Re: 21.5 Test Suite Failures - lisp tests
Steve Youngs <youngs(a)xemacs.org> wrote:
|--==> "JJ" == Jerry James <james(a)xemacs.org>
writes:
JJ> These stem from Ben's recent synch with FSF Emacs:
JJ> <URL:http://list-archive.xemacs.org/xemacs-patches/200303/msg00004.html>;.
JJ> They apparently decided to do away with including empty strings at the
JJ> beginning and end of the list if the string starts or ends with a
JJ> separator (see the docstring). We need to decide whether we want the
JJ> old behavior, in which case split-string needs to be fixed, or the new
JJ> behavior, in which case the tests need to be fixed.
Ben, can you comment on this.
My feeling is that if `split-string' works as advertised, we fix the
tests. But I'd like to wait to find out what Ben thinks.
I haven't heard from Ben, so I did a little snooping through the Lisp
code that we ship (both in the lisp directory and in packages). In most
of the cases where I understood what was going on, it looked like the
empty strings would never be generated anyway. There were a few
functions that immediately deleted any empty strings in the result. The
one case that left me wondering is BBDB. It possibly depends on the old
behavior (see especially bbdb-get-field in bbdb.el), but I'm not sure.
Is anybody familiar enough with BBDB's inner workings to tell?
Incidentally, quite a few packages have their own split-string
functions, almost all identical (APEL, AUCTeX, clearcase, dictionary,
and EUDC to name a few). I don't know that we need to do anything about
it, necessarily.
If BBDB is not a problem or can be fixed easily, I propose that we patch
the tests to make them conform to the new behavior.
--
Jerry James
http://www.ittc.ku.edu/~james/