Re: buffer-substring vs. buffer-substring-no-properties

[Kevin Layer]

eliz@is.elta.co.il (Eli Zaretskii) writes:

> On 13 Nov 2001, Kevin Layer wrote:
> 
> > That is, seeing a value returned from buffer-substring of
> > 
> > #("(def" 0 1 (fontified t) 1 4 (face font-lock-keyword-face fontified t))
> > 
> > instead of "(def" is freaking out the software I'm "porting" to emacs
> > 21.
> 
> How is it ``freaking out'' your code, exactly?  Can you show an
> example of such breakage?  Perhaps there's a way to do what you want
> without switching to buffer-substring-no-properties.  AFAIK, the Emacs
> primitives generally DTRT even though properties are there.

Huh?  I have a program built in elisp (13,000+ lines) that freaks out
when it doesn't see a string when it expects it.  This program is used
by many people and is already "in the field".  When they upgrade to
emacs 21, they will be contacting me, because the program doesn't work
on emacs 21.  I can certainly fix the problem, but I do not see the
reason for such an incompatible change.  Any software that was not
packaged with emacs 21, and was thus fixed before release, is at risk
for being broken by an emacs 21 upgrade.  Right?  Doesn't this bother
anyone but me?

The broken program in question has existed since the mid 80's, and
this is the most incompatible changes I've seen, in *any* version of
emacs (emacs, epoch, lucid emacs, and xemacs).