagged perhaps in a dedicated smcl format in order
for -search-/-findit- to easily parse, for example, {kw:date}
{kw:conversion}. Maybe it would speed the search if they were set-off in
their own paragraph in the help file, much as for the key words section of
manuscripts.
This was prompted yesterday when Svend Juul pointed out Nick Cox's -todate-
as a much more convenient alternative to a "first principles" approach that
I had suggested earlier. This is certainly true, provided that you know
about its existence. A few experimental tries with -findit- and -search-
(using terms such as "date" "conversion" "string" and so on, variously
singly or in combination) either pulled up hundreds of irrelevant hits or
missed it.
I recall that the problem of avoiding -search-/-findit- overlooking
user-written contributions has come up on the list before, but I don't
recall whether there has been any resolution other than for interested users
to notify StataCorp ad hoc of user-written contributions that seem to be
overlooked or difficult to catch. If that's as it stands, then perhaps a
more systematic approach would be better.
Obviously, the concept of set-off or smcl-identified key words is no
panacea. Its success will still depend upon both programmer and future user
formulating a problem in terms of the same key words, which doesn't always
happen for a variety of reasons. But it seems that a systematized approach
would go a long way toward helping a user quickly determine whether there's
a user-written contribution already available and tailored to the problem at
hand.
Joseph Coveney
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
