[Python-Dev] Re: Dafanging the find() gotcha

[David Abrahams]

A relevant discussion came up recently on the boost list. In our regex
library there's a function which tells you whether a string was a partial
match to the beginning of a pattern. The example used in the docs was a
credit-card number validator which watches what you type and beeps at you
if there's a mistake. Unfortunately, the implementation would return false
if the input string was empty. Of course that required special-casing for
the empty string. Eventually complaints from users caused the library
maintainer to change his mind about the response to the empty string.

http://aspn.activestate.com/ASPN/search?query=another+regex+partial+match+b
ug&type=Archive_boost_list&x=0&y=0

FWIW-ly yr's,
Dave

-----------------------------------------------------------
           David Abrahams * Boost Consulting
dave@boost-consulting.com * http://www.boost-consulting.com


----- Original Message -----
From: "Guido van Rossum" <guido@python.org>
To: "Andrew Koenig" <ark@research.att.com>
Cc: "Michael Chermside" <mcherm@destiny.com>; "python-dev"
<python-dev@python.org>
Sent: Tuesday, August 06, 2002 11:36 AM
Subject: Re: [Python-Dev] Re: Dafanging the find() gotcha


> > Michael> I think this is an extremely unconvincing example. You have
> > Michael> pushed the API up to the user of a program and supposed that
> > Michael> they expect the behavior which you are trying to defend. In
> > Michael> practice, what users expect in cases where a field is left
> > Michael> blank is for that field to be IGNORED, not for it to be
> > Michael> processed, but its contents treated as containing an empty
> > Michael> string.
> >
> > I understand.  My point is that in this particular example, what the
> > user perceives as ignoring the request is obtained by the
> > implementation technique of treating it as an empty string.  The user
> > doesn't have to know about this implementation technique, of course.
>
> I think it's a poor implementation technique. :-)  Opening the file to
> search for an empty string is very inefficient.
>
> My own potential example was some kind of graph traversal algorithm,
> representing paths by sequences of letters (the letters labeling
> edges), and involving paths that are subpaths of other paths.
> Certainly the empty path should be considered a valid subpath of other
> paths.
>
> BTW, a more fool-proof (though unfortunately slower) way of testing
> for substring containment in existing Python would be s2.count(s1) --
> this returns the number of occurrences.  And of course,
> 'abc'.count('') returns 4.
>
> --Guido van Rossum (home page: http://www.python.org/~guido/)
>
> _______________________________________________
> Python-Dev mailing list
> Python-Dev@python.org
> http://mail.python.org/mailman/listinfo/python-dev
>