bug#5672: 23.1; (intern-soft "") returns non-nil

bug#5672: 23.1; (intern-soft "") returns non-nil

[Katsumi Yamaoka]

Hi,

There's a program assuming (intern-soft STR) always return nil
if STR is the null string.  It works in Emacs 22, but doesn't in
Emacs 23.  That problem is easily solved by modifying the code
into: (unless (equal STR "") (intern-soft STR))

However, isn't it an Emacs 23 bug?  It's not so serious though.
It does not happen in Emacs that is launched using the arg -Q
but happens after loading the ~/.emacs file.  I found at least
`internal-set-lisp-face-attribute' that `set-face-attribute'
uses does intern "".  For instance:

$ emacs -Q
(intern-soft "")
 => nil
(internal-set-lisp-face-attribute
 'bold :font
 "-*-gothic-bold-r-*-*-16-*-*-*-*-*-iso8859-1"
 0)
 => bold
(intern-soft "")
 =>

I don't know what in `internal-set-lisp-face-attribute' does it.

Regards,

bug#5672: 23.1; (intern-soft "") returns non-nil

[Andreas Schwab]

Katsumi Yamaoka <yamaoka@jpl.org> writes:

> I don't know what in `internal-set-lisp-face-attribute' does it.

Inside font_parse_xlfd, where the adstyle field is usually empty.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

bug#5672: 23.1; (intern-soft "") returns non-nil

[Stefan Monnier]

> There's a program assuming (intern-soft STR) always return nil
> if STR is the null string.

That is an invalid assumption.

> It works in Emacs 22, but doesn't in Emacs 23.

Try it with Emacs-22 after doing (intern ""), you'll get the same result
as in Emacs-23.

> However, isn't it an Emacs 23 bug?

No.


        Stefan

bug#5672: 23.1; (intern-soft "") returns non-nil

[Katsumi Yamaoka]

>>>>> Stefan Monnier wrote:
>> There's a program assuming (intern-soft STR) always return nil
>> if STR is the null string.

> That is an invalid assumption.

Agreed.

>> It works in Emacs 22, but doesn't in Emacs 23.

> Try it with Emacs-22 after doing (intern ""), you'll get the same result
> as in Emacs-23.

I've already verified it but that's the case of doing it purposely,
whereas Emacs 23's case doesn't seem so.  That's why I reported it.

>> However, isn't it an Emacs 23 bug?

> No.

Ok.

bug#5672: 23.1; (intern-soft "") returns non-nil

[Stefan Monnier]

>>> It works in Emacs 22, but doesn't in Emacs 23.
>> Try it with Emacs-22 after doing (intern ""), you'll get the same result
>> as in Emacs-23.
> I've already verified it but that's the case of doing it purposely,

Right, but it could be done purposefully by some random other package.

> whereas Emacs 23's case doesn't seem so.  That's why I reported it.

It's not done "to define the empty symbol", so in a sense it's not
done "purposefully", but it's not the result of a bug either.


        Stefan